This is default featured slide 5 title

This is default featured slide 5 title

You can completely customize the featured slides from the theme theme options page. You can also easily hide the slider from certain part of your site like: categories, tags, archives etc. More »

 

Studies natural language processing and machine

Regina Barzilay, a professor in MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) who does research in natural language processing and machine learning, is a recipient of a 2017 MacArthur Fellowship, sometimes referred to as a “genius grant.”

The fellowships carry a five-year, $625,000 prize, which recipients are free to use as they see fit. Twenty-one current MIT faculty members and three staff members have won MacArthur Fellowships, which were established in 1981 and are usually given out to roughly 25 people each year.

In accepting the award, Barzilay credited MIT for being an energizing and encouraging community.

“I have been blessed to work with amazing students and colleagues who challenge my thinking, inspire me, and give me a new perspective on research,” Barzilay says. “From my first days at MIT, it was clear to me that you don’t have to conform to existing standards in the field. You are free to explore any direction you like.”

The Delta Electronics Professor of Electrical Engineering and Computer Science, Barzilay does research in natural language processing (NLP) and machine learning. Her research covers multiple areas of NLP, from syntactic parsing and the deciphering of dead languages, to developing new ways to train neural networks that can provide rationales for their decisions.

“I’m rarely interested in providing yet another solution to traditional NLP tasks,” she says. “I’m most excited about solving problems not within the mainstream of the field that require new perspectives.”

She has also been active in applying machine learning methods to oncology and drug design, arguing that data-driven approaches will soon revolutionize early detection and treatment of cancer.

The MacArthur Foundation cited Barzilay for making “significant contributions to a wide range of problems in computational linguistics, including both interpretation and generation of human language.”

automatically evaluates proposals from far-flung data scientists

In the analysis of big data sets, the first step is usually the identification of “features” — data points with particular predictive power or analytic utility. Choosing features usually requires some human intuition. For instance, a sales database might contain revenues and date ranges, but it might take a human to recognize that average revenues — revenues divided by the sizes of the ranges — is the really useful metric.

MIT researchers have developed a new collaboration tool, dubbed FeatureHub, intended to make feature identification more efficient and effective. With FeatureHub, data scientists and experts on particular topics could log on to a central site and spend an hour or two reviewing a problem and proposing features. Software then tests myriad combinations of features against target data, to determine which are most useful for a given predictive task.

In tests, the researchers recruited 32 analysts with data science experience, who spent five hours each with the system, familiarizing themselves with it and using it to propose candidate features for each of two data-science problems.

The predictive models produced by the system were tested against those submitted to a data-science competition called Kaggle. The Kaggle entries had been scored on a 100-point scale, and the FeatureHub models were within three and five points of the winning entries for the two problems.

But where the top-scoring entries were the result of weeks or even months of work, the FeatureHub entries were produced in a matter of days. And while 32 collaborators on a single data science project is a lot by today’s standards, Micah Smith, an MIT graduate student in electrical engineering and computer science who helped lead the project, has much larger ambitions.

FeatureHub — like its name — was inspired by GitHub, an online repository of open-source programming projects, some of which have drawn thousands of contributors. Smith hopes that FeatureHub might someday attain a similar scale.

“I do hope that we can facilitate having thousands of people working on a single solution for predicting where traffic accidents are most likely to strike in New York City or predicting which patients in a hospital are most likely to require some medical intervention,” he says. “I think that the concept of massive and open data science can be really leveraged for areas where there’s a strong social impact but not necessarily a single profit-making or government organization that is coordinating responses.”

Smith and his colleagues presented a paper describing FeatureHub at the IEEE International Conference on Data Science and Advanced Analytics. His coauthors on the paper are his thesis advisor, Kalyan Veeramachaneni, a principal research scientist at MIT’s Laboratory for Information and Decision Systems, and Roy Wedge, who began working with Veeramachaneni’s group as an MIT undergraduate and is now a software engineer at Feature Labs, a data science company based on the group’s work.

The years most significant innovations address

R&D 100 Awards have been presented to six technologies that were developed either solely by technical staff from MIT Lincoln Laboratory or through their collaborations with researchers from other organizations. These awards, given annually by R&D Magazine, recognize the 100 most significant inventions introduced in the past year.

A panel composed of R&D Magazine editors and independent reviewers selects the recipients from hundreds of nominees from industry, government laboratories, and university research institutes worldwide. The awards were announced during a banquet at the 2017 R&D 100 Conference last month in Orlando, Florida. The six winning technologies from this year bring to 38 the total number of R&D 100 Awards that Lincoln Laboratory has received since 2010.

Measuring metabolism to guide health decisions

Lincoln Laboratory, in collaboration with the U.S. Army Research Institute of Environmental Medicine and the Marine Expeditionary Rifle Squad, developed a low-cost personal sensor that allows individuals to make on-demand metabolic measurements simply by breathing into the apparatus. The Carbon dioxide/Oxygen Breath and Respiration Analyzer (COBRA) measures the respiratory exchange ratio in exhaled breath (i.e., the ratio of carbon dioxide produced to oxygen consumed) and the volume rate of oxygen consumption. These measurements enable the calculation of both energy expenditure and the levels of carbohydrates and fats burned by the body to meet energy-expenditure demands.

This information about the rate of energy expenditure and the dietary sources of metabolic energy can help military commanders, doctors, physical trainers, and coaches set reasonable standards for physical exertion. For example, limits on the distance and speed of foot marches can be established by quantifying metabolic workloads of soldiers. Clinicians could also use COBRA metabolic data to tailor dietary and exercise regimens for the one-third of Americans who are struggling to manage obesity and avoid the high blood glucose levels that precede the onset of Type 2 diabetes.

“The COBRA system promises to cost-effectively provide RER and energy-expenditure measurements comparable to those provided by clinical sensors costing as much as $40,000, and with ease of use that makes personal ownership feasible,” says Gary Shaw, principal investigator on the Laboratory’s COBRA team.

Enabling remotely piloted aircraft to avoid mid-air collisions

Lincoln Laboratory worked with the U.S. Army, SRC Inc., and Kutta Technologies to develop a system that enables remotely piloted aircraft to meet Federal Aviation Administration (FAA) and international standards for seeing and avoiding other aircraft.

The Ground-Based Sense-and-Avoid (GBSAA) System for Unmanned Aircraft Systems (UAS) uses both existing FAA radars to locate transponder-equipped airplanes that communicate with air traffic control personnel and 3-D radars with special processing algorithms to locate general-aviation airplanes that do not carry transponders. The GBSAA system’s purpose-built algorithms process the surveillance data from the various radars to track aircraft and to estimate the risk the remotely piloted systems pose to nearby aircraft; the system then issues warnings and provides maneuver guidance for UAS pilots.

The next generation of artificial intelligence

On a recent Monday morning, Vivienne Sze, an associate professor of electrical engineering and computer science at MIT, spoke with enthusiasm about network architecture design. Her students nodded slowly, as if on the verge of comprehension. When the material clicked, the nods grew in speed and confidence. “Everything crystal clear?” she asked with a brief pause and a return nod before diving back in.

This new course, 6.S082/6.888 (Hardware Architecture for Deep Learning), is modest in size — capped at 25 for now — compared to the bursting lecture halls characteristic of other MIT classes focused on machine learning and artificial intelligence. But this course is a little different. With a long list of prerequisites and a heavy base of assumed knowledge, students are jumping into deep water quickly. They blaze through algorithmic design in a few weeks, cover the terrain of computer hardware design in a similar period, then get down to the real work: how to think about making these two fields work together.

The goal of the class is to teach students the interplay between two traditionally separate disciplines, Sze says. “How can you write algorithms that map well onto hardware so they can run faster? And how can you design hardware to better support the algorithm?” she asks rhetorically. “It’s one thing to design algorithms, but to deploy them in the real world you have to consider speed and energy consumption.”

“We are beginning to see tremendous student interest in the hardware side of deep learning,” says Joel Emer, who co-teaches the course with Sze. A professor of the practice in MIT’s Department of Electrical Engineering and Computer Science, and a senior distinguished research scientist at the chip manufacturer NVidia, Emer has partnered with Sze before. Together they wrote a journal article that provides a comprehensive tutorial and survey coverage of recent advances toward enabling efficient processing of deep neural networks. It is used as the main reference for the course.

In 2016, their group unveiled a new, energy-efficient computer chip optimized for neural networks, which could enable powerful artificial-intelligence systems to run locally on mobile devices. The groundbreaking chip, called “Eyeriss,” could also help usher in the internet of things.

Association for Computing Machinery Fellows

Today four MIT faculty were named among the Association for Computing Machinery’s 2017 Fellows for making “landmark contributions to computing.”

Honorees included School of Science Dean Michael Sipser and three researchers affiliated with MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL): Shafi Goldwasser, Tomás Lozano-Pérez, and Silvio Micali.

The professors were among fewer than 1 percent of Association for Computing Machinery (ACM) members to receive the distinction. Fellows are named for contributions spanning such disciplines as graphics, vision, software design, algorithms, and theory.

“Shafi, Tomás, Silvio, and Michael are very esteemed colleagues and friends, and I’m so happy to see that their contributions have recognized with ACM’s most prestigious member grade,” said CSAIL Director Daniela Rus, who herself was named an ACM Fellow in 2014. “All of us at MIT are very proud of them for receiving this distinguished honor.”

Goldwasser was selected for “transformative work that laid the complexity-theoretic foundations for the science of cryptography.” This work has helped spur entire subfields of computer science, including zero-knowledge proofs, cryptographic theory, and probabilistically checkable proofs. In 2012 she received ACM’s Turing Award, often referred to as “the Nobel Prize of computing.”

Lozano-Pérez was recognized for “contributions to robotics, and motion planning, geometric algorithms, and their applications.” His current work focuses on integrating task, motion, and decision planning for robotic manipulation. He was a recipient of the 2011 IEEE Robotics Pioneer Award, and is also a 2014 MacVicar Fellow and a fellow of the Association for the Advancement of Artificial Intelligence (AAAI) and of the IEEE.

Like Goldwasser, Micali was also honored for his work in cryptography and complexity theory, including his pioneering of new methods for the efficient verification of mathematical proofs. His work has had a major impact on how computer scientists understand concepts like randomness and privacy. Current interests include zero-knowledge proofs, secure protocols, and pseudorandom generation. He has also received the Turing Award, the Goedel prize in theoretical computer science, and the RSA prize in cryptography.

Answer plain language business queries in minutes instead of months

Companies often employ number-crunching data scientists to gather insights such as which customers want certain services or where to open new stores and stock products. Analyzing the data to answer one or two of those queries, however, can take weeks or even months.

Now MIT spinout Endor has developed a predictive-analytics platform that lets anyone, tech-savvy or not, upload raw data and input any business question into an interface — similar to using an online search engine — and receive accurate answers in just 15 minutes.

The platform is based on the science of “social physics,” co-developed at the MIT Media Lab by Endor co-founders Alex “Sandy” Pentland, the Toshiba Professor of Media Arts and Sciences, and Yaniv Altshuler, a former MIT postdoc. Social physics uses mathematic models and machine learning to understand and predict crowd behaviors.

Users of the new platform upload data about customers or other individuals, such as records of mobile phone calls, credit card purchases, or web activity. They use Endor’s “query-builder” wizard to ask questions, such as “Where should we open our next store?” or “Who is likely to try product X?” Using the questions, the platform identifies patterns of previous behavior among the data and uses social physics models to predict future behavior. The platform can also analyze fully encrypted data-streams, allowing customers such as banks or credit card operators to maintain data privacy.

“It’s just like Google. You don’t have to spend time thinking, ‘Am I going to spend time asking Google this question?’ You just Google it,” Altshuler says. “It’s as simple as that.”

Financially backed by Innovation Endeavors, the private venture capital firm of Eric Schmidt, executive chairman of Google parent company Alphabet, Inc., the startup has found big-name customers, such as Coca-Cola, Mastercard, and Walmart, among other major retail and banking firms.

Recently, Endor analyzed Twitter data for a defense agency to detect potential terrorists. Endor was given 15 million data points containing examples of 50 Twitter accounts of identified ISIS activists, based on identifiers in the metadata. From that, they asked the startup to detect 74 with identifiers extremely well hidden in the metadata. Someone at Endor completed the task on a laptop in 24 minutes, detecting 80 “lookalike” ISIS accounts, 45 of which were from the pool of 74 well-hidden accounts named by the agency. The false positive rate was also extremely low (35 accounts), meaning that human analysts could afford to have experts investigating the accounts.

Finds patterns in materials recipes

Last month, three MIT materials scientists and their colleagues published a paper describing a new artificial-intelligence system that can pore through scientific papers and extract “recipes” for producing particular types of materials.

That work was envisioned as the first step toward a system that can originate recipes for materials that have been described only theoretically. Now, in a paper in the journal npj Computational Materials, the same three materials scientists, with a colleague in MIT’s Department of Electrical Engineering and Computer Science (EECS), take a further step in that direction, with a new artificial-intelligence system that can recognize higher-level patterns that are consistent across recipes.

For instance, the new system was able to identify correlations between “precursor” chemicals used in materials recipes and the crystal structures of the resulting products. The same correlations, it turned out, had been documented in the literature.

The system also relies on statistical methods that provide a natural mechanism for generating original recipes. In the paper, the researchers use this mechanism to suggest alternative recipes for known materials, and the suggestions accord well with real recipes.

The first author on the new paper is Edward Kim, a graduate student in materials science and engineering. The senior author is his advisor, Elsa Olivetti, the Atlantic Richfield Assistant Professor of Energy Studies in the Department of Materials Science and Engineering (DMSE). They’re joined by Kevin Huang, a postdoc in DMSE, and by Stefanie Jegelka, the X-Window Consortium Career Development Assistant Professor in EECS.

Sparse and scarce

Like many of the best-performing artificial-intelligence systems of the past 10 years, the MIT researchers’ new system is a so-called neural network, which learns to perform computational tasks by analyzing huge sets of training data. Traditionally, attempts to use neural networks to generate materials recipes have run up against two problems, which the researchers describe as sparsity and scarcity.

Any recipe for a material can be represented as a vector, which is essentially a long string of numbers. Each number represents a feature of the recipe, such as the concentration of a particular chemical, the solvent in which it’s dissolved, or the temperature at which a reaction takes place.

Since any given recipe will use only a few of the many chemicals and solvents described in the literature, most of those numbers will be zero. That’s what the researchers mean by “sparse.”

Similarly, to learn how modifying reaction parameters — such as chemical concentrations and temperatures — can affect final products, a system would ideally be trained on a huge number of examples in which those parameters are varied. But for some materials — particularly newer ones — the literature may contain only a few recipes. That’s scarcity.

“People think that with machine learning, you need a lot of data, and if it’s sparse, you need more data,” Kim says. “When you’re trying to focus on a very specific system, where you’re forced to use high-dimensional data but you don’t have a lot of it, can you still use these neural machine-learning techniques?”

Neural networks are typically arranged into layers, each consisting of thousands of simple processing units, or nodes. Each node is connected to several nodes in the layers above and below. Data is fed into the bottom layer, which manipulates it and passes it to the next layer, which manipulates it and passes it to the next, and so on. During training, the connections between nodes are constantly readjusted until the output of the final layer consistently approximates the result of some computation.

Artificial intelligence systems that process language

Neural networks, which learn to perform computational tasks by analyzing huge sets of training data, have been responsible for the most impressive recent advances in artificial intelligence, including speech-recognition and automatic-translation systems.

During training, however, a neural net continually adjusts its internal settings in ways that even its creators can’t interpret. Much recent work in computer science has focused on clever techniques for determining just how neural nets do what they do.

In several recent papers, researchers from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) and the Qatar Computing Research Institute have used a recently developed interpretive technique, which had been applied in other areas, to analyze neural networks trained to do machine translation and speech recognition.

They find empirical support for some common intuitions about how the networks probably work. For example, the systems seem to concentrate on lower-level tasks, such as sound recognition or part-of-speech recognition, before moving on to higher-level tasks, such as transcription or semantic interpretation.

But the researchers also find a surprising omission in the type of data the translation network considers, and they show that correcting that omission improves the network’s performance. The improvement is modest, but it points toward the possibility that analysis of neural networks could help improve the accuracy of artificial intelligence systems.

“In machine translation, historically, there was sort of a pyramid with different layers,” says Jim Glass, a CSAIL senior research scientist who worked on the project with Yonatan Belinkov, an MIT graduate student in electrical engineering and computer science. “At the lowest level there was the word, the surface forms, and the top of the pyramid was some kind of interlingual representation, and you’d have different layers where you were doing syntax, semantics. This was a very abstract notion, but the idea was the higher up you went in the pyramid, the easier it would be to translate to a new language, and then you’d go down again. So part of what Yonatan is doing is trying to figure out what aspects of this notion are being encoded in the network.”

Sensors could be sensitive enough for self-driving cars

For the past 10 years, the Camera Culture group at MIT’s Media Lab has been developing innovative imaging systems — from a camera that can see around corners to one that can read text in closed books — by using “time of flight,” an approach that gauges distance by measuring the time it takes light projected into a scene to bounce back to a sensor.

In a new paper appearing in IEEE Access, members of the Camera Culture group present a new approach to time-of-flight imaging that increases its depth resolution 1,000-fold. That’s the type of resolution that could make self-driving cars practical.

The new approach could also enable accurate distance measurements through fog, which has proven to be a major obstacle to the development of self-driving cars.

At a range of 2 meters, existing time-of-flight systems have a depth resolution of about a centimeter. That’s good enough for the assisted-parking and collision-detection systems on today’s cars.

But as Achuta Kadambi, a joint PhD student in electrical engineering and computer science and media arts and sciences and first author on the paper, explains, “As you increase the range, your resolution goes down exponentially. Let’s say you have a long-range scenario, and you want your car to detect an object further away so it can make a fast update decision. You may have started at 1 centimeter, but now you’re back down to [a resolution of] a foot or even 5 feet. And if you make a mistake, it could lead to loss of life.”

At distances of 2 meters, the MIT researchers’ system, by contrast, has a depth resolution of 3 micrometers. Kadambi also conducted tests in which he sent a light signal through 500 meters of optical fiber with regularly spaced filters along its length, to simulate the power falloff incurred over longer distances, before feeding it to his system. Those tests suggest that at a range of 500 meters, the MIT system should still achieve a depth resolution of only a centimeter.

Kadambi is joined on the paper by his thesis advisor, Ramesh Raskar, an associate professor of media arts and sciences and head of the Camera Culture group.

Slow uptake

With time-of-flight imaging, a short burst of light is fired into a scene, and a camera measures the time it takes to return, which indicates the distance of the object that reflected it. The longer the light burst, the more ambiguous the measurement of how far it’s traveled. So light-burst length is one of the factors that determines system resolution.

The other factor, however, is detection rate. Modulators, which turn a light beam off and on, can switch a billion times a second, but today’s detectors can make only about 100 million measurements a second. Detection rate is what limits existing time-of-flight systems to centimeter-scale resolution.

Associate department head for strategic directions

The Department of Electrical Engineering and Computer Science (EECS) has announced the appointment of two new associate department heads, and the creation of the new role of associate department head for strategic directions.

Professors Saman Amarasinghe and Joel Voldman have been named as new associate department heads, effective immediately, says EECS Department Head Asu Ozdaglar. Ozdaglar became department head on Jan. 1, replacing Anantha Chandrakasan, who is now dean of the School of Engineering. Professor Nancy Lynch will be the inaugural holder of the new position of associate department head for strategic directions, overseeing new academic and research initiatives.

“I am thrilled to be starting my own new role in collaboration with such a strong leadership team,” says Ozdaglar, who is also the Joseph F. and Nancy P. Keithley Professor of Electrical Engineering and Computer Science. “All three are distinguished scholars and dedicated educators whose experience will contribute greatly to shaping the department’s future.”

Saman Amarasinghe leads the Commit compiler research group at the Computer Science and Artificial Intelligence Laboratory (CSAIL). His group focuses on programming languages and compilers that maximize application performance on modern computing platforms. It has developed the Halide, TACO, Simit, StreamIt, StreamJIT, PetaBricks, MILK, Cimple, and GraphIt domain-specific languages and compilers, which all combine language design and sophisticated compilation techniques to deliver unprecedented performance for targeted application domains such as image processing, stream computations, and graph analytics.

Amarasinghe also pioneered the application of machine learning for compiler optimization, from Meta optimization in 2003 to OpenTuner extendable autotuner today. He was the co-leader of the Raw architecture project with EECS Professor and edX CEO Anant Agarwal. Recently, his work received a best-paper award at the 2017 Association for Computing Machinery (ACM) Object-Oriented Programming, Systems, Languages, and Applications (OOPSLA) conference and a best student-paper award at the 2017 Big Data conference.

Amarasinghe was the founder of Determina Inc., a startup based on computer security research pioneered in his MIT research group and later acquired by VMware. He is the faculty director for MIT Global Startup Labs, whose summer programs in 17 countries have helped launch more than 20 startups.