EECS SuperUrops All SuperUROP Projects
spacing
EECS SuperUrops
 
Home
EECS
Letters of Recommendation
Login

SuperUROP Projects

"Power Electronics to Enable Energy Access in the Developing World: Consumer Module"
Faculty Advisor: David Perreault
Mentor(s): Wardah Inam
Contact e-mail: wardah@mit.edu
Research Area(s): Circuits, Energy
The lack of electricity is one of the most pressing concerns in the developing world. Today, about one out of every 5 people, or a total of 1.3 billion people in the world remain in the dark and are denied basic standard of living due to the lack of electricity. While there is a pressing need to provide electrification, the current technologies have not been able to scale to serve these areas. This project is focused on developing technology that enables a peer-to-peer scalable DC microgrid that creates a marketplace for electricity. This results in people getting access to electricity who were not able to afford generating assets. And it also incentivizes people who can afford generating assets to invest in more generation as now they are able to sell the excess power. The research will include designing and testing of low cost and high efficiency dc/dc power converter for the consumer module. The technology developed will be piloted in India. Power Electronics to Enable Energy Access in the Developing World: Consumer Module Reuse
"Design and Implementation of a Planar Magnetics Analysis System (P-MAS)"
Faculty Advisor: David Perreault
Mentor(s): Minjie Chen
Contact e-mail: minjie@mit.edu
Research Area(s): Applied Physics, Circuits, Energy, Numerical Methods
The Power Electronics Research Group has recently developed a systematic approach to modeling and analyzing planar magnetic devices utilizing lumped circuit models. This method enables losses, current sharing and other aspects of planar magnetic devices to be rapidly and accurately analyzed. It is more powerful than conventional analytical approaches, and is much faster than numerical methods.
To date, use of this modeling approach has been carried out in a two-step process:
step 1: planar magnetic device structures are described using circuit netlists through MATLAB;
step 2: circuit netlists are solved by a SPICE engine to calculate performance and device parameters.
The goal of this SuperUROP project is to integrate these two stages and implement a standalone software package for analyzing planar magnetics that is easy to use and widely distributable. The project has the following two phases:
(1) Based on the original MATLAB+SPICE implementation, develop an integrated back-end engine in Python (or C++, or in other languages, if it has a SPICE library).
(2) Design and develop a front-end GUI.
Students will gain experience in developing a practical and useful CAD tool, and gain exposure to mathematical modeling of physical systems. This software, if successfully developed, has the potential to be adopted for developing planar magnetic designs.
Design and Implementation of a Planar Magnetics Analysis System (P-MAS) Reuse
"Rapid Visualization of Pictures and Medical Images"
Faculty Advisor: Polina Golland
Mentor(s): Adrian Dalca
Contact e-mail: adalca@mit.edu
Research Area(s): Artificial Intelligence, Graphics and Human-Computer Interfaces
The goal of this project is to build an interactive visualization tool to quickly explore large collections of pictures and medical images. The main idea is to use the movement of the pointer on a canvas to control the image being shown. Applications range from medical image analysis to photo collections and time lapse visualization. Aside from being usable as an artistic canvas, such a visualization tool will facilitate biomedical and computer vision research. We have several exciting directions to explore, and there is significant room for students to choose the most interesting paths for them. The tool will be developed for modern web browsers. Candidates should have familiarity with javascript, HTML and CSS.
An early design demonstrating the main goal is available here: Link
Time lapse demo: Link
Medical data demo: Link
Rapid Visualization of Pictures and Medical Images Reuse
"Mass-customization and personalization of 3D printed objects"
Faculty Advisor: Wojciech Matusik
Mentor(s):
Contact e-mail: wojciech@csail.mit.edu
Research Area(s): Artificial Intelligence, Computer Systems, Graphics and Human-Computer Interfaces, Numerical Methods
The goal of this project is to develop web-based tools for mass-customization and mass-personalization of objects that can be 3D printed. We will investigate different application scenarios where this model provides significant benefits over competing technologies. Reuse
"Building a Mobile Personal Capnograph"
Faculty Advisor: Thomas Heldt
Mentor(s): Becky Asher, George Verghese
Contact e-mail: thomas@mit.edu
Research Area(s): BioEECS, Circuits, Signals and Systems
A capnograph, or CO2 monitor, in an instrument for effort-independent noninvasive recording of
the profile of CO2 concentration in exhaled breath as a function of time, using a nasal cannula. Capnography is an essential element of modern anesthesia and respiratory care, and capnographs are available in every operating room, intensive care unit, emergency department, and ambulance system in the United States. Some current research in our group is aimed at extracting information about cardiorespiratory health from the shape of the CO2 waveform. We hope to eventually use capnography to monitor treatment in asthma patients. This project is aimed at designing and constructing a basic smartphone-based personal capnograph. The project will involve researching CO2 sensors for suitability in this application, incorporating the sensor into appropriate electronics and airflow tubing, working out the computation and display, and carrying out initial tests. You will work with a team of engineers and physicians on this project.
Reuse
"Computer Theorem-Proving for Undergrad CS Education"
Faculty Advisor: Adam Chlipala
Mentor(s):
Contact e-mail: adamc@csail.mit.edu
Research Area(s): Computer Systems, Graphics and Human-Computer Interfaces, Theoretical Computer Science
We are building web-based software to teach courses centered on mathematical proofs. The idea is to build an IDE usable by MIT freshmen, where it is possible to state and prove mathematical theorems, such that the system catches logical errors, much in the way that a programming IDE catches syntax or type errors. Our backend is the Coq proof assistant software, which is based on functional programming; but we want to build something much more accessible to newcomers than the common Coq interfaces. There are many opportunities to work on the general infrastructure (which is implemented in the Ur/Web language) or its application to particular course modules. Computer Theorem-Proving for Undergrad CS Education Reuse
"Next-Generation Stellar Replacement in a Fancy Programming Language"
Faculty Advisor: Adam Chlipala
Mentor(s):
Contact e-mail: adamc@csail.mit.edu
Research Area(s): Computer Systems, Graphics and Human-Computer Interfaces, Theoretical Computer Science
We are working on building a replacement for MIT's Stellar system, with more customization options and generally more neat stuff. We're using Ur/Web, a new functional programming language that supports unusual modularity features. The idea is that static type-checking will validate the reasonableness of the feature mash-up associated with each new class. There are many opportunities to test the limits of our architecture by implementing new feature plugins, which involves coming up with very expressive interfaces using static types. Next-Generation Stellar Replacement in a Fancy Programming Language Reuse
"Automatic and Sound Caching for Web Applications"
Faculty Advisor: Adam Chlipala
Mentor(s):
Contact e-mail: adamc@csail.mit.edu
Research Area(s): Computer Systems, Graphics and Human-Computer Interfaces, Theoretical Computer Science
Ur/Web is a specialized programming language for SQL-database-backed Web applications. Often, to maximize performance, programmers manually write code to cache results of common database queries, within the application. Mistakes in this code can introduce all sorts of zany bugs. The proposed project is to implement a compiler optimization that automatically finds and implements caching opportunities. Automatic and Sound Caching for Web Applications Reuse
"Statistical Analysis of Phonetic Inventories of the Worlds Languages"
Faculty Advisor: Jim Glass
Mentor(s):
Contact e-mail: glass@mit.edu
Research Area(s): Artificial Intelligence
This project will involve analyzing the distribution of sounds across
the worlds languages. The goal is to develop a phonetic prior model
that can serve to guide a computer as it tries to learn the phonetic
inventory of an unknown language. The CSAIL Spoken Language Systems
groups has been developing non-parametric Bayesian methods to learn
speech sounds. The SuperUROP involved in this project would work with
graduate students to refine these models.
Reuse
"A Library of Lattice-based Cryptographic Primitives"
Faculty Advisor: Vinod Vaikuntanathan
Mentor(s):
Contact e-mail: vinodv@mit.edu
Research Area(s): Computer Systems, Theoretical Computer Science
Lattices are mathematical objects have enabled a revolution in the field of cryptography in the past few years. We have been able to realize advanced forms of encryption such as fully homomorphic encryption, functional encryption and program obfuscation, all thanks to lattices.

The goal of this project is to build a library of lattice-based cryptographic primitives. This will include fast algorithms for public and secret-key encryption, identity-based encryption as well as advanced forms of attribute-based encryption and functional encryption.

Pre-requisites: Strong mathematical background (equivalent of 18.06 and 18.701) and strong programming skills.
A Library of Lattice-based Cryptographic Primitives Reuse
"Using Attribute-based Encryption to Secure Software Defined Networks"
Faculty Advisor: Vinod Vaikuntanathan
Mentor(s): Sergey Gorbunov
Contact e-mail: vinodv@mit.edu
Research Area(s): Computer Systems, Theoretical Computer Science
Prerequisites: Strong Programming Skills and Mathematical Sophistication. Knowledge of Cryptography is plus.

Software Defined Networking (SDN) is an emerging paradigm that allows building flexible networks for dynamic environments. On the high level, it introduces a centralized (or distributed) programmable controller that manages applications, forwarding and security policies of the entire network. Switches (and other devices) use OpenFlow protocol to communicate with the controller to make routing decisions. Unfortunately, the controller introduces a central point of failure in the network. In the simplest case, an adversary compromising the controller can fool the switches to route packets to a wrong destination. This project will investigate how to protect Software Define Networks in case the controller gets compromised. In particular, we will design, implement and evaluate a novel solution based on recent developments in Attribute-Based Encryption (ABE). One of the features of an ABE is that it provides a mechanism to efficiently authenticate computation.

Using an ABE, we will enhance OpenFlow protocol by making the controller to provide a short proof of computation. Each switch will be able to efficiently verify the routing decisions. The security of the ABE guarantees that even in the case the controller gets compromised, the switch will not accept a fake routing decision.

Put together, this project will significantly help to ensure that the new generation of computer networks is mathematically proven to be secure by design.
Reuse
"Network Code Design for Non-multicast Case"
Faculty Advisor: Muriel Medard
Mentor(s): Ying Cui
Contact e-mail: medard@mit.edu, yingcui@mit.edu
Research Area(s): Communications
Project Description: An UROP position is available in the Network Coding and Reliable Communications Group at EECS, working with professor Muriel Médard, post-doctoral associate Dr. Ying Cui and several collaborators in other universities. The aim of this project is to find a general, distributed and low-complexity method to design network codes in the non-multicast case for arbitrary networks. It will have a great impact on network performance improvement for today’s Internet and future network design. We have already made some progress and obtained some early results. We would like to evaluate the performance of the current theoretical design using numerical experiments. Based on the simulation results, we will further improve the design and the theoretical analysis. We are looking for a self-motivated undergraduate with good mathematical background and basic programming skills to help us with the numerical experiments and further theoretical study.

Student Role: The student participating in this project will be responsible for conducting numerical experiments and algorithmic developments.

Background: An ideal candidate for this position will be a junior/senior with mathematical background in algebra, probability and basic graph theory or knowledge of networks. In addition, basic programming skills (Matlab, C) are required.
Reuse
"Development of Superconducting Nanowire-Based Electronics"
Faculty Advisor: Karl K. Berggren
Mentor(s): Karl Berggren
Contact e-mail: berggren@mit.edu
Research Area(s): Applied Physics, Circuits, Materials and Devices, Nanotechnology
Superconducting nanowires have been used for some time as single-photon detectors. These nanowires switch between a superconductive and normal-metal state based on the addition of a small increment of energy (a photon). We have recently discovered that a small current can also be used to stimulate the transition, resulting in a new superconductive three-terminal device which we call the nTron. The nTron has opened up a host of interesting circuit-design challenges in digital computing and sensor readout. Because device fabrication is simple, a UROP could learn the entire technology, from device design, to fabrication, to test, and have an opportunity to be one of the first designers for this new technology in the world. Reuse
"Programming Robotics for Automating Genome Engineering"
Faculty Advisor: Timothy Lu
Mentor(s):
Contact e-mail: timlu@mit.edu
Research Area(s): BioEECS, Computer Systems, Control
One of the major limitations in synthetic biology is the effort it takes to construct engineered strains, since most steps are carried out by hand. We are working to establish a pipeline for automated genetic engineering leveraging robotic systems and algorithms. We seek a talented student to work at the intersection of robotic programming and synthetic biology to help develop software to automate the high-throughput construction of genetic constructs and the introduction of these genetic modifications into target organisms. The student will work closely with Prof. Lu and his team to carry out the work, with the goal of having a deployable system by the end of the SuperUROP period and have a publishable paper. Programming Robotics for Automating Genome Engineering Reuse
"Evaluation Platform for Low Power Vision"
Faculty Advisor: Vivienne Sze
Mentor(s):
Contact e-mail: sze@mit.edu
Research Area(s): Artificial Intelligence, Computer Systems, Signals and Systems
We are developing energy-efficient hardware for computer vision algorithms such as object detection, recognition and tracking. This project involves developing a user-friendly simulation platform for benchmarking the complexity and performance of a collection of computer vision algorithms. For instance, the platform should automatically evaluate the performance metrics (e.g false positive rate, false negative rate, precision-recall curves) for a given algorithm under different configurations that map to different complexity constraints. This will aid in determining the various complexity-performance trade-offs to enable efficient hardware designs. Following the completion of the platform, the student will be involved in the complexity evaluation of various vision algorithms as well as proposing low power optimizations.

We're looking for a student who has programming experience with MATLAB/C/Python; familiarity with computer vision is a plus.
Evaluation Platform for Low Power Vision Reuse
"Language and Compiler for High Performance Physics Simulations"
Faculty Advisor: Saman Amarasinghe
Mentor(s): Fredrik Kjolstad
Contact e-mail: fred@csail.mit.edu
Research Area(s): Computer Systems
Physical simulations include large graph processing systems that are used in many systems such as 3D movies, games, weather simulation and robotics. We are developing a programmable system to help programmers quickly develop really fast timesteppers to simulate physics that will be optimized and run on desktops to supercomputers. Longer term we also hope to express iterative simulation-like algorithms from other fields such as Machine Learning on big data.

We are looking for a student with compiler background to help us develop a front-end (lexer, parser) and a proof-of-concept back-end for our prototype. Longer term expect the student to actively partake in solving the research problems we face. Prerequisites: Strong C++ or Python programming background and algorithm/data-structure mastery. Familiarity with compilers a plus.
Reuse
"A single-cell pipettor"
Faculty Advisor: Joel Voldman
Mentor(s):
Contact e-mail: voldman@mit.edu
Research Area(s): Applied Physics, BioEECS
Pipettors are ubiquitous instruments in biological laboratories, able to deliver precise quantities of liquids into containers. What they are unable to do, however, is to deliver precise numbers of cells. Apportioning single cells or precise numbers of cells to wells is critical for basic biology, biotech, and pharma, yet is in large measure an unsolved problem. In this project, we’ll solve that problem by creating a “smart pipettor” that can deliver desired quantities of cells rather than liquids. Using a combination of existing pipettors, microfluidics, optics, and control strategies, we’ll create the first cell pipettor. A single-cell pipettor Reuse
"Statistics and experimental design for new model system genetic studies"
Faculty Advisor: David Gifford
Mentor(s):
Contact e-mail: gifford@mit.edu
Research Area(s):
Computational models and algorithms for executing genome-wide association studies in yeast. Most genetic studies in yeast cross a small number of strains (typically 2, rarely up to 4) and look at the variation among the "children" in that cross. However, that doesn't include as much genetic variation and doesn't correspond to what happens in nature as closely. We could analyze through simulation or statistical models the best way to conduct studies with up to 50 or 100 separate strains. This would be complicated by the fact that some yeast strains have related ancestry patterns, which can lead to false genetic hits. Depending on the timeframe, this could lead to biological experiments where we perform a multi-strain association study, guided by our computational predictions and supported by improved analysis algorithms. Reuse
"Machine learning algorithms for high-throughput genetic studies"
Faculty Advisor: David Gifford
Mentor(s):
Contact e-mail: giford@mit.edu
Research Area(s):
Developing a new algorithm and modeling approach for mapping multiple linked associations in pooled genetic studies. This would involve statistical model building and implementing a software approach to analyze very efficient sequencing studies that combine the DNA of thousands of individuals (usually yeast) into a single experiment. Our group has pursued directions that improve on current approaches to analyze this data, but further work is needed in refining the model, working out all the algorithmic details, and completing a stable and efficient implementation for a wide user base. Reuse
"Design and Simulation of multi-GHz Coupled MEMS Oscillators"
Faculty Advisor: Dana Weinstein
Mentor(s):
Contact e-mail: dana@mtl.mit.edu
Research Area(s): Applied Physics, Circuits, Materials and Devices
Coupled oscillators exhibits several interesting phenomena that make them suitable for unique applications. For example, the lift of states degeneracy resulting from the coupling creates multiple distinct states in the oscillator array. Using suitable excitation and detection techniques, this capability can be used to implement associative, human-like memory. Such application would require a very large array of fast oscillators. To maintain reasonable footprint and power consumption of such arrays, oscillators based on MEMS resonators naturally lend themselves as potential candidates for such implementation.
The aim of this project is to design, simulate and fully characterize arrays of coupled MEMS oscillators. The project is divided into 4 phases:
1. Developing a VerilogA compact model for a multi-GHz MEMS resonator.
2. Developing a detailed, time-domain, large-signal model for a Trans-Impedance-Amplifier (TIA). The resonator and TIA models are to be used together to actually design a MEMS based oscillator.
3. Verification and full characterization of the designed oscillator. This includes oscillator phase noise and start-up time estimation alongside with sensitivity to process variations in the MEMS resonator.
4. Studying the coupling dynamics of the designed MEMS oscillator for different coupling mechanisms. Synchronization and de-synchronization times as well as injection locking are of interest to this study.

By the end of the project the superUROP would have developed a deep understanding of MEMS resonators, MEMS based oscillators and coupling dynamics. The superUROP will also develop skills in VerilogA compact modeling, analog circuits simulation along with some tools from the Cadence IC design suites.
Design and Simulation of multi-GHz Coupled MEMS Oscillators Reuse
"Oscillator and amplifier design for active piezoelectric MEMS resonators"
Faculty Advisor: Dana Weinstein
Mentor(s):
Contact e-mail: dana@mtl.mit.edu
Research Area(s): Applied Physics, Circuits, Materials and Devices
The past decade has seen a rapidly growing demand for multi-GHz RF components such as filters and oscillators for potential applications in communications, navigation, inertial sensing and microprocessor clocking etc. Micromechanical resonators have been demonstrated to be advantageous over traditional LC tanks and quartz crystals due to their small size and weight, low power consumption and high Quality factor (Q~10000) for such applications. Piezoelectric materials (such as AlN, LN, PZT etc.) have been widely studied to provide the drive and sense mechanism in resonators due to their low insertion loss. Recent efforts in our group have focused on integrating an active sensing element such as a transistor into a piezoelectric resonator to reduce the input to output feedthrough which allows scaling to high frequencies.

This project focuses on the design of RF front-end circuit topologies that take advantage of the high-Q, low insertion loss and non-linear operation of the active piezoelectric resonator. You will construct electrical equivalent models, analyze and simulate the expected performance and non-linearities that may be leveraged for new circuit topologies. You will also identify and implement at least one such topology (such as an amplifier or an oscillator) in a simulation environment such as Cadence.

Some experience with analog circuit modeling and Cadence/SPICE or similar simulation environments will be useful. This project would be original research and has the opportunity to generate publishable results.
Reuse
"Deep learning on the genome"
Faculty Advisor: David Gifford
Mentor(s):
Contact e-mail: gifford@mit.edu
Research Area(s):
Machine learning project; convolutional neural nets with max-pooling have now become one of the dominant modeling paradigms in computer vision and natural language processing benchmarks. We have characterized some successful shallow architectures that can model regulatory structures in the genome and hope to use recent training techniques for deep nets to construct interpretable, deep architectures that can go directly from sequence to epigenetic structure. Reuse
"Direct vs indirect binding detection in ChIP-seq"
Faculty Advisor: David Gifford
Mentor(s):
Contact e-mail: gifford@mit.edu
Research Area(s):
statistical modeling project; Detection of transcription factor binding by ChIP-seq followed by computational analysis is now a relatively solved problem; however one major open problem is identifying which binding sites represent direct protein-DNA interaction and which are the result of indirect protein-protein interactions. Through combining various epigenetic assays such as DNAse-seq and Histone modification ChIP we hope to construct a reliable model of direct DNA binding that allows us to understand the regulatory structure of the genome. Reuse
"Sketch Synthesis of Algorithmic Choices"
Faculty Advisor: Saman Amarasinghe
Mentor(s): Shoaib Kamil
Contact e-mail: skamil@mit.edu
Research Area(s): Computer Systems
The Sketch synthesis system enables "computer-aided programming" by letting users specify an incomplete program and automatically generating the missing parts using program synthesis. This is particularly useful for complicated low-level pieces of code that are often performance-critical. The current system only guarantees that the generated code is correct, not necessarily performant. In this project, you will implement a system that synthesizes a large number of correct implementations from an incomplete sketch; these versions can each be run to find the best-performing one. This project requires knowledge of Java, some understanding of concepts from 6.172, and some knowledge of compilers as you will be modifying the existing Sketch compiler. Experience with program synthesis and Sketch is a plus, but is not required. Reuse
"Parallel Code Generation"
Faculty Advisor: Saman Amarasinghe
Mentor(s): Shoaib Kamil
Contact e-mail: skamil@mit.edu
Research Area(s): Computer Systems
As supercomputers are becoming more and more complex in terms of their hierarchical designs and interconnection networks, the problem of determining optimal data distributions for parallel programs is becoming more difficult. Much work in auto-parallelizing compilers has tried to apply simplified machine models to generate auto-parallelized code for distributed machines, but these models too simple to always generate the best data distribution. Instead, we're working on a system that automatically generates all possible data distributions from parallel code. The student will assist in writing candidate programs as well as possibly helping implement the system in LLVM. This project requires familiarity with parallel/distributed programming, compilers, and the LLVM infrastructure. Reuse
"OpenTuner"
Faculty Advisor: Saman Amarasinghe
Mentor(s): Shoaib Kamil
Contact e-mail: skamil@mit.edu
Research Area(s): Artificial Intelligence, Computer Systems
Autotuning is a technique whereby programs with many parameters that control performance can be tuned automatically. OpenTuner is the first generalized framework for allowing users to autotune their own programs by applying an ensemble of machine learning techniques to search for parameters that obtain the best performance. With OpenTuner, developers express the search space of their problem using a variety of parameter types; these parameters are searched over during autotuning. One common issue is that the efficacy of search often depends on the particular encoding of the search space. This project is to build a tool for OpenTuner that helps users determine which of a set of candidate encodings will be most amenable to searching over. This project requires familiarity with Python, some systems programming, and ideally some experience with machine learning. Reuse
"Designing for offline tolerance in mobile apps"
Faculty Advisor: Lalana Kagal
Mentor(s):
Contact e-mail: lkagal@csail.mit.edu
Research Area(s): Computer Systems
This project deals with the problem of limited connectivity in mobile apps. It involves designing an offline mode for apps, developing a simple proof-of-concept messaging application that is offline tolerant, and finally developing offline functionality in our app framework [1] that would allow a developer to easily employ an offline mode in any of their applications.

[1] App Inventor, Link
Designing for offline tolerance in mobile apps Reuse
"Customizable Mobile Geofencing"
Faculty Advisor: Lalana Kagal
Mentor(s):
Contact e-mail: lkagal@csail.mit.edu
Research Area(s): Computer Systems
The ubiquity of handheld computing technology has been found to be especially useful in disaster management and relief operations. Our focus is to enable developers to quickly deploy mobile applications that take advantage of key sources such Twitter feeds, Facebook posts, and government data, that empower citizens to contribute via crowdsourcing, and that provide up-to-date information to decision makers.

As part of this project, we are interested in incorporating geofencing capability into our framework to enable users to define a geographic area of interest. This could be done by decision makers to get information about a certain region or by citizens to outline the affected area. Google has recently released a geofencing
library ( Link ) for Android. This project involves investigating how the library could be incorporated into our app framework.
Customizable Mobile Geofencing Reuse
"Privacy concerns of mobile apps"
Faculty Advisor: Lalana Kagal
Mentor(s):
Contact e-mail: lkagal@csail.mit.edu
Research Area(s): Computer Systems
With the advent of “smart” mobile phones and wireless data communication technology, the pace at which people generate, access, and acquire data has accelerated significantly. Mobile phones are now rich deposits of memories and information that chronicle peoples’ lives. As a result, mobile privacy is rapidly emerging as a contested field of study for mobile developers, industry leaders and the public.

There are several projects available that address different aspects of mobile privacy including the development of an app store that allows users to search for apps that comply with their privacy preferences; performing user studies to gauge interest/concern around privacy notices of apps; and supporting the development of policy-compliant mobile apps.
Privacy concerns of mobile apps Reuse
"Characterization of High-Field Terahertz Generation Systems"
Faculty Advisor: Franz X. Kaertner
Mentor(s): Koustuban Ravi and Wenqian Ronny Huang
Contact e-mail: kaertner@mit.edu
Research Area(s): Applied Physics
Developing efficient electromagnetic sources with extremely high peak fields in the frequency range 0.1-10 THz, known as the Terahertz (THz) region remains elusive. These unique sources will be critical for understanding many important phenomena in material science, chemistry and biology [1]. In addition, their properties will open up a new paradigm of applications in the area of compact particle accelerators, imaging and medical therapy. Recently, we have demonstrated THz sources with world record generation efficiencies [2]. A key impediment to further increasing power levels is an insufficient reporting of material properties used for THz generation in the literature. This gap in knowledge deeply impacts the optimal design of these sources. In this project, the student will assist in accurate measurements of these properties and will develop skills in optical setup construction, optical measurements and numerical simulations. Characterization of High-Field Terahertz Generation Systems Reuse
"Deep Neural Networks for Speech Processing"
Faculty Advisor: Jim Glass
Mentor(s):
Contact e-mail: glass@mit.edu
Research Area(s): Artificial Intelligence
The CSAIL Spoken Language Systems Group has been investigating the use
of Deep Neural Networks (DNNs) for automatic speech recognition, as
well as speaker, language, and dialect recognition. The SuperUROP on
this project will join a team of graduate students and research
scientists to further develop the DNN functionality, as well as create
a DNN toolkit that can be used by other students for DNN-based speech
research. The opportunity exists to begin this research over the
summer as a UROP, and continue the SuperUROP in the fall.
Reuse
"Speech Interface for Food Diary"
Faculty Advisor: Jim Glass
Mentor(s):
Contact e-mail: glass@mit.edu
Research Area(s): Artificial Intelligence, Graphics and Human-Computer Interfaces
In collaboration with nutritionists at Tufts University, the CSAIL
Spoken Language Systems group is developing a nutrition-based
application to support spoken descriptions of food consumption (e.g.,
what I had for breakfast, etc). The SuperUROP on this project will
join a team of graduate students and research scientists developing
the speech prototype, will assist in crowdsourced data collection and
annotation activities. Depending on the student's interest,
opportunities exist to participate in language and dialogue aspects of
the project, or help develop a mobile speech interface on an Android
and/or iOS device.
Reuse
"Speech Interfaces for Smartphones"
Faculty Advisor: Jim Glass
Mentor(s):
Contact e-mail: glass@mit.edu
Research Area(s): Graphics and Human-Computer Interfaces
The CSAIL Spoken Language Systems group has developed many prototype
speech interfaces that provide information about weather, flights,
movies, restaurants etc., that can run in a Chrome browser
environments on desktop or mobile devices. This project involves the
refinement of the audio interface to upgrade it to the newly available
HTML5 protocol. Applicants should have some background in HCI, web
programming, and familarity with web protocols. Experience with
Android and iOS would also be helpful.
Reuse
"On-Line Spectrogram Reading Tutorial"
Faculty Advisor: Victor Zue, Jim Glass
Mentor(s):
Contact e-mail: glass@mit.edu
Research Area(s): Graphics and Human-Computer Interfaces, Signals and Systems
This project will involve the development of an on-line version of a
speech spectrogram reading course that teaches students the
acoustic-phonetic properties of American English, and their
spectral-temporal realization in a speech spectrogram. The project
can leverage existing resources that have been previously developed
for the course, but will involve the design and implementation of an
on-line version. Applicants should have some background in HCI, web
programming, and ideally, some signal processing, and an interest in
speech. Option to start as a regular UROP over the summer, and
continue into the superUROP in the fall.
Reuse
"BRINGING DOWN THE NETWORK"
Faculty Advisor: Prof. Vincent Chan
Mentor(s): Matt Carey
Contact e-mail: chan@mit.edu, mcarey@mit.edu
Research Area(s): Communications, Computer Systems
The objective of this study is to explore known network-disabling attacks and then to identify novel approaches to bringing down the networked system.

As network access has evolved to include new access technologies, traditional Internet protocols have been adapted incrementally to satisfy emerging challenges. Near-term advances, such as large optical flows and multihop wireless, will push this incremental-change approach to the breaking point and require the development of a clean slate architecture. This greenfield design presents the unique opportunity to architect a system robust to known attack planes, as well as to posit future vulnerabilities and proactively shore up weaknesses. Our interest extends from terminal hacking for data extraction to large-scale, coordinated efforts to inhibit the proper operation of the network.

This SuperUROP program will begin by categorizing attacks both in the wild and discussed in current literature. From here, the study will proceed with the definition of a network model and analysis of the effect and scope of new attack proposals. The interested researcher will have a passion for exploiting protocol weaknesses and characterizing chaos (in a controlled, theoretical framework!). Simulation opportunities may exist. A candidate with strong problem formulation, analysis skills, and exposure to 6.02 and 6.041/6.042 desired.
BRINGING DOWN THE NETWORK Reuse
"TRANSPORT LAYER PROTOCOL FOR LARGE OPTICAL “ELEPHANT” FLOWS"
Faculty Advisor: Prof. Vincent Chan
Mentor(s): Henna Huang
Contact e-mail: chan@mit.edu, hphuang@mit.edu
Research Area(s): Communications, Computer Systems
In data networks, the transport layer is an end user peer process. A generic transport layer protocol provides congestion control, bandwidth matching between transmitter and receiver, and the reliable delivery of data. Currently, the Transport Control Protocol (TCP) is the dominant protocol in practice. The bulk of Transport Layer Protocol research has been to modify TCP. However, the need for a disruptive change becomes apparent as data rates and data volume per transaction continues to increase. TCP has serious performance problems generally for large bandwidth-delay product networks and those with high contiguous packet loss. Adopting incremental changes to TCP alone is insufficient to combat the drastic performance degradation in future networks.

The goal of this project is to implement and analyze the performance of a new transport layer protocol for scheduled flow switched optical networks. Flow control and fair resource allocation are done via scheduling. Reliable delivery of data will be realized through a combination of framing, forward error correction, and automatic repeat request (ARQ). This new protocol will dramatically decrease the delay of large data transfer applications such as in data centers and video content delivery networks. Exposure to concepts from 6.02 and 6.041 is helpful.
TRANSPORT LAYER PROTOCOL FOR LARGE OPTICAL “ELEPHANT” FLOWS Reuse
"Acoustoelectric effects in piezoelectric MEMS device"
Faculty Advisor: Dana Weinstein
Mentor(s):
Contact e-mail: dana@mtl.mit.edu
Research Area(s): Applied Physics, Materials and Devices, Nanotechnology
This project involves investigation of the acoustoelectric effect and its application in MEMS devices composed of piezoelectric semiconductor materials (e.g. III-V semiconductors such as GaN). The acoustoelectric effect originates from the interaction between electrons and acoustic phonons. GaN’s large piezoelectric coefficients and the possibility of achieving large free carrier density make it a good material option for studying the physics at the intersection of mechanical, electrical and even optical domains. The following picture shows a Surface Acoustic Wave (SAW) device, which provides the perfect platform for acoustoelectric effect to take place (source: Link Acoustoelectric effects in piezoelectric MEMS device Reuse
"Optimization of effective electromechanical coupling coefficient for MEMS resonators"
Faculty Advisor: Dana Weinstein
Mentor(s):
Contact e-mail: dana@mtl.mit.edu
Research Area(s): Applied Physics, Materials and Devices, Nanotechnology, Numerical Methods
MEMS resonators are great candidates for timing and RF signal processing. One particularly exciting application is the RF bandpass filter, which is a crucial component in radio front ends. Critical specifications for a MEMS RF bandpass filter include off-band rejection ratio and bandwidth. Both are directly related to the effective electromechanical coupling coefficient (k2) of the resonating device, which depends strongly on design. This project’s goal is to design a pipeline for calculating and optimizing k2 based upon basic device structure and resonator mode shape. It involves numerical simulation and generating sophisticated equivalent circuit models for the MEMS resonator, which could be applied to common resonator and driving electrode schemes. Optimized resonators will be fabricated and characterized within the group. Optimization of effective electromechanical coupling coefficient for MEMS resonators Reuse
"Nanophotonic Phased Arrays"
Faculty Advisor: Michael Watts
Mentor(s): Jie Sun
Contact e-mail: mwatts@mit.eud
Research Area(s): Applied Physics, Communications, Nanotechnology, Signals and Systems
Nanophotonic phased arrays have numerous applications ranging from free-space communications to chip-based LIDAR for automotive accident avoidance technologies to biological/neurological sensing. Implemented within a state-of-the-art integrated silicon photonic platform, very large-scale optical phased array can be made in a compact silicon chip to include millions of optical antennas to generate high-resolution far-field patterns. However, one of the issues in optical phased arrays is the existence of sidelobes which lead to cross-talk and limit the application of optical phased arrays in both communications and chip-based LIDAR. This Super-UROP project is to address this sidelobe problem. The UROP student will design an optical phased array using advanced numerical techniques to address this important problem and then implement the corrected design in our custom 300mm silicon photonics process. The ideal candidates should have basic knowledge in Signals and Systems (6.003) as well as electromagnetics (6.013). Nanophotonic Phased Arrays Reuse
"Subtext: A new programming paradigm"
Faculty Advisor: Daniel Jackson
Mentor(s): Jonathan Edwards
Contact e-mail: edwards@csail.mit.edu
Research Area(s): Computer Systems
Subtext [ Link ] is an experimental programming language to simplify building applications like websites and mobile apps. It does this by unifying previously distinct concepts: static/dynamic types, imperative/functional programming, objects/relations/documents, editing/testing/debugging. There are a number of research opportunities. One is to build a UI library that encapsulates HTML and CSS. The major research problem is to replace CSS with a sane combinator library. Another opportunity is in the IDE, which requires advanced UI programming to support live incremental code editing. The research problem here is either to use advanced incremental parsing techniques within a text editor, or to develop a structural editor that operates by transformation and refactoring of the AST. Reuse
"Using Capnography to Assess Asthma Severity and Response to Treatment"
Faculty Advisor: Thomas Heldt & George Verghese
Mentor(s): Becky Asher and Dr. Baruch Krauss (BCH)
Contact e-mail: verghese@mit.edu
Research Area(s): BioEECS
This study builds on lessons learned during an earlier SuperUROP project, and considerably expands its scope.

Capnography is the measurement of CO2 concentration in exhaled breath and is routinely carried out in a variety of clinical settings. However, capnography is currently used well below its potential to reveal the respiratory state of a patient.

To quantify the potential of capnography as an effort-independent noninvasive tool for continuous assessment of asthma severity and response to treatment, we will

(a) determine the relationship between changes in capnogram shape and changes in spirometry during methacholine challenge testing; and

(b) validate the relationship between asthma severity and capnogram shape in patients being treated for acute asthma exacerbation.

The project will be carried out with collaborators at Boston Children's Hospital, and will also involve data collection in the pulmonary function lab and the emergency department.
Using Capnography to Assess Asthma Severity and Response to Treatment Reuse
"Manufacturing of Hierarchical Carbon Nanotube Electrodes for Energy Storage"
Faculty Advisor: Luis Fernando Velasquez-Garcia
Mentor(s): A. John Hart
Contact e-mail: lfvelasq@mit.edu, ajhart@mit.edu
Research Area(s): Energy, Materials and Devices, Nanotechnology
The high-throughput fabrication of nanostructured materials with controlled morphology is essential to the implementation of advanced energy storage and power conversion systems at low cost. Through miniaturization of the characteristic dimensions, the composite surface-to-volume ratio is greatly increased and therefore, diffusion-limited processes become more efficient. One of the most exciting possibilities is the exploration of systems with hierarchical nanostructure, i.e., arrays of elements with local and global nanoscaled structure. We are interested in combining two scalable manufacturing techniques to create hierarchically structured electrodes: interference lithography (IL) to batch-produce arrays of 250nm-level features, and chemical vapor deposition (CVD) to grow ultra-long dense vertical carbon nanotubes (CNTs). The templates made using IL will be used to define catalyst pads composed of nanosized particles that serve templates for CNT growth. After making the structures we plan to characterize their electrical and electrochemical performance.
The candidate should have strong interest in micro/nanofabrication; hands-on experience in cleanroom fabrication, materials synthesis, and electrical characterization is a plus. The project will require considerable ingenuity and perseverance, and would ideally result in a conference or journal publication at the end of the year. The student will collaborate with both research groups, and therefore be expected to communicate fluently across disciplines.
Interested students should contact Professor John Hart at ajhart@mit.edu and Dr. Luis Velásquez-García at lfvelasq@mit.edu. Please include in your initial email your resume in PDF format.
Manufacturing of Hierarchical Carbon Nanotube Electrodes for Energy Storage Reuse
"3D Printed Multiplexed Electrospinning Sources for High-Throughput Additive Manufacture of Nanostructured Composites"
Faculty Advisor: Luis Fernando Velasquez-Garcia
Mentor(s):
Contact e-mail: lfvelasq@mit.edu
Research Area(s): Applied Physics, Energy, Materials and Devices, Nanotechnology
Electrospinning is the process to generate nanofibers using strong electric fields and polar, high-molecular weight polymer solutions. Electrospinning is the only known process that can generate nanofibers of arbitrary length with controlled morphology, and it has unmatched versatility as it can create non-woven/aligned mats of metallic, dielectric, and semiconducting fibers. However, the practical application of these nanofibers is limited to high-end products because of the very low throughput of standard electrospinning sources. Increasing electrospinning throughput by multiplexing emission sources has been actively investigated for over a decade, including MEMS versions [1], [2] that consume less power and operate at lower voltage than macrosized sources. We recently demonstrated a novel MEMS-based modular architecture for nanofiber generation [2] that circumvents the problems associated with miniaturization of these sources: clogging, large operational pressure, and lack of field enhancers to further reduce the operational voltage and localize the emission. We recently realized that the characteristic dimensions of optimized devices (roughness, height, aspect ratio) line up nicely with the resolution capabilities of high-end 3D printers. This project explores the development of multiplexed electrospinning sources using additive manufacture while bypassing the use of expensive cleanrooms.
The candidate should have strong interest in micro/nanofabrication;hands-on experience on high voltage experiments and/or experimental characterization of micro/nanodevices is a plus. The work will require considerable ingenuity and perseverance, and would ideally result in a conference or journal publication at the end of the year.
Interested students should contact Dr. Luis Velásquez-García at lfvelasq@mit.edu. Please include in your initial email your resume in PDF format.
References:
[1] Y. Srivastava, M. Marquez, and T. Thorsen, “Multijet electrospinning of conducting nanofibers from microfluidic manifolds” J. Appl. Polym. Sci. vol. 106 pp. 3171-3178, 2007.
[2] P. Ponce de Leon, F. A. Hill, and L. F. Velásquez-García, “Batch-microfabricated arrays of electrospinning emitters for high throughput generation of nanofibers”, Technical Digest 12th International Workshop on Micro and Nanotechnology for Power Generation and Energy Conversion Applications (PowerMEMS), Atlanta GA, USA, pp. 227 – 230, December 2-5, 2012.
3D Printed Multiplexed Electrospinning Sources for High-Throughput Additive Manufacture of Nanostructured Composites Reuse
"The acoustics of child speech"
Faculty Advisor: Stefanie Shattuck-Hufnagel
Mentor(s): Elizabeth Choi
Contact e-mail: sshuf@mit.edu
Research Area(s): Communications, Computer Systems, Signals and Systems
SuperUROP proposal – child data 2014 2/18/2014 Jeung-Yoon Choi
The Speech Communication Group is working on analyzing children’s speech, from both typically and atypically developing children, using the landmark/acoustic cue system, developed in our lab. Landmarks are robustly-detectable locations in the signal that indicate classes of speech sounds, i.e. vowels, glides, and consonants. Other acoustic cues near landmarks specify other distinctive features that further define each sound. The specific realizations of landmarks and acoustic cues are strongly impacted by prosodic events. Speech analysis using this approach applies to a wide range of tasks, including clinical speech analysis, and speech recognition/synthesis.
The SuperUROP candidate will be involved in labeling of children’s speech with landmarks and prosodic events using Praat, and with analysis of those labels using decision tree software and/or statistical analysis packages. He/she will also be involved in software development, and should be able to program in Python or similar. Familiarity with phonetics/phonology is a plus.
Reuse
"Reducing false alarms in critical care"
Faculty Advisor: Thomas Heldt
Mentor(s): Thomas Heldt
Contact e-mail: thomas@mit.edu
Research Area(s): BioEECS, Signals and Systems
Patients in intensive care are monitored very closely through a variety of bedside devices. Each device has the capability of alerting the clinical staff to potentially serious conditions. Unfortunately, these devices issue an overwhelming number of false or 'nuisance' alarms. These alarms lead to desensitization of the clinical staff to the alarms and potential neglect of truly life-threatening conditions. By analyzing multiple signal streams, we seek to develop signal-processing approaches to reducing the high incidence of false alarms. This work will be developed in close collaboration with clinical colleagues. Reducing false alarms in critical care Reuse
"Speech Analysis System based on feature cues"
Faculty Advisor: Stefanie Shattuck-Hufnagel
Mentor(s): Elizabeth Choi
Contact e-mail: sshuf@mit.edu
Research Area(s): Communications, Computer Systems, Signals and Systems
SuperUROP proposal 2/18/2014 Jeung-Yoon Choi

The Speech Communication Group is developing a speech analysis system modeled on human speech understanding, based on acoustic cues called landmarks (LMs). LMs are robustly-detectable locations in the signal that indicate classes of speech sounds, i.e. vowels, glides, and consonants. Other acoustic cues near LMs specify distinctive features (DFs) that further define each sound. The specific realizations of LMs/DFs are strongly impacted by prosodic events. Speech analysis using this approach applies to a wide range of tasks, including clinical speech analysis, and speech recognition/synthesis.
Most of the individual components in the system are nearing completion (from 2 SuperUROP projects through 2012-2014), and development of the remaining modules and integration of the complete system will be carried out in 2014-2015. Ongoing analysis of a variety of speech data will guide how modules are deployed in the overall system.
The SuperUROP candidate will choose among several subprojects needed to complete this system. He/she will be involved in software development, and should be able to program in either Java, Python, Matlab, C, or similar. Completion of a signal processing and/or machine learning course is encouraged.
Reuse
"Diagnosing Anemia using the Electrical Properties of Blood"
Faculty Advisor: George Verghese & Collin Stultz
Mentor(s):
Contact e-mail: verghese@mit.edu
Research Area(s): Applied Physics, BioEECS
Using the techniques of impedance spectroscopy and dielectric characterization, we seek to use the high-frequency electrical properties of a small sample of blood to determine parameters relevant to anemia.

This project involves designing a setup to make the high-frequency electrical measurements and developing a theoretical model to relate the measured electrical properties to the parameters of interest.
Diagnosing Anemia using the Electrical Properties of Blood Reuse
"Monitoring Procedural Sedation"
Faculty Advisor: George Verghese & Thomas Heldt
Mentor(s): Becky Asher & Dr. Baruch Krauss (BCH)
Contact e-mail: verghese@mit.edu
Research Area(s): BioEECS, Signals and Systems
Physicians widely administer sedative and analgesic drugs to carry out procedures such as fracture reductions, spinal taps, and CT scans. Although such procedural sedation is relatively safe and effective, these drugs can slow spontaneous breathing (respiratory depression) or stop breathing altogether (apnea). Current patient safety monitoring systems identify such serious adverse events at the time they occur, but provide no form of advance warning. Working with a collaborator at Boston Children's Hospital, we are interested in exploring innovative ways to predict respiratory adverse events during procedural sedation, using high-resolution vital sign monitoring (including capnography, which measures the CO2 concentration in exhaled breath), coupled with computational models of the relevant human physiology. Monitoring Procedural Sedation Reuse
"Mobile Medical Platform"
Faculty Advisor: Anantha Chandrakasan
Mentor(s):
Contact e-mail: anantha@mtl.mit.edu
Research Area(s): Circuits, Computer Systems
The goal of this project is to develop a complete mobile health platform for use in rural areas. The vision of this project is to interface medical sensors (ECG, glucose monitoring, blood pressure monitoring, etc.) to a mobile device (e.g., tablet). The project will also involve developing a software environment for easy data collection, analysis and storage. This will include documenting vaccination records for children.
The project will include the following components:
1) Developing the hardware sensor devices for medical monitoring
2) Developing an app for user-friendly data collection and analysis
3) Efficient data storage and access (including patient identification)
Reuse
"Investigating Differential Privacy: Big Data and its Access Issues"
Faculty Advisor: Una-May O'Reilly
Mentor(s): Kalyan Veeramachaneni
Contact e-mail: alfa-apply@csail.mit.edu
Research Area(s): Artificial Intelligence, Theoretical Computer Science
With big data comes big responsibilities! There is a delicate balance between privacy and utilization of data.  One open question is how to allow access to data while maintaining privacy. We are interested in algorithmic approaches that can support open access to data while still conforming to privacy policies. The project goal is to concretely determine, in the context of a real-world dataset, how different anonymization techniques such as replacement-with-substitution, suppression, generalization or perturbation impact the investigative scope of data mining questions and sharing policies by data owners. (image from Link Investigating Differential Privacy: Big Data and its Access Issues Reuse
"Scalable Machine Learning via Volunteer Computing"
Faculty Advisor: Una-May O'Reilly
Mentor(s): Kalyan Veeramachaneni
Contact e-mail: alfa-apply@csail.mit.edu
Research Area(s): Artificial Intelligence
Have you heard of Einstein@Home or DrugDesign@Home? They are two examples of volunteer computing which is a means of harnessing idle cycles throughout the world toward solving important problems. Volunteer compute implies special resource management challenges and algorithm requirements. The ALFA group has a system under development that can build a large scale probabilistic model via volunteer computing. To be able to exploit these ad-hoc resources efficiently, we have strategies for efficient distribution of data, relaxed distributed algorithms, and centralized but asynchronous coordination. Our current framework executes unsupervised learning. It builds a latent variable model which can identify clusters in data. Join us in expanding to build other kinds of latent variable models and learn about volunteer computing systems, machine learning and data mining. Scalable Machine Learning via Volunteer Computing Reuse
"Machine Learning and Data Science for Medicine"
Faculty Advisor: Una-May O'Reilly
Mentor(s): Kalyan Veeramachaneni
Contact e-mail: alfa-apply@csail.mit.edu
Research Area(s): Artificial Intelligence
This project centers on the interpretation of large quantities of physiological waveforms collected from humans. The waveforms can range from arterial blood pressure readings from an invasive probe, collected in an ICU setting to heart rate data collected from mobile devices which fulfill our quest for a "quantified self". We are addressing challenges related to identifying meaningful features in this kind of data, engineering agile signal targeting, and integrating machine learning, *at scale* in order to achieve a vision of knowledge mining and forecasting at one's finger tips. All the major components of the data science process need attention. A super-urop will work within a team and reference the ALFA group's large suite of machine learning algorithms and statistical methods which range from rule based classification, to logistic regression and latent variable models. Machine Learning and Data Science for Medicine Reuse
"Educational Data Mining to Analyze MOOCs"
Faculty Advisor: Una-May O'Reilly
Mentor(s): Kalyan Veeramachaneni
Contact e-mail: alfa-apply@csail.mit.edu
Research Area(s): Artificial Intelligence
This project investigates ways to improve delivery of online education with new technologies and tools. With the rise of MOOCs, there is a wealth of behavioral data now available that captures in an explicit, detailed way how students study online. Our goal is to investigate fundamental questions in a MOOC or blended teaching (a.k.a digital education) setting. Some example involve understanding how well a student has learned, how affordances in the course architecture provide motivation, how resource usage correlates to achievement, and how emergent social structures signal collaborative learning. A super-urop project within this project will focus in on a particular research goal that contributes to overall goals. Educational Data Mining to Analyze MOOCs Reuse
"Modular Compilation of Bluespec SystemVerilog (BSV)"
Faculty Advisor: Arvind
Mentor(s): Murali Vijayaraghavan
Contact e-mail: arvind@mit.edu
Research Area(s): Computer Systems
Ability to refine a system in a modular fashion is essential for designing complex systems. Modular refinement requires modular compilation. The goal of this project is to implement a modular compilation scheme for Bluespec SystemVerilog (BSV) presented in Link

Currently, the BSV modules with interface parameters (i.e. parameters which are the methods defined in other modules) are inlined because the scheduling constraints of these parameters are not known at the time of compiling such a module. By providing the conflict information regarding the interface parameters, it is possible to do separate compilation of each such module. The scheduling based on our scheme provably obeys one-rule-at-a-time semantics.

We are looking for a student who is interested in compilers and knows Haskell or some other functional language. Some knowledge of Bluespec hardware description language may help, but is not required.
Modular Compilation of Bluespec SystemVerilog (BSV) Reuse
"Hardware-accelerated Map-Reduce on Distributed Flash Storage"
Faculty Advisor: Arvind
Mentor(s): Sang Woo Jun
Contact e-mail: arvind@mit.edu
Research Area(s): Computer Systems
This project builds on top of our ongoing hardware-accelerated distributed flash storage project, BlueDBM. (link: Link BlueDBM implements a networked flash controller inside a programmable FPGA, and provides an extremely low latency network connection between controllers. The platform provides abstract network and storage interfaces, and users can use these interfaces to implement hardware accelerators inside the FPGA
controller.

The goal of this project is to implement a framework for users to write programs in the map-reduce style. (link: Link The program will be compiled into hardware accelerators that will run inside the FPGA-based flash controller. The map-reduce program in the controller will run with a low performance overhead, because processing can be done on the data as it is being streamed out from the flash storage, without any additional job invocation or data copying.

Some knowledge of the Bluespec hardware description language and map-reduce will be useful, but is not required.
Hardware-accelerated Map-Reduce on Distributed Flash Storage Reuse
"Development of Phase Plate for Intense and Coherent Electron Vortex Beam Generation"
Faculty Advisor: Karl K Berggren
Mentor(s):
Contact e-mail: berggren@mit.edu
Research Area(s): Applied Physics, Materials and Devices, Nanotechnology
The generation of vortex electron beams, spiraling like a tornado, in existing electron microscope can offer completely new possibilities for electron microscopy and spectroscopy such as displaying objects and investigating material specific properties, particularly in angular resolved electron energy loss spectroscopy and mapping of moment with atomic resolution [1, 2]. Currently, every vortex generator, phase plate, has been fabricating by focused ion beam (FIB) technology, therefore there exist some issues originated by nature of ion sputtering to be overcome, including the prevention of the thermal diffuse scattering and radiation effect from dislocated crystal phase plate fabricated by FIB, in order to generate the most intense, clean and controllable electron vortex beam as well as high angle diffraction [3, 4]. With the help of helium ion microscope (HIM), FIB fabricated phase plate will be able to modulate and improve the geometry and property of vortex beam generator for new microscopy and spectroscopy applications.

[1] J Verbeeck, et al., Nature 467 (2010) 201.
[2] B J Mcmorran, et al., Science 331 (2011) 192.
[3] E Humphery, et al, The 15th European Microscopy Congress, Mancester Central, UK, 16th-21st Sep 2012.
[4] V Grillo, et al., Applied Physics Letters 104 (2014) 043109.

This project will deal with design, fabrication (gallium and helium ion beams), and analysis (electron microscopy) of phase plate.
Development of Phase Plate for Intense and Coherent Electron Vortex Beam Generation Reuse
"Ultrafast High-Field Laser Source Abstract:"
Faculty Advisor: Franz X. Kaertner
Mentor(s): Peter Krogen
Contact e-mail: kaertner@mit.edu
Research Area(s): Applied Physics
There are many applications in spectroscopy and nonlinear optics that require access to extremely short (on the order of a single optical cycle, mere femtoseconds long) and extremely energetic (with peak powers approaching a terawatt) pulses. We are looking for a SuperUROP to take up a variety of circuit design and fabrication tasks necessary to increase the precision with which these ultrafast and intense waveforms of light can be sculpted for applications, and to test the operation of newly built components in the laser system. A possible project is to build an extremely precise timing stabilization system with sub-femtosecond (i.e., “attosecond" time scale) class timing jitter, which will involve electrical, electromechanical, and optical design work. Ultrafast High-Field Laser Source Abstract: Reuse
"Attosecond Electronics on a Chip"
Faculty Advisor: Franz X. Kaertner
Mentor(s): Billy Putnam
Contact e-mail: kaertner@mit.edu
Research Area(s): Applied Physics
To study the dynamics of fast physical processes, we need ultrafast probes. Franz Kartner's group in RLE has been developing ultrafast optical probes for over the past decade. We have developed laser sources producing optical pulses only a few femtoseconds (10-15 s) in duration. Such short pulses correspond to only a few oscillations of the optical electromagnetic field. These femtosecond pulse sources have helped unraveled the temporal dynamics of numerous physical, chemical, and biological systems. However, to push to the next frontier of ultrafast science we need to develop even shorter probe pulses: pulses only attoseconds (10-18 s) in duration. To this end, we are developing chips with carefully designed arrays of nanoplasmonic antennas that can concentrate the optical energy in femtosecond light pulses and produce bursts of electrons only attoseconds in duration. These nano-devices could allow for the future probing and study of temporal dynamics at the attosecond scale, all on a convenient chip-scale platform. We seek a superUROP candidate with a strong physics background to assist us in better understanding the operation of our nanoplasmonic electron sources through simulation and modeling work as well through experimental investigation. Attosecond Electronics on a Chip Reuse
"Electron energy-loss near-edge spectroscopy (ELNES) for electron beam lithography toward the atomic scale"
Faculty Advisor: Karl Berggren
Mentor(s):
Contact e-mail: berggren@mit.edu
Research Area(s): Applied Physics, Nanotechnology
Electron beam lithography (EBL) has facilitated the production of structures with dimensions of ~1 nm. Fundamental to the understanding of the resolution limit of EBL is the understanding of the interaction of high-energy beams of electrons with matter, in particular with electron beam resists.

In electron energy-loss near-edge spectroscopy (ELNES), inner shell electrons ionized by a high-energy incident electron beam may be excited to empty electronic energy levels if they do not have sufficient energy to escape to vacuum. The empty energy levels are related to the local electron distribution due to the atom and it's neighbors in a solid. Consequently, these energy levels, which are found in the ‘near-edge’ region of electron energy loss spectra, may provide information on the local bonding of an atom in a material. The SuperUROP will perform ELNES analysis of electron energy loss spectra, in conjunction with atomistic simulations, to explore chemical changes in high-resolution EBL resists such as hydrogen silsesquioxane (HSQ), thus allowing us to understand fundamental limits in nanofabrication by EBL.

In this project the student will develop an understanding of the fundamental interactions of high-energy electrons with matter, and receive practical training in the use of an atomic-resolution transmission electron microscope.
Electron energy-loss near-edge spectroscopy (ELNES) for electron beam lithography toward the atomic scale Reuse
"Understanding language to see the world, and seeing the world to understand language"
Faculty Advisor: Boris Katz
Mentor(s): Andrei Barbu
Contact e-mail: boris@csail.mit.edu
Research Area(s): Artificial Intelligence
Why is human vision so much better than machine vision? How do we describe what we see? How do you recognize an event that someone is describing? How do you learn language when all you hear are sentences with words you don't understand?

We are interested in understanding these questions by developing models that jointly address vision and language. These models should allow you to perform tasks that come naturally to children but are difficult for machines: describing what you see, seeing what is being described, asking questions about what you see, determining if what you are seeing is reasonable, using knowledge from language to change your mind about what you are seeing, etc.

This project aims to develop new models, extend current models to novel environments, or answer novel questions about language and vision.
Reuse
"Nanometer-scale InGaAs MOSFETs"
Faculty Advisor: Jesus A. del Alamo
Mentor(s): Dr. Alon Vardi
Contact e-mail: alamo@mit.edu
Research Area(s): Materials and Devices, Nanotechnology
This is a project to support the experimental demonstration of nanometer-scale InGaAs MOSFETs currently under research in our group at the Microsystems Technology Laboratories. The broader context of this project is the difficulties that Si MOSFETs are facing to scale down to gate dimensions below 10 nm. This threatens to slow down Moore’s Law and all the progress that it has brought to human society. The outstanding properties of III-V compound semiconductors, most notably InGaAs, offer a path forward. In our group, we are investigating planar, trigate and nanowire InGaAs MOSFETs. Common to all these devices is the need for a high-quality oxide/semiconductor interface using new high-dielectric constant dielectrics. This project is to experimentally investigate oxide/InGaAs MOS structures for future InGaAs MOSFETs. The research will involve fabrication of MOS structures in the nanofabrication facilities of the Microsystems Technology Laboratories and characterization of the resulting devices by means of electrical measurements. This project is for students that seek a hands-on experience in nanotechnology and nano-MOSFET design, fabrication and characterization. Nanometer-scale InGaAs MOSFETs Reuse
"Hardware and Runtime Support for Irregular Parallel Programs"
Faculty Advisor: Daniel Sanchez
Mentor(s):
Contact e-mail: sanchez@csail.mit.edu
Research Area(s): Computer Systems
New application areas like big-data analytics, machine learning, and data mining are shifting the focus of parallel computing from regular to irregular applications. While regular applications (e.g., dense linear algebra) are easy to analyze and parallelize at compile-time, in irregular applications tasks are dynamically created and dependencies among tasks cannot be determined in advance. Irregular applications can only be parallelized at run-time. Unfortunately, these applications require two runtime parallelization techniques that are expensive and scale poorly in current parallel architectures: speculative task execution and fine-grained communication and synchronization. For example, because of their overheads, state-of-the-art software runtimes cannot obtain any speedup on a discrete-event simulation application with 150x parallelism.

We are developing a combination of hardware support and runtime techniques to scale these parallel applications efficiently. By innovating at both the hardware and the software level, our goal is to achieve performance, scalability, and efficiency that neither hardware-only nor software-only techniques can provide. Because of the broad scope of the project, there are many areas where you can contribute, depending on your interests. For example, you can:

- Port challenging irregular parallel applications to use this parallel programming model and runtime.

- Develop new runtime techniques, for example, to predict what work should be executed speculatively (what tasks are unlikely to be dependent?), to minimize the amount of wasted work, maximizing performance and energy efficiency.

- Prototype the new hardware features required to support these runtimes efficiently.
Reuse
"Unified Placement of Memory and Computation in Large-Scale Multicores"
Faculty Advisor: Daniel Sanchez
Mentor(s):
Contact e-mail: sanchez@csail.mit.edu
Research Area(s): Computer Systems
Memory accesses and communication already limit the performance and energy efficiency of current multicore systems: a single main memory access costs the same energy of a thousand compute operations. What's worse, with more cores, these costs increase further. To ameliorate this problem, memories and caches adopt a non-uniform organization, providing fast access to physically close data, and slower access to far-away data. To scale efficiently, it is crucial that most data is placed close to where it is used. Unfortunately, hardware alone does a poor job at deciding where to place data, because it does not have enough information about what the software is doing. We've recently shown that a combination at hardware and OS-level software mechanisms can do a much better job at placing data close to computation, reducing network traffic by 10x transparently to the programmer. But a fundamental problem remains: some data is shared among many far-away threads, so we cannot place it close to the many threads where it's used.

To solve this problem, in this project you will explore the following question: what is the fundamental relationship between data and the computation that operates on that data in parallel applications? Specifically, you will start with a state-of-the-art parallel programming model and runtime, which already executes many of high-performance applications (e.g., raytracing, MapReduce, crypto, etc.) on current multicores, and will modify it to schedule both data and computation simultaneously. Instead of threads, this runtime uses short tasks that access limited, well-defined amounts of data. By dynamically deciding both where to place the data and the tasks that operate on that data, you will enable many challenging memory-intensive applications to scale to multicore chips with hundreds to thousands of cores --- by simply leveraging program structure, and transparently to the programmer.
Reuse
"Mixed Morphology Nanopatterns with Block Copolymer Self-assembly"
Faculty Advisor: Karl Berggren
Mentor(s): Sam Nicaise
Contact e-mail: berggren@mit.edu
Research Area(s): Materials and Devices, Nanotechnology
Interested in nanotechnology, materials for devices, and obtaining hands-on lab experience?

This project will work on block copolymer (BCP) self-assembly, a technology which produces nanoscale patterns and designs from the bottom up. We will investigate the blending of two different BCPs which form independent morphologies, nanocylinders and spheres. Next generation integrated circuits, sensors, bio-devices, and NEMS may be dependent on the advancement this technology. The placement of these two morphologies is expected to be controlled with top-down lithography. Lithographically defined templates of chemical layers can be patterned on the surface before BCP self-assembly. With this patterning, it’s possible to define exactly where the spheres and cylinders will respectively assemble. We expect that this work will culminate in a published research paper in collaboration with other authors.

Skills: wet chemistry, clean room instruments, electron-beam microscopy, thin films processing, creativity, self-motivation, constant and effective communication.

Background:
“Aligned Sub-10-nm Block Copolymer Patterns Templated by Post Array” ACS Nano 6(3) 2071 (2012)
“Sacrificial-Post Templating Method for Block Copolymer Self-Assembly” small 10(3), 418 (2014)

Students will work directly with Graduate Student Sam Nicaise, Professors Caroline Ross and Karl Berggren, and other researchers in the lab.

Questions can be sent to snicaise@mit.edu.
Mixed Morphology Nanopatterns with Block Copolymer Self-assembly Reuse
"Constrictions in Superconducting Nanowire Devices"
Faculty Advisor: Karl Berggren
Mentor(s):
Contact e-mail: berggren@mit.edu
Research Area(s): Applied Physics, Materials and Devices, Nanotechnology
Constrictions are hypothesized to be a major limiting factor for the reproducible fabrication of detectors based on superconducting nanowires, but we haven't observed any obvious structural defects in many of our "constricted" devices. It's possible that the constrictions are too small be observed with electron microscopy, that they are material defects, or that they exist between the substrate and the nanowire, but more data on the resistance and critical current of nanowires is necessary before we can narrow down the list of possibilities. This project will involve fabricating and testing some structures and analyzing data from previous devices to look for relationships among the device characteristics. Reuse
"App Development for Activity Recognition"
Faculty Advisor: Vivienne Sze
Mentor(s):
Contact e-mail: sze@mit.edu
Research Area(s): Artificial Intelligence, Signals and Systems
Activity recognition can play a significant role in providing feedback for a healthy life style as well as assist the sick and disabled. Data is typically collected from various sensors (e.g. accelerometers) and activity recognition is done using machine learning.

We are looking for a strong and motivated student to develop an app to enable annotation of accelerometer data that will be used in activity recognition. The main objective of the project is to develop an app that can efficiently store accelerometer data and provide an easy-to-use interface to annotate the data.

Developing the app will be just the first step; there will be an opportunity for the student to get deeper into research by analyzing the data and developing machine-learning algorithms for activity recognition.

Good C programming skills and experience in app development for iOS/android required; background (or interest) in machine learning is a plus.
App Development for Activity Recognition Reuse
"An Energy-Efficient FPGA Vision platform for UAV navigation"
Faculty Advisor: : Vivienne Sze & Russ Tedrake
Mentor(s):
Contact e-mail: sze@mit.edu, russt@mit.edu
Research Area(s): Artificial Intelligence, Circuits, Computer Systems, Signals and Systems
Description: Unmanned Aerial Vehicles (UAV) use a variety of imaging sensors to navigate and avoid obstacles. Real-time processing of the sensor data is necessary in order for the UAV to react quickly to changing environments; in particular, high speed vision will enable faster detection and allow more time for course correction. However, the energy available to process the data is limited as the weight of the battery must be kept to a minimum.

The goal of this project is to create an energy-efficient FPGA platform for vision processing on an UAV; specifically, we would like to produce hardware that would fly on a real airplane using vision-based control for high-speed autonomous flight ( Link ).

There are several opportunities to get involved in this project. One is to build an FPGA platform that can interface with high speed image sensors. Another is to explore various architectures to enable high speed vision processing on the FPGA.

Experience with FPGA, PCB and RTL design (e.g. 6.111) is a plus.
An Energy-Efficient FPGA Vision platform for UAV navigation Reuse
"Monte Carlo Simulation for Quantum Electron Microscope"
Faculty Advisor: Karl K. Berggren
Mentor(s):
Contact e-mail: berggren@mit.edu
Research Area(s): Applied Physics, Numerical Methods, Theoretical Computer Science
Interaction-free measurement with electrons was recently proposed to reduce radiation damage in electron microscopy and to build a quantum electron microscope (QEM). In a free-space approach to QEM, conventional transmission electron microscope (TEM) will be modified by adding a beam splitter and a storage ring. We proposed the idea of using a thin film crystal as the beam splitter. Therefore, QEM performance largely depends on the probability of decoherence events in the crystal beam splitter, including thermal diffuse scattering, plasmon excitation, inner-shell ionization, and so on.

The student’s task is to theoretically model the crystal beam splitter with decoherence. This includes calculating various inelastic scattering cross sections and hence decoherence probabilities, comparing the calculation with experimentally obtained data, and building a Monte Carlo model with calculated or measured decoherence probabilities to evaluate the performance of QEM with decoherence events.
Reuse
"A short-throw black-board camera"
Faculty Advisor: Fredo Durand
Mentor(s):
Contact e-mail: fredo@mit.edu
Research Area(s): Graphics and Human-Computer Interfaces
The goal of this project is to create a camera that can record writing on a black board from a very short distance. For this, we will use the optics of a short-throw projector and will replace the projecting element by a camera. Candidates should not be afraid of tinkering with hardware and have basic notions of optics. A short-throw black-board camera Reuse
"Video magnification with Halide"
Faculty Advisor: Fredo Durand
Mentor(s):
Contact e-mail: fredo@mit.edu
Research Area(s): Graphics and Human-Computer Interfaces
Video magnification is a new technique that can reveal tiny motion and
variations in regular videos. It can show the reddening of a face with
each heartbeat or make tiny motion visible, such as the mechanical
vibrations of a camera
Link
Link

The new version relies on a special kind of image wavelets, complex valued
steerable pyramids, and directly modifies the local phase of videos. This
dramatically reduces noise and enables new levels of magnification.

This super-urop project is a real-time extension of video magnification
that can work on handheld devices such as smart phones or tablets. For
this, we will use the new Halide language and compiler developed in our
lab. Halide is embedded in C++ but enables up to a 10x speedup for image
processing, while keeping code concise and readable.
Video magnification with Halide Reuse
"Authoring of Handwritten Video Lectures"
Faculty Advisor: Fredo Durand
Mentor(s):
Contact e-mail: fredo@mit.edu
Research Area(s): Graphics and Human-Computer Interfaces
We want to facilitate the authoring of handwritten educational video lectures and seek to combine the advantages of vector graphics and video editing. Handwritten videos are currently recorded using screen capture and regular sketching software, which makes editing challenging.
Our work focuses on the non-sequential and iterative creation of one particular style of educational content, handwritten video lectures.
Super urops are available on a variety of aspects of this work, such as speech recognition and transcript editing, handwriting beautification, and audio processing.
Authoring of Handwritten Video Lectures Reuse
"Dynamic walking with an electric humanoid"
Faculty Advisor: Russ Tedrake
Mentor(s):
Contact e-mail: russt@mit.edu
Research Area(s): Artificial Intelligence, Control
The Robot Locomotion group has been working hard on dynamic walking control algorithms for the hydraulic ATLAS humanoid robot as a part of MIT's entry in the DARPA Robotics Challenge. We have developed a number of new methods, and would now like to thoroughly evaluate them on the other humanoid robot in our lab - and electric robot named Hubo. In particular, we will design careful experiments to evaluate the performance differences due to the different actuation strategies. It should be an excellent opportunity to play with a state-of-the-art robot and algorithms.

Requires strong programming skills (ideally in MATLAB, C++, and a little Java). Comfort with hardware is also a plus; for instance we anticipate having to replace a faulty PC-104 board on the robot to get started.
Dynamic walking with an electric humanoid Reuse
"Portable Water Desalination"
Faculty Advisor: J. Han
Mentor(s):
Contact e-mail: jyhan@mit.edu
Research Area(s): Applied Physics, Energy, Materials and Devices
Water is increasingly scarce globally, and recent climate fluctuations strains the existing water supply model in many parts of the world. A small scale, portable and on-site water desalination system may be a solution to meet the ever-increasing and fluctuating demand for water for many uses, including human consumption and agriculture. In this project, we will develop a small scale water desalination system that can meet the economic and technical specification for rural and remote areas' water needs. We will base our ICP desalination methodology, which does not require any pretreatment of water, to enable such system and characterize the performance. Reuse
"Combinatorial Hardness of Approximation"
Faculty Advisor: Dana Moshkovitz
Mentor(s):
Contact e-mail: dmoshkov@mit.edu
Research Area(s): Theoretical Computer Science
Hardness of approximation for NP-hard problems is based on a deep theorem in Complexity Theory called the Probabilistically Checkable Proofs (PCP) Theorem. The theorem says that any mathematical proof can be compiled into a format that can be checked probabilistically by verifying only one random implication in the proof (instead of all implications). The goal of the project is to help facilitate research in hardness of approximation: make it easy for approximation algorithms researchers to use the PCP theorem with its different parameters, and make it easy for PCP researchers to find out about open problems in hardness of approximation. In particular, the following sub-projects are available: update an existing data base of approximation problems, read and survey existing combinatorial hardness results, try to solve some open problems in combinatorial hardness of approximation. Combinatorial Hardness of Approximation Reuse
"Standard Packaging Design for Cryogenic Photodetectors"
Faculty Advisor: Prof. Karl K. Berggren
Mentor(s):
Contact e-mail: berggren@mit.edu
Research Area(s): Materials and Devices
The Quantum Nanostructures and Nanofabrication (QNN) group at RLE is one of the world leading groups in the design and fabrication of Superconducting Nanowire Single Photon Detectors (SNSPDs). These devices represent one of the most promising technologies for applications like optical communications, low intensity sources characterization and spectroscopy, and middle infrared imaging. With more than ten years of experience the QNN group has created a state of the art fabrication process for detectors with internal detection efficiency > 80% and sub 10 ns reset time. The goal for the future is to create collaborations around the world to increase the use of SNSPD for cutting edge research.
The student’s task is to create a compact packaging for SNSPDs that minimizes the stray light hitting the detector, maximizes the light coupled to the detector, and is compatible with state of the art cryogenic systems. This project allows the student to acquire skills in mechanical designing and machining for optical applications, nanostructures designing, and cryogenic testing. The work will be supervised by Prof. Berggren and a graduate student in the QNN group.
Reuse
"Secure Distributed Multi-User Web Applications"
Faculty Advisor: Frans Kaashoek
Mentor(s): w. Nickolai Zeldovich
Contact e-mail: kaashoek@mit.edu
Research Area(s): Computer Systems
Your goal is to build secure multi-user distributed web applications. We have built infrastructure to make it easy for developers to build such applications, and like to test it with applications such a distributed calendar applications. Building an application is likely to require changes to the underlying infrastructure. This project requires knowledge of security and javascript. Ideally a student has taken 6.858. Reuse
"Graphene-based Strain Sensors"
Faculty Advisor: Jing Kong
Mentor(s):
Contact e-mail: jingkong@mit.edu
Research Area(s): Materials and Devices
Strain gauges are a ubiquitous sensor type that is used in many different applications where the mechanical deformation has to be analyzed. Conventional strain gauges use metallic thin films (relying on geometrical change of conduction path) or piezoelectric materials such as Si. However, these sensors are not suitable for emerging application areas, such as human motion detectors, force sensitive touch screens, or high resolution deformation analysis, which requires sensors that exhibit stretchability, transparency or can be applied to complex surfaces.

During the past couple of years, our group has demonstrated a novel class of strain gauges that combines cheap and scalable production with high sensitivity and novel, attractive properties using graphene. In this project, we will carry out further investigations to develop low cost reliable sensors on various types of substrates, focusing on optimizing the processing conditions to obtain high gauge factors and on achieving uniformity of the strain sensors.
Reuse
"Modeling the intracranial pressure morphology"
Faculty Advisor: Thomas Heldt
Mentor(s): Thomas Heldt
Contact e-mail: thomas@mit.edu
Research Area(s): BioEECS, Signals and Systems
Intracranial pressure (ICP) is the pressure of the cerebrospinal fluid in the cranial space. ICP rises in response to traumatic brain injury, brain hemorrhage, tumors, hydrocephalus, and many other conditions. While brain-related treatment decisions are mostly based on trends in mean ICP, some decisions are based on the detailed morphology of the ICP waveform. In the latter case, physicians attribute observable changes in the waveform morphology to changes of the physical characteristics of cerebrospinal fluid space. This link, however, is currently poorly understood. In this project, we will i) develop models of the cerebrospinal fluid space to understand the intra-beat dynamics of ICP pulsations; ii) determine which parameter combinations in the model results in waveform changes that are seen clinically; iii) determine whether such models can be used to estimate physical parameters of the cerebrospinal fluid space from clinical waveform data. Our research is conducted in close collaboration with neurosurgeons and neurocritical care physicians at Boston Children's Hospital, Beth Israel Deaconess Medical Center, and Boston Medical Center. Modeling the intracranial pressure morphology Reuse
"Oxygenation patterns in the preterm neonate"
Faculty Advisor: Thomas Heldt
Mentor(s): Thomas Heldt
Contact e-mail: thomas@mit.edu
Research Area(s): BioEECS, Signals and Systems
Oxidative stress during fetal development, delivery, or early postnatal life is a major cause of neuropathology, as both too little oxygen (hypoxia) and too much oxygen (hyperoxia) can significantly injure the developing brain. To help reduce prematurity-related brain injury, our group is collaborating with neonatologists at Beth Israel Deaconess Medical Center and Boston Children's Hospital to determine antecedents of neural injury in the neonate, including hypo- and hyperoxia. To that end, we are collecting bedside monitoring data from newborns in neonatal critical care. Our database currently contains over 5,000 patient records and continues to grow. This superUROP project will focus on investigating oxygenation patterns in preterm infants and relating such patterns to neurological outcome measures. Oxygenation patterns in the preterm neonate Reuse
"Fabrication of Multi-Layer Self-Assembled Sub-100-Nanometer Patterns"
Faculty Advisor: Karl K. Berggren
Mentor(s): Amir Tavakkoli
Contact e-mail: amirtkg@mit.edu
Research Area(s): Materials and Devices, Nanotechnology
Fabrication of multilayer structures (more than 2 layers). The challenge in this project is how to image more than two layers (possibly cross-sectioning) and fabrication of multilayer structures. Reuse
"Computational Glass Blowing"
Faculty Advisor: Erik Demaine
Mentor(s): Martin Demaine
Contact e-mail: demaine@mit.edu
Research Area(s): Computer Systems, Graphics and Human-Computer Interfaces
We are currently developing a C++ software system for experimentation with new designs in glass blowing, to explore the design space before actually making something. Our current focus is to develop a system for glass "cane", and to develop new artistic designs of cane, which has been relatively stagnant field for several years. The project currently involves computer graphics (OpenGL, meshing), geometry (for glass-blowing transformations), Qt user interface design (a kind of visual programming language for glassblowing), and multithreading (for background rendering of complex designs). The project might also grow to include physical simulation and color modeling. You would work with Erik Demaine and Martin Demaine, both of whom blow glass at MIT. See Link for more information on the project. Computational Glass Blowing Reuse
"Font Art through Algorithms"
Faculty Advisor: Erik Demaine
Mentor(s): Martin Demaine
Contact e-mail: demaine@mit.edu
Research Area(s): Graphics and Human-Computer Interfaces, Theoretical Computer Science
We are designing a collection of "puzzle fonts" based on geometric algorithms and open problems. See Link for the fonts we've designed so far. We are looking for a student interested in design, algorithms, and programming to explore the creation and implementation of new mathematical and puzzle fonts. Font Art through Algorithms Reuse
"Unfolding Printable Robots"
Faculty Advisor: Erik Demaine
Mentor(s):
Contact e-mail: edemaine@mit.edu
Research Area(s): Computer Systems, Theoretical Computer Science
The Printable Robots project [Link aims to make cheap (<$100) and immediate (<hour) construction of robots a reality. The concept is that a designed robot can be laser-cut out of sheet material and folded into the desired structure.

This project will explore the underlying geometric algorithms of this process. Given a 3D geometry of a robot, we need to "unfold" it into a 2D geometry suitable for cutting from sheet material. We have some algorithms in mind, based on the computational origami design algorithm "Origamizer", but they need to be implemented and experimented with to evaluate how well they perform. The research will involve algorithm design and implementation.
Unfolding Printable Robots Reuse
"Energy-Efficient Algorithm Compiler"
Faculty Advisor: Erik Demaine
Mentor(s):
Contact e-mail: edemaine@mit.edu
Research Area(s): Computer Systems, Theoretical Computer Science
We are developing a new field of algorithms where we measure the amount of energy it consumes in addition to the time and space it uses. In our model, reversible computation (where the inputs can be recovered from the outputs) is free, but creating or destroying information costs energy. This model will become the dominant energy cost in computers within the next 10-20 years.

We would like to be able to measure the energy savings our algorithms achieve by implementing a language and simple compiler. (The compiler needn't be efficient -- you can think of it as a simulation.) We have already designed most of a pseudocode language, but it would be cool to see it actually "running" (in a simulated sense) on a real computer.

Long term, we envision that this language specification would be a model for the future of energy-efficient computing.
Energy-Efficient Algorithm Compiler Reuse
"CMOS Nanophotonics"
Faculty Advisor: Rajeev Ram
Mentor(s):
Contact e-mail: rajeev@mit.edu
Research Area(s): Applied Physics, Communications, Materials and Devices, Nanotechnology
Apply your understanding of electromagnetism to explore new waveguide based devices embedded within high performance VLSI circuits. This is an opportunity to explore a new generation of devices that are currently being deployed for high-performance computing. The ideal student would have A's in 6.012 and 6.013 (or equivalent). CMOS Nanophotonics Reuse
"Motion Microscopy"
Faculty Advisor: Bill Freeman
Mentor(s): Neal Wadhwa
Contact e-mail: billf@mit.edu
Research Area(s): Applied Physics, Artificial Intelligence, Graphics and Human-Computer Interfaces, Signals and Systems
We have developed a "motion microscope" that enlarges small, invisible
motions to make them visible. See
Link
for examples.

Tasks for the superUROP:

(a) find or take more videos of scenes with small motions and magnify
them, aiming toward mechanical engeineering or medical diagnostic or
civil engineering applications. What would reveal important
information, motion magnified? Can you reveal the Zakim bridge
swaying under the load of traffic? Ability to use, or learn to use,
video equipment needed. Some physics knowledge desirable.

(b) develop automatic method to select the frequency bands of the
motions to be amplified. Now, that selection is done by hand, and
we'd like to develop an automatic way to select the frequencies to
magnify, and the proper amounts.

Needed: experience in signal processing. 6.003 at a minimum, 6.431
desirable. Ability to use video camera, tripod, and willingness to learn how to
use a high-speed video camera.
Reuse
"WikiScout: Automatic Generation of Annotations for Question Answering"
Faculty Advisor: Boris Katz
Mentor(s): Sue Felshin and Alvaro Morales
Contact e-mail: boris@csail.mit.edu
Research Area(s): Artificial Intelligence
The WikiScout project aims to increase the coverage of the START question answering system by automatically generating natural language annotations (generalized facts) from free text in Wikipedia and other online resources. The annotations enrich Wikipedia data with semantic context, thus making it possible for START to answer thousands of new types of questions.

The student will work on developing a probabilistic model for generating and correcting annotations based on semantic relatedness and article connections. This is a good opportunity for someone interested in natural language processing of large data sets.
Reuse
"Optical Health Monitor"
Faculty Advisor: Rajeev Ram
Mentor(s):
Contact e-mail: rajeev@mit.edu
Research Area(s): Applied Physics, BioEECS, Materials and Devices
Our goal is to miniaturize a high-resolution spectroscopy lab capable of obtaining molecular fingerprints of chemicals through the skin. The ultimate embodiment of this work is a non-invasive wrist-band sized monitor for measuring blood sugar in real-time. The ideal candidate should have obtained A's in 6.007 and 6.012. Optical Health Monitor Reuse
"Automatic error correction for non-native English texts"
Faculty Advisor: Boris Katz
Mentor(s): Yevgeni Berzak
Contact e-mail: boris@csail.mit.edu
Research Area(s): Artificial Intelligence
The project deals with Natural Language Processing (NLP) for non-native English texts, with particular focus on automatic correction of grammatical errors. Current grammar checkers typically handle a small number of error types, and leave much room for improvement of coverage and performance. The project will involve analysis of errors committed by English learners, text processing with existing NLP tools, and developing new models for error correction and style checking, integrating Machine Learning and Linguistics. Reuse
"HotSpot vectorization"
Faculty Advisor: Saman Amarasinghe
Mentor(s): Jeff Bosboom
Contact e-mail: jbosboom@csail.mit.edu
Research Area(s): Computer Systems
Looking for a student to implement additional vectorization optimizations in HotSpot, the OpenJDK Java JIT compiler. HotSpot currently vectorizes simple loops with vector destinations, but cannot vectorize anything with a scalar on the left-hand side (e.g., dot product), some forms of index math, or loops with varying tripcount (e.g., 'triangular' nested loops: i in [0, N], j in [i, N]). Besides vectorization passes, you'll probably need to enhance the analysis to prove transformations to be legal. You'll need a strong grasp of compilers, knowledge of C++ (HotSpot's implementation language), and the bravery to work with a large (but well-maintained) code base. We are not personally familiar with HotSpot's internals so you'll have to interact with development mailing lists yourself. If you do a good job, the OpenJDK devs might integrate and ship your changes to millions of users. Reuse
"SunnyUI: An application agnostic UI builder for a model-based Rails framework"
Faculty Advisor: Daniel Jackson
Mentor(s): Aleks Milicevic
Contact e-mail: aleks@csail.mit.edu
Research Area(s): Computer Systems
We have implemented a model-based web framework on top of Ruby on Rails for building event-driven interactive web applications. The core of our framework is an information-rich model which enables the framework to automatically synthesize many parts of the application at runtime, on the fly. For example, there are no controllers, and the programmer does not have to write database migrations, ActiveRecrod association ends (reflections), routes, etc. Furthermore, the framework keeps track of all connected clients and the data they are displaying, so whenever the value of a piece of data changes, the framework immediately propagates that change to all the relevant clients (using the "server push" mechanism). The clients then know how to automatically refresh their views (by updating the DOM), thanks to our smart views (written as regular templates but data binding enabled); as a result, the programmer does not have to write any Ajax code to get a responsive, auto-refreshing GUI.

Currently, the programmer has to manually write the views for his or her application (which are conceptually no different from the standard ERB templates commonly used in Rails applications). This part could be automated as well, which is exactly the main goal of this UROP project: provide a generic user interface, out of the box, directly generated from the core application model, which simply allows the user to view and query the current object state. Alternatively, the student may choose to implement a WYSIWYG UI builder that lets the programmer quickly design GUIs automatically tied to the domain model.

The student joining this project should already be comfortable with Javascript, and ideally familiar with some Javascript UI libraries, as this project is mostly about designing snappy looking widgets for visualizing the application object state. The most challenging part of the project is to come up with a clever, generic, model-based UI, preferably customizable so that it can be tweaked and fine tuned differently for different concrete applications.

This is not a building a one-off GUI for a given app kind of project; rather, this project is about building a set of generic, template-based widgets, that reflectively read the application model (e.g., "what are the all different model classes in this system?", "what are the fields of this particular model class and what are their types?", etc.), dynamically query the application state (e.g., "give me all the instances of this class", "what is the value of this field in this concrete instance", etc.), and display all that data in a meaningful and intuitive way. Fortunately, all these technical and infrastructural problems have already been taken care of, so the student will be able to focus solely on being creative and designing a nice general-purpose GUI.
Reuse
"Optical techniques for minimally invasive blood cytometry"
Faculty Advisor: Martha Gray
Mentor(s): Carlos Castro; Ian Butterworth
Contact e-mail: ccastro@mit.edu; ibutter@mit.edu
Research Area(s): BioEECS, Circuits, Computer Systems
Prototyping and testing optical techniques to assess blood concentrations of neutrophils (i.e. the most common type of white blood cells) in a minimally invasive way.
We’re looking for someone with experience or a keen interest in optics and microscopy to develop prototypes for in vivo blood cytometry. In particular, we will work adapting several microscopy setups, such as oblique back-illumination microscopy or dark-field microscopy (see figures attached), to our particular application. Testing will be performed in vivo on human capillaries. Prototyping may include designing and 3D printing optical components to suit our particular needs, working towards a home-based use and integration with mobile, hand-held devices. Experience with LabView would be valuable but not necessary.
This work is part of a project for neutrophil monitoring in oncology patients that is being funded by the Madrid-MIT M+Vision consortium.
Optical techniques for minimally invasive blood cytometry Reuse
"Making software analysis scale using massively parallel techniques"
Faculty Advisor: Daniel Jackson
Mentor(s): Eunsuk Kang
Contact e-mail: eskang@csail.mit.edu
Research Area(s): Computer Systems
The Alloy Analyzer is a tool for modeling and analyzing software design, used in numerous applications ranging from telephone switches and medical devices to web security and network configuration. The analysis works by translating an input model into a constraint solving problem, which is then handed off to a powerful SAT solver. For complex models, the number of the constraints generated can be huge, and take much longer to solve than the user is willing to wait. In this project, you will explore and develop new techniques to scale the Alloy Analyzer beyond what anyone has ever achieved, employing large server farms to parallelize the solving process. You will apply these techniques to the analysis of realistic systems, such as web services and peer-to-peer protocols. Reuse
"Optical-Based Technologies for Catheter-Infection Prevention"
Faculty Advisor: Martha Gray
Mentor(s): Aurélien Bourquard (primary supervisor), Carlos Castro
Contact e-mail: aurelien@mit.edu, ccastro@mit.edu
Research Area(s): BioEECS, Circuits, Computer Systems
This research project involves the prototyping and testing of innovative optical-based technologies for the in situ sterilization of catheter materials. This work is part of a global research effort that aims at reducing catheter infections in hemodialysis patients and that is funded by the Madrid-MIT M+Vision consortium.

We are looking for someone with experience or interest in applied optics, photomedicine, and/or microbiology. The focus of the project will be on the experimental framework required for growing bacteria on catheter phantoms, and on the testing of distinct sterilization procedures involving mostly UV and NIR light and lasers. Fluorescence microscopy will be used as a tool to measure bacterial death rates and determine the sterilization efficiency of the various methods under investigation. The opportunity in this project is to participate in the research towards new sterilization methods that can have substantial impact in clinical practice and healthcare.
Optical-Based Technologies for Catheter-Infection Prevention Reuse
"aRby: A domain-specific language for Alloy embedded in Ruby"
Faculty Advisor: Daniel Jackson
Mentor(s): Aleks Milicevic
Contact e-mail: aleks@csail.mit.edu
Research Area(s): Computer Systems
Alloy is a first-order, relational, modeling language. Alloy is declarative (based on logic), and as such, it is not executable per se; it is instead used primarily to formally model systems for the purpose of checking various logical properties against them. Being purely declarative, however, makes a number of tasks difficult or impossible.

aRby---an embedding of Alloy in Ruby---demonstrates the benefits of having a declarative modeling language (backed by an automated solver) embedded in a traditional object-oriented imperative programming language. aRby aims to bring the two opposing paradigms (imperative and declarative) closer together in a novel and unique way. We show that having the other paradigm available within the same language is beneficial to both the modeling community of Alloy users and the object-oriented community of Ruby programmers. aRby provides elegant solutions to several well-known, outstanding Alloy problems: (1) mixed execution, (2) specifying partial instances, (3) staged model finding.

We have already implemented a working prototype. The student joining this project would work on exploring some exciting new posibilities this new paradigm (of using imperative code to produce declarative specifications) enables. This project spans across many topics of programming language research (listed above; familiarity is by no means a prerequisite), and being involved in such a broad project would be of great benefit to someone interested in pursuing graduate education in the area of programming languages and program analysis.
Reuse
"Conceptual redesign of software"
Faculty Advisor: Daniel Jackson
Mentor(s): Santiago Perez De Rosso
Contact e-mail: sperezde@csail.mit.edu
Research Area(s): Computer Systems
We're developing a new approach to software design that emphasizes the conceptual models that underlie complex systems. As part of this effort, we're engaged in a series of case studies in which we redesign widely used applications to experiment with conceptual transformations. The goal is to remove rough edges and unnecessary complexity in its design. Past case studies include Dropbox, Gmail and Git. Potential case studies include Google Analytics, Facebook and CSS but you are welcome to propose your own. Reuse
"Optical Spectroscopy Based Hydration Monitoring"
Faculty Advisor: Martha Gray
Mentor(s): Nick Durr & Ryan Coe
Contact e-mail: hydration@mit.edu
Research Area(s): BioEECS, Circuits, Computer Systems
Prototyping and testing of an early stage non-invasive human hydration sensor, based on optical spectroscopy. Working towards a wearable design.
We’re looking for someone with experience or a keen interest in optical spectroscopy, to develop prototypes for human hydration assessment. The prototype will be used on a range of skin areas around the body to monitor relative hydration changes via fluctuations in optical absorption spectra. We’ll be building a measurement and logging system (in LabView, or similar) that the prototype would ideally interface. Experience of working with optical spectroscopy would be valuable but not necessary.
During development, testing will be done on ex-vivo animal and in-vivo human tissue, which will then likely be followed by a more specific dehydration study, where other established and novel approaches will be compared.
This is part of a larger project for hydration monitoring in elderly care that is being funded by the Madrid-MIT M+Vision consortium.
Optical Spectroscopy Based Hydration Monitoring Reuse
"Wearable RF Absorption Based Hydration Monitoring"
Faculty Advisor: Martha Gray
Mentor(s): Ian Butterworth
Contact e-mail: hydration@mit.edu
Research Area(s): BioEECS, Circuits, Computer Systems
Prototyping and testing of a wearable non-invasive human hydration sensor, based on high-accuracy RF absorption.
We’re looking for someone with experience of RF antenna design, to develop prototypes of a low-GHz based RF absorption method for human hydration assessment. The prototype will be used on a range of features around the body to monitor relative hydration changes via fluctuations in RF absorption. We’ll be building a measurement and logging system (in LabView, or similar) that the prototype would ideally interface. Experience of working with network analysers would be valuable but not necessary.
During development, testing will be done on water/saline solutions and in-vivo, which will then likely be followed by a more specific dehydration study, where other established and novel approaches will be compared.

This is part of a larger project for hydration monitoring in elderly care that is being funded by the Madrid-MIT M+Vision consortium.
Wearable RF Absorption Based Hydration Monitoring Reuse
"Speed of Sound Based Hydration Monitoring"
Faculty Advisor: Martha Gray
Mentor(s): Ian Butterworth & Luca Giancardo
Contact e-mail: hydration@mit.edu
Research Area(s): BioEECS, Circuits, Computer Systems
Prototyping and testing of a non-invasive human hydration sensor, based on high-accuracy speed of sound measurement. Towards a portable tool, or wearable sensor.
We’re looking for someone who’s good with prototyping small devices. For instance, enhancing the function of existing equipment, 3D printing, and very basic electronics. The device will be used on a range of features around the body to monitor relative hydration changes, through measurement of ultrasonic transmission, and high accuracy temperature and propagation distance, to give bulk speed of sound. We’ll be building a measurement and logging system (in LabView, or similar) that the prototype would ideally interface.
During development, testing will be done in water/saline solutions and in-vivo, which is then likely be followed by a more specific dehydration study, where other established and novel approaches will be compared.

This is part of a larger project for hydration monitoring in elderly care that is being funded by the Madrid-MIT M+Vision consortium.
Image credit: Sarvazyan A, et al. Ultrasonic assessment of tissue hydration status. Ultrasonics. 2005 Aug;
Speed of Sound Based Hydration Monitoring Reuse
"Neurological changes detected by interaction with mobile devices"
Faculty Advisor: Martha Gray
Mentor(s): Luca Giancardo and Alvaro Sanchez-Ferro
Contact e-mail: neuroqwerty@mit.edu
Research Area(s): BioEECS
In 2013, the average US adult spent 2:21 hours interacting with a mobile device. Every day. Not counting voice calls. In this project, you will contribute to the design and implementation of a bioengineering study that takes advantage of the widespread and routine use of mobile devices. Specifically, through this project we are studying the extent to which changes in brain function are associated with a change in the way someone interacts with a mobile device. If successful, we will be able to employ mobile devices as a “window” into the human brain that will allow for the early detection of neurological diseases.
We are looking for an outstanding, highly motivated student with cross disciplinary interests in computer science and medicine. Your initial role will be the development of low latency, transparent data acquisition software for the Android platform. If successful, you will start to contribute to the development of machine learning and data mining algorithms, which will be employed in clinical studies in Boston and Madrid.
Android development skills are required. Matlab, C++ and experience with scientific data analysis are desirable.
Reuse
"Jade: Design Tools for MOOCs"
Faculty Advisor: Chris Terman
Mentor(s):
Contact e-mail: cjt@mit.edu
Research Area(s): Computer Systems
Hands-learning is both effective and fun! This goal of this project is to develop Jade, a browser-based integrated design environment for digital circuits suitable for use by MOOCs. Instances of Jade can be embedded in online course material to provide interactive figures, hands-on lab assignments and design projects, and so on.

Jade includes diagram capture, simulation at various levels (device, gate, ...), design analysis (timing, rules-check, ...), testing, etc. There are many possible projects depending on your interests:

* editor for IC masks, design rule checker, place-and-route, circuit extraction, ...

* new simulation techniques: design compilers, lazy evaluation, ...

* remote collaboration support, e.g., allow remote mentors to share the design environment with a student to provide "over the shoulder" help, including A/V chat link

* Server-side support (uses Google App Engine/Python)

* Other computer-aided design tool projects of mutual interest

Our goal is to use Jade for on-campus and MITx courses.

The project has room for up to 4 SuperUROPs -- please get in touch if you're interested and want to learn more.
Jade: Design Tools for MOOCs Reuse
"Thermal characterization of semiconductor devices with nanosecond time resolution"
Faculty Advisor: Tomas Palacios
Mentor(s):
Contact e-mail: tpalacios@mit.edu
Research Area(s): Materials and Devices
The maximum performance of many semiconductor devices is typically limited by their power dissipation. For example, although GaN transistors should theoretically allow the fabrication of power amplifiers with an output power density of more than 100 W per millimeter of gate width, heat dissipation limits the performance to little more than 10 W per millimeter. To push the performance of electronic devices to their ultimate thermal limit, it is very important to understand how heat is generated in the device and how the temperature increases as a function of the different operating conditions.

In this project, we will study how the temperature varies in GaN transistors operated under switching conditions. We will focus on how the temperature changes at the nanosecond scale. For this, we will fabricate a new thermoreflectance measurement system that is able to measure the device temperature by analyzing small changes in the surface reflectivity of GaN transistors.
Thermal characterization of semiconductor devices with nanosecond time resolution Reuse
"Smart 3D-Printed Systems"
Faculty Advisor: Tomas Palacios
Mentor(s):
Contact e-mail: tpalacios@mit.edu
Research Area(s): Materials and Devices
The use of three-dimensional printing is quickly changing the way products are designed and manufactured. However, in spite of the tremendous flexibility and potential of current 3D printers, most of the 3D objects that can be printed today lack of any kind of electronics and intelligence.

At the same time, the recent development of graphene and other two-dimensional electronic materials offers tremendous new opportunities to fabricate electronic circuits on almost any surface. In this project, we will embed these new materials into 3D-printed systems to add sensing and information processing capabilities. The "smart" 3D systems that will be demonstrated will be an important step towards ubiquitous electronics.

The main tasks of this project are:

1. Identify new technologies to embed two-dimensional electronic materials (e.g. graphene) into 3D printed systems.
2. Fabrication of "smart" 3D systems at the Microsystems Technology Laboratories.
3. Characterization of the fabricated 3D systems.
Smart 3D-Printed Systems Reuse
"Information Policy"
Faculty Advisor: Hal Abelson
Mentor(s):
Contact e-mail: hal@mit.edu
Research Area(s):
The Internet is now the central nervous system of our global economy and the essential infrastructure for communication, commerce, and civic discourse. Yet at this transformative moment, many important public debates concerning information policy occur without adequate technical understanding and scholarship. The new MIT Information Policy Project seeks to fill this gap with technically-informed research and politically-engaged dialogue, aimed at guiding Internet policymakers around the world.

Based within MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL), the Information Policy Project builds on MIT’s proven approach to engineering research and education. We view policy development, like technology development, as a design discipline that should be driven by methodical study, and we aims to train a new generation of technology policy leaders in government, civil society, academia and industry.

In its first three years, the Project will tackle research challenges such as:

Privacy and surveillance, including accountable systems and their potential; mobile apps privacy; civil liberties in the age of big data; EU-US privacy agreements; and surveillance and national security.

Network architecture, including the evolution of wireless and wireline access; the economics and flow of payments across network layers; and alternate service provision models.

Internet governance, including assessments of existing global mechanisms and the growth of policy expertise within the global internet community.

Other research priorities will include cybersecurity, online copyright protection, Internet content regulation, and intermediary liability worldwide.

There are several SuperUROP projects for students who want to be part of this new effort. It would be good have some familiarity with the material covered in 6.805/STS085 and also 6.033, but these are not necessarily prerequisites.
Information Policy Reuse
"Design of wireless power transfer and data communication for an implantable medical device"
Faculty Advisor: Charles Sodini
Mentor(s):
Contact e-mail: brunogr@mit.edu
Research Area(s): BioEECS
Neurologists dealing with patients with epilepsy currently have two issues. First, some patients must stay in the hospital several weeks in order to capture a seizure on EEG (recording of the electrical activity on the scalp). Second, in order to tailor the medication, the doctor relies on the patient to self-report the number of seizure he or she had even though self-reporting is known to be very inaccurate. To solve these two problems we are designing a long-term EEG recording and seizure detection system that will be implanted between the patient’s scalp and skull. This system will be powered through an implanted battery that will need to be charged daily. In order to charge the battery and get the data from the device, the patient will have to wear an external device for about 10 minutes every day. The goal of this Super UROP will be to design the external device board and test the data/power communication to a mockup of the implanted system. The external board will consist basically of a microcontroller, power amplifier, and some circuits to receive and transmit data. Data and power will be transferred through a set of coils that will also need to be designed. Design of wireless power transfer and data communication for an implantable medical device Reuse
"Bacterial-CMOS integration"
Faculty Advisor: Anantha Chandrakasan, Timothy Lu
Mentor(s): Phillip Nadeau
Contact e-mail: anantha@mtl.mit.edu
Research Area(s): BioEECS, Circuits
This project explores the interface between microelectronic circuits and synthetic biological sensors. We are designing a self-powered microelectronics package to enable readout of luminescent biological sensors in situ, creating an opportunity for localized real-time measurements of biomarkers outside of the laboratory setting.

There are several important opportunities to get involved. One is modeling the parameter space of the system (concentration and volume of cells/analyte, temperature, time constants, signal strength, etc). Another is in the energy and power-electronics design, exploring new sources of energy and associated electronics to power the system in or around the body.

Student should have a strong interest in circuits/electronics and some interest in signal processing. Interest or familiarity with biology concepts helpful but not required.
Bacterial-CMOS integration Reuse
"Parallel processing for next generation Ultra-HD video compression with H.265/HEVC"
Faculty Advisor: Vivienne Sze & Charles Leiserson
Mentor(s):
Contact e-mail: sze@mit.edu
Research Area(s): Computer Systems, Signals and Systems
Today, video content exceeds 50% of the Internet bandwidth and continues to grow; accordingly, video compression is critical for supporting the video demands of tomorrow. The latest video coding standard High Efficiency Video Coding (H.265/HEVC) was recently completed in January 2013, and delivers 50% higher compression than its successor H.264/AVC, today’s most widely used standard. In addition to improved compression, high throughput implementations are required in order to deliver the high resolutions (e.g. 4K and 8K ultra-high definition) and frame rates expected for next generation video. To address the speed requirements, HEVC also has several new built-in parallel features.

We are looking to explore the speed and performance impact of these new forms of parallelism using Cilk-P and to develop encoder implementation running on a multi-core platform.
Reuse
"Fabricate a test chip"
Faculty Advisor: Vicky Diadiuk
Mentor(s): Bernard Alamariu
Contact e-mail: bernard@mtl.mit.edu
Research Area(s): Nanotechnology
This project involves the design, fabrication and test of a Silicon semiconductor test structure, used for the qualification of our
semiconductor fabrication systems. The test chip includes Solar cell,
MOS transistors, diodes, capacitors, resistors and pattern transfer control structures. We will also test the concentration of the mobile ions, and of the deep level metals by electrical methods as well as processing parameters of the clean room machines. The analyzed data
report will specify also the effect of contamination level on the electrical
parameters of active devices.
The student will learn semiconductor fabrication techniques including
the clean room safety and operation procedures.
Reuse
"Smart power converters for rural electrification: Generator Module Development"
Faculty Advisor: David Perreault
Mentor(s): Wardah Inam
Contact e-mail: wardah@mit.edu
Research Area(s): Circuits, Energy
More than 1.3 billion people around the world live without any electricity and even a greater population does not have access to reliable electricity. This energy poverty results in a lower quality of life and is a significant hurdle in the economic development of these areas. This project is focused on developing power electronics that tackle the challenges associated with rural electrification. The goal is to develop low-cost power converters which enables people, who are unable to afford expensive generating sources, an access to electricity by creating an ad-hoc and scalable microgrid. The research will include evaluating and testing of techniques and devices which will result in low-cost and high efficiency power converters. The technology developed will be piloted in India over IAP 2015. This particular SuperUROP project focuses on the "Generator Module" aspect of the system. Reuse
"ASICs for Security Applications"
Faculty Advisor: Anantha Chandrakasan
Mentor(s): Chiraag Juvekar
Contact e-mail: chiraag@mit.edu
Research Area(s): Circuits, Computer Systems
A large number of applications ranging from automobiles to battery chargers require embedded computers that must support cryptographic operations like encryption, signing and message authentication. The embedded nature of these systems makes them especially vulnerable to side channel attacks. A careful analysis of implementation details is critical to prevent any information leakage without compromising performance.
This project spans across circuits, micro-architecture and architecture abstractions with the final goal of building a secure ASIC primitive to support fast cryptographic operations. Based on the specific interests of the candidate the work can involve one or more of the following:
1. Analyzing new crypto (for e.g. SHA3) for implementation specific vulnerabilities.
2. Proposing hardened versions to be implemented in real silicon.
3. Using innovative circuit techniques to implement new TRNG designs.
4. Implementing fundamentally new techniques to allow computation over encrypted data
ASICs for Security Applications Reuse
"A wearable, long term ECG measurement system"
Faculty Advisor: Charles Sodini
Mentor(s):
Contact e-mail: maggied@mit.edu
Research Area(s): BioEECS
Cardiovascular disease is the leading cause of death in the United States. Researchers here at MIT are developing risk stratification methods that could determine if and when an individual will have a cardiac event. A key aspect of this research is algorithm development on electrocardiogram (ECG) traces from at risk individuals. Researchers in the Sodini Group have developed a wearable, long term ECG measurement system that can record a single lead ECG and 3-axis acceleration data for a period of up to one week. While the prototype device performs well under a variety of conditions, both mechanical and software improvements are needed before the device can be worn by patients. The goal of this Super UROP will be to design the next generation of the prototype system. Special attention to mechanical stability and comfort will need to be taken into account, along with ensuring high data quality during periods of activity. Additionally, the software for the system should allow for easy use by clinical staff. A wearable, long term ECG measurement system Reuse
"Algorithms for E. coli Genome Engineering"
Faculty Advisor: Timothy Lu
Mentor(s):
Contact e-mail: timlu@mit.edu
Research Area(s): Artificial Intelligence
The Lu lab is interested in developing computational algorithms to guide efficient targeted genome engineering in E. coli and in validating these algorithms with high-throughput experimental techniques. We seek a SuperUROP to work on expanding our modeling and simulation framework, designing experimental validation strategies, and implementing these in living cells. SuperUROPs will work closely with graduate students, postdocs, and Prof. Lu to achieve these goals. Reuse
"Soft Robots"
Faculty Advisor: Daniela Rus
Mentor(s):
Contact e-mail: rus@csail.mit.edu
Research Area(s): Artificial Intelligence, Materials and Devices, Numerical Methods, Theoretical Computer Science
In this project we address the design, fabrication, and control of a soft robots such as robot fish and robot manipulators. Soft robot generally consist of several segments actuated using bidirectional fluidic elastomer actuators. The robots are fabricated using molding and 3d printing processes. We also develop the associated computation and control systems enable the robots to move. Reuse
"Printable Robots"
Faculty Advisor: Daniela Rus
Mentor(s):
Contact e-mail: rus@csail.mit.edu
Research Area(s): Artificial Intelligence, Graphics and Human-Computer Interfaces, Materials and Devices
Designing and fabricating new robotic systems is typically limited to experts, requiring engineering background, expensive tools, and considerable time. In contrast, to facilitate everyday users in developing custom robots for personal use, this project aims to create algorithms, systems, and tools to easily create printable robots from high-level structural specifications. From that, the system generates complete mechanical drawings suitable for fabrication, instructions for the assembly of electronics, and software to control and drive the final robot.This project aims to develop steps towards creating a hardware compiler. Printable Robots Reuse
"iDiary"
Faculty Advisor: Daniela Rus
Mentor(s):
Contact e-mail: rus@csail.mit.edu
Research Area(s): Artificial Intelligence, Computer Systems, Theoretical Computer Science
Information extraction from mobile phones, smart glasses, or robots sensors such as GPS and video over long periods of time enable the autonomous analysis of the activity stream for the mobile agent that generated these capabilities. The collected data is valuable for mapping, situation awareness, and modeling behaviors, but requires efficient tools that can extract the right information at the right time efficiently. In this project we develop algorithm and system for autonomously extracting activities from GPS and video streams. iDiary Reuse
"Mobile apps for education"
Faculty Advisor: Hal Abelson
Mentor(s):
Contact e-mail: hal@mit.edu
Research Area(s): Computer Systems
MIT App Inventor is a Web-based development environment that let's anyone build mobile apps for Android phones and tablets -- even kids who have never programmed before. Our group in CSAIL and he Media Lab run App Inventor as a worldwide service, currently with 3 million users. The system is used by hobbyists, teachers, and school kids, and there are several curriculum project that use App Inventor and mobile app development as a basis for teaching computing.

There are several projects available, including getting involved in back end maintenance and development of a major cloud computing service; building new capabilities into the platform to support new curriculum and project ideas; working with kids and teachers around the work in creating mobile apps; and performing data mining and applying machine learning methods on log information to gain insight into how students learn about programming and app creation.
Mobile apps for education Reuse
"Compute Trend Prediction"
Faculty Advisor: Devavrat Shah
Mentor(s): George Chen
Contact e-mail: devavrat@mit.edu
Research Area(s): Artificial Intelligence, Computer Systems
We have developed a statistical method to predict trends - it is established to be quite effective in predicting trends - for example, in Twitter, it predicts trends more than 1 hour in advance. Mathematically, we have established various desirable statistically properties.

The goal of the project is to develop a scalable computation architecture and implementation for this method.

Proficiency with Python and interest in Scientific computing are desirable.
Reuse
"Open Data for MIT Students"
Faculty Advisor: David Karger
Mentor(s):
Contact e-mail: karger@mit.edu
Research Area(s): Graphics and Human-Computer Interfaces
This project aims at improving the ability of MIT students to access and share data that can help them at MIT. On the academic stide, starting from a project like the MIT course picker (picker.mit.edu), we envision creating broader tools that enable students to share information about their overall academic goals and progress (four-year plans), share experience about courses to take (or not), accurately assess workloads, and coordinate curricula with friends. On the recreational side, we'll consider tools that make it easier to locate and use resource at MIT---such as improvements in event planning and room scheduling. Some of this work will involve coordinating with the MIT registrar and data warehouse to open up access to data that can be of value to students; another part will be to build applications that can collect new kinds of data from students and make good use of it. Reuse
"Assistive Devices for Healthcare"
Faculty Advisor: Dorothy Curtis
Mentor(s): Prof Pino & Prof Aqueveque, University of Concepcion
Contact e-mail: dcurtis@csail.mit.edu
Research Area(s): Computer Systems, Graphics and Human-Computer Interfaces, Signals and Systems
This project involves the design and implementation of a non-invasive monitoring and alerting system for people with advanced Multiple Sclerosis (MS). Several sensors are deployed on an electric wheelchair acquiring vital signs, activity level and ambient information. An initial sensing platform was been built by MISTI Chile collaborators.

At this time, we are looking for assistance from students in the analysis of the data that has been collected from the monitoring platform. This algorithm development will be based on 260 hours of data collected from MS volunteers at The Boston Home (TBH). The data includes Ballistocardiogram (BCG), pressure signals, accelerometers and ambient conditions (temperature and humidity). Based on these signals, we are planning to compute vital signs, and mobility (activity) level.

Support in building and deploying a central monitoring station for caregivers may also be needed.
Assistive Devices for Healthcare Reuse
"Consumer Credit Risk Models Via Machine Learning Algorithms"
Faculty Advisor: Andrew W. Lo
Mentor(s):
Contact e-mail: alo@mit.edu
Research Area(s): Artificial Intelligence, Control, Numerical Methods, Signals and Systems
Current credit bureau analytics, such as credit scores, are based on slowly varying consumer characteristics, and therefore are not as relevant for tactical risk management decisions by chief risk officers and policy makers.

We will apply machine-learning techniques to construct nonlinear nonparametric forecasting models of consumer credit risk. “Machine-learning” in the computer science literature refers to a set of algorithms specifically designed to tackle computationally intensive pattern-recognition problems in extremely large datasets. By combining customer transactions and credit bureau data, we expect to be able to construct out-of-sample forecasts that significantly improve the classification rates of credit-card-holder delinquencies and defaults. We will then apply the models and methods to two specific challenges in consumer credit-risk management: deciding when and how much to cut individual-account credit lines, and forecasting aggregate consumer credit delinquencies for the purpose of enterprise-wide and macroprudential risk management.
Consumer Credit Risk Models Via Machine Learning Algorithms Reuse
"Systemic Risk in the Financial System"
Faculty Advisor: Andrew W. Lo
Mentor(s):
Contact e-mail: jcummin@mit.edu
Research Area(s): Control, Numerical Methods, Signals and Systems, Theoretical Computer Science
Following up on the survey of 31 financial indicators by Bisias, Flood, Lo, and Valavanis (2012), we will compare the performance of currently available financial indicators to identify which one(s) could have been more informative for policymakers to navigate through previous crises. The ultimate goal is to construct new indicator(s) that aggregates the information from the plethora of existing measures, purge redundant information, and improve performance. In doing so, we expect to be able to help better understand the risks in the U.S. financial system as well as those in other regions of the world; understand the transmission mechanism of systemic events; and to help the U.S. gauge the likelihood of whether financial distress in other countries spread into the U.S. and whether turbulence in one asset class will spill over and become a systemic event. Systemic Risk in the Financial System Reuse
"Financing Biomedical Innovation"
Faculty Advisor: Andrew W. Lo
Mentor(s):
Contact e-mail: jcummin@mit.edu
Research Area(s): BioEECS, Control, Numerical Methods, Signals and Systems
Biomedical innovation has become riskier, more expensive and more difficult to finance with traditional sources such as private and public equity. In this project, we will consider new financial structures in which a large number of biomedical programs are funded by a single financial entity to substantially reduce the risk of the portfolio. This portfolio entity can finance its activities by issuing debt, a critical advantage because there is a much larger pool of capital that can be invested in debt versus equity. By employing financial engineering techniques such as securitization, it can raise even greater amounts of more-patient capital. Using historical data for new molecular entities in oncology, we will perform simulations of the risk and reward of various financing vehicles to assess the practical relevance of securitization for funding cancer research and drug development. Financing Biomedical Innovation Reuse
"artificial sphincter"
Faculty Advisor: Vicky Diadiuk
Mentor(s):
Contact e-mail: diadiuk@mit.edu
Research Area(s): BioEECS
Electronics and micromechanical systems on flexible substrates might make it possible to fabricate an artificial sphincter. This would greatly enhance the quality of life on incontinent people (elderly, some post-pregnancy women) by preventing leakage and allowing for controlled release of accumulated fluid. In this project we will design and attempt to fabricate a prototype using the MTL fabrication cleanroom facilities. Reuse
"3D Print Preview"
Faculty Advisor: Wojciech Matusik
Mentor(s):
Contact e-mail: wojciech@mit.edu
Research Area(s): Graphics and Human-Computer Interfaces, Numerical Methods
Our group is building a novel, multi-material 3D printer with a modular hardware architecture and an extensive software stack that will allow both expert and novice users to fully exploit its printing potential. We seek a student to help implement a "print preview" feature that would allow for a rapid simulation of the print results before launching a 3D print. Print preview can dramatically improve the productivity of 3D printing since a typical 3D print takes hours or even days and uses high quantities of expensive materials. The perfect candidate would have taken 6.837 (computer graphics) and have very strong math and physics background. 3D Print Preview Reuse
"Blood cell separator for diagnostics"
Faculty Advisor: Jongyoon Han
Mentor(s):
Contact e-mail: jyhan@mit.edu
Research Area(s): BioEECS
Blood contains many cells (red blood cells, white blood cells, platelets) with various functions, as well as foreign pathogens (bacteria, virus, and other non-natural cellular components such as circulating tumor cells), therefore has a lot of information regarding a person’s physiological conditions and potentially important early markers for disease development. However, blood is such a complicated mixture of different cells (approximately 45% of blood is composed of various cells, mostly red blood cells), and it is essential to have an efficient method to separate blood cells into different subcomponents, so that one can analyze them properly. For example, detection of a very small amount of bacterial cells from a person’s blood can diagnose deadly diseases such as sepsis (blood infection). In this project, we will develop a simple microfluidic device (based on inertial microfluidic principles) to take a raw blood and separate them into different subgroups (red blood cells, white blood cells, platelets, bacteria/virus) in a rapid and continuous flow manner. This will be useful for many disease diagnostics application. Blood cell separator for diagnostics Reuse
"Microfluidic Activity Assay for Monitoring the Quality of Drug Molecules"
Faculty Advisor: Jongyoon Han
Mentor(s):
Contact e-mail: jyhan@mit.edu
Research Area(s): BioEECS
With the issue of fake drugs in some countries, and the fact that more and more drugs are not simple small molecules but biologics (protein drugs) with limited lifetime, it is now more important to be able to rapidly and quickly test validate the efficacy of a drug without going through an extensive laboratory procedure, potentially at the site of the care. In this project, we will develop a simple microfluidic device that can test the effectiveness of a protein drug molecule, by utilizing Han group’s biomolecule concentration technology. Drug molecules will be bound to its substrate (either binding or reaction partner) outside of the device, and then analyzed in the microfluidic device. The device will separate the bound (reacted) and unbound (unreacted) species, therefore quantifying the efficacy of the drug rapidly. Students will take a model case drug and design the probe, design and fabricate the device, and test / quantify the drug efficacy using a simple microfluidic device to demonstrate that it is possible to test the drug efficacy rapidly and on the field, without the need for extensive labs and technicians. Microfluidic Activity Assay for Monitoring the Quality of Drug Molecules Reuse
"Neural Network Training Algorithms"
Faculty Advisor: Cardinal Warde
Mentor(s):
Contact e-mail: warde@mit.edu
Research Area(s): Control
The Photonic Systems Group is developing a Compact Opto-electronic Integrated Neural (COIN) network co-processor. The COIN employs arrays of photodetectors and thresholding circuits in silicon, in combination with arrays of light emitters and holographic optical interconnection elements sandwiched together into a brick. Our current training algorithm is based on a gradient-descent algorithm using nearest-neighbor interconnections. The goal of the project is to implement this or other suitable algorithm on the native hardware (discrete component based at this time) and demonstrate its effectiveness in an application (e.g., face recognition) with the system. Neural Network Training Algorithms Reuse
"2-D Bistable Optical Array"
Faculty Advisor: Cardinal Warde
Mentor(s):
Contact e-mail: warde@mit.edu
Research Area(s): Materials and Devices
The Photonic Systems Group is developing a Compact Opto-electronic Integrated Neural (COIN) network co-processor. The COIN employs arrays of photodetectors and thresholding circuits in silicon, in combination with arrays of light emitters and holographic optical interconnection elements sandwiched together into a brick. One of the key hardware subsystems is a dense 2-D array of bistable optical devices. We are exploring the development of such a subsystem, first at the discrete component level, by driving an array of LEDs with an array of photodetectors and thresholding electronics. This project involves: (1) modeling of the components, (2) design and modeling of the bistable optical subsystem, and (3) assembly and testing of a hybrid bistable optical array. Follow-on work would focus on the design and fabrication of the final compact VLSI integrated-circuit version of the COIN (2 students). 2-D Bistable Optical Array Reuse
"Creosote Thickness Monitor"
Faculty Advisor: Vicky Diadiuk
Mentor(s):
Contact e-mail: diadiuk@mit.edu
Research Area(s): Energy
With the increasing popularity of wood-burning stoves, measuring creosote build-up, to know when the chimney should be cleaned, & thus avoid fires, is very important. In this project we will design a sensor system that can be installed in the upper chimney & will send a signal when the creosote thickness reaches 0.125". The challenges are a harsh environment & low cost. Reuse
"QoSBench: A benchmark suite to quantify quality of service"
Faculty Advisor: Daniel Sanchez
Mentor(s):
Contact e-mail: sanchez@csail.mit.edu
Research Area(s): Computer Systems
Computation is often interactive and has real-time requirements (e.g., search, web services, real-time analytics, gaming, multimedia). However, computer system performance is still evaluated with non-interactive benchmark suites and using long-term average metrics (e.g., instructions per cycle, queries per second) that ignore short-term fluctuations. This was often good enough for uniprocessors, but the move to multicore systems and increasingly parallel programs has made it much easier to degrade quality of service (QoS), since concurrent threads share several hardware and software resources, often competitively and unfairly, and have frequent dependencies among them. In this project, you will take several representative, widely-used parallel applications (such as a game or a search engine), modify them to capture an application-specific quantitative QoS metric (e.g., a histogram of response latencies or frame rates), package them into an easy-to-use benchmark suite, and produce aggregate QoS metrics that enable us to score systems not on their average performance, but their performance distribution. This will enable system designers to easily quantify QoS, and pave the way for the design of future architectures, runtimes and operating systems with robust QoS. Reuse
"Distributed Access Control for Social Media"
Faculty Advisor: David Karger
Mentor(s):
Contact e-mail: karger@mit.edu
Research Area(s): Computer Systems
The goal of this project is to construct a fully distributed access control protocol, and implement a system, that supports the kind of friend-oriented sharing offered by Facebook and other social media tools. The protocol should allow users to define a set of "friends" and indicate that certain resources should only be accessible to those friends. It must also provide a mechanism that allows individuals to authenticate that they are members of a given "friends" group to gain access permissions. The authentication mechanism must be distributed (so anyone who wants to can implement and "authenticator" system. All of this must be done while simultaneously preserving privacy (of who is in what friends lists). Reuse
"Algorithms for extraction and validation of arousal markers from photo plethysmograph (PPG) recordings during disordered sleep"
Faculty Advisor: George Verghese
Mentor(s): Thomas Heldt, Robert Daly
Contact e-mail: verghese@mit.edu
Research Area(s): BioEECS, Signals and Systems
Sleep apnea, or intermittent cessation of breathing during sleep, deprives the afflicted individual of proper rest. It is also associated with significant cardiovascular and neurological derangements that can have serious health consequences. To diagnose sleep apnea, patients undergo laboratory testing in which a large number of physiological variables are measured continuously, while the patient is sleeping.

The premise of this project is that patients with sleep apnea resulting from instability of the various feedback mechanisms that regulate breathing during sleep may experience physiological arousal events considerably more frequently than they manifest the apneic and hypopneic events that get scored in a sleep study. These frequent arousals can themselves potentially be the source of health issues arising from disordered sleep.

The project aims to develop methods and tools to recognize, validate and quantify the arousals that a patient experiences during sleep. The challenge
ultimately is to do this recognition with data from a simple and nonintrusive photo plethysmograph on the finger. The dataset for this study was collected
during sleep studies on children diagnosed with ADHD.
Algorithms for extraction and validation of arousal markers from photo  plethysmograph (PPG) recordings during disordered sleep Reuse
"Using images to reverse engineer man-made objects"
Faculty Advisor: Wojciech Matusik
Mentor(s):
Contact e-mail: wojciech@csail.mit.edu
Research Area(s): Graphics and Human-Computer Interfaces
The task for this research project can be stated as follows: “given a photograph of a man-made object (e.g., a bicycle, a furniture piece), automatically produce a complete 3D design (e.g., a CAD model) of the corresponding object that can be assembled and manufactured”. To solve this seemingly impossible task, we will rely on a database of component templates created by a group of mechanical engineering students. We use a synthesis approach that builds new designs from pieces of the designs that are in the example data set. The main goal for the UROP project is to develop an algorithm for finding the best match to the input photographs. Using images to reverse engineer man-made objects Reuse
"From Mockup to Web App: Building the Next-Generation Web Template Language"
Faculty Advisor: David Karger
Mentor(s): Ted Benson
Contact e-mail: karger@mit.edu
Research Area(s): Graphics and Human-Computer Interfaces
Think web frameworks like Node and Backbone are cool? Then help us develop the future of web templates. We are working on a web template language that continues to have benefits long after the page is rendered, including: rich copy-and-paste of data between websites, in-browser WYSIWYG editing, automatically-generated APIs, and site themes that are trivially transportable from site to site. We aim to empower casual web users with the ability to make professional web sites: from just a mockup, we hope to infer the data-backend and editing interface; by just pointing at another site, we hope to import that site's style for reuse on one's own. Experience with Javascript (or Coffeescript) and web development is a plus, as is good performance in 6.813/831. From Mockup to Web App: Building the Next-Generation Web Template Language Reuse
"Transparent Web Browsing"
Faculty Advisor: David Karger
Mentor(s):
Contact e-mail: karger@mit.edu
Research Area(s): Graphics and Human-Computer Interfaces
Nowadays, all sorts of shady companies are collecting information about your browsing activities and using it for their own mysterious purposes. How could that information be used to your benefit? We propose to build Eyebrowse, a web browser extension that gathers information about your web browsing activities and shares that information (under your control) with others to mutual benefit. Potential applications include discovering interesting new web sites based on the browsing activity of your friends, improving web navigation by blazing trails to the important parts of web sites, supporting chance encounters when you and your friends are visiting the same web site, collaboratively browsing the web, identifying links between pages that ought to exist but don't, reporting on global web activity trends, tagging sites and pages according to the interests of people who visit them, and other exciting applications that you will come up with. Experience with Javascript, Django and general web application design is a plus, as is good performance in 6.813/831. Reuse
"The Future Textbook"
Faculty Advisor: David Karger
Mentor(s):
Contact e-mail: karger@mit.edu
Research Area(s): Graphics and Human-Computer Interfaces
Now that we can put textbooks on the web, how can we change them to make them better? How can we make them more dynamic, more adaptable to individual students, more sociable, or more informative? We've tackled some of these questions with Nb (Link a tool that lets students hold forum-type discussions in the margins of their online reading material. Nb is currently in use in roughly 25 classes at 6 universities. We have a long list of improvements to implement and assess in Nb, including social moderation, key-question highlighting, organization via tagging, chat and wiki functionality, support for sketching diagrams and other non-text annotations, hot-spot mapping for faculty, html and video annotation, and many others that students like you think of. Experience with Javascript and Django is a plus, as is good performance in 6.813/831. The Future Textbook Reuse
"Information Scraps, Quick Notetaking, and Personal Information Organization"
Faculty Advisor: David Karger
Mentor(s):
Contact e-mail: karger@mit.edu
Research Area(s): Graphics and Human-Computer Interfaces
Our lives are filled with small, random scraps of information that seem to have no natural home. Where do we put them, and how do we find them later? We've created List.it (Link a fast, lightweight browser extension for capturing and organizing such scraps. Listit has over 25,000 active users who have recorded more than 100,000 scraps. Analyzing them we've discovered important subpopulations such as packrats, minimalists, and spring cleaners. To advance our study of personal information organization, we want to study the activity of our current users (both the content they create and their interaction with it), add useful functionality to the tool (such as sharing, reminding, and context sensitive retrieval) and study the way users react to the new functionality. Information Scraps, Quick Notetaking, and Personal Information Organization Reuse
"Interactive Data Visualization for Journalists using Wordpress"
Faculty Advisor: David Karger
Mentor(s):
Contact e-mail: karger@mit.edu
Research Area(s): Graphics and Human-Computer Interfaces
There's a new movement in journalism to incorporate rich data visualization in news stories, but many journalists lack that skills to create their own "news apps" for this purpose. We've prototyped a data visualization framework, Datapress (Link to support authoring (not programming) such visualizations in Wordpress, a popular platform for journalism. Your job is to flesh out this prototype. This will involve a cyclic process of contacting and working with journalists to support their use of Datapress, discovering what improvements need to be made to enhance the value of the tool, and implementing those improvements to the Datapress platform. Interactive Data Visualization for Journalists using Wordpress Reuse
"Interactive Data Visualization for Everyone the Web"
Faculty Advisor: David Karger
Mentor(s):
Contact e-mail: karger@mit.edu
Research Area(s): Graphics and Human-Computer Interfaces
Exhibit (Link is an open source Javascript library that helps non-programmers author and publish rich interactive data visualizations on the web. We use Exhibit to push the boundaries of web authoring without programming, with our ultimate goal being to enable end-users to WYSIWYG-author complete web applications. Exhibit has been adopted on over a thousand web sites by hobbyists, scientists, merchants, and journalists and has served several million page views. Opportunities to advance Exhibit include (i) incorporating new types of visualizations such as heat maps or network layouts, or entire visualization frameworks such as theJit or dojo GFX into Exhibit, (ii) incorporating Exhibit into common web platforms such as Mediawiki (see Link or Wordpress (see Link (iii) enhancing performance using powerful Javascript libraries such as datavore, and (iv) studying Exhibit's thousands of uses on the web to learn more about how people manage information and what could make Exhibit more useful. Interactive Data Visualization for Everyone the Web Reuse

Total: 139

Technical questions related to this site? Please contact Search-Master@superurop.eecs.mit.edu