Researchers at MIT’s Computer Science and Artificial Intelligence Laboratory have developed a new computational model of a neural circuit in the brain, which could shed light on the biological role of inhibitory neurons — neurons that keep other neurons from firing.
The model describes a neural circuit consisting of an array of input neurons and an equivalent number of output neurons. The circuit performs what neuroscientists call a “winner-take-all” operation, in which signals from multiple input neurons induce a signal in just one output neuron.
Using the tools of theoretical computer science, the researchers prove that, within the context of their model, a certain configuration of inhibitory neurons provides the most efficient means of enacting a winner-take-all operation. Because the model makes empirical predictions about the behavior of inhibitory neurons in the brain, it offers a good example of the way in which computational analysis could aid neuroscience.
The researchers will present their results this week at the conference on Innovations in Theoretical Computer Science. Nancy Lynch, the NEC Professor of Software Science and Engineering at MIT, is the senior author on the paper. She’s joined by Merav Parter, a postdoc in her group, and Cameron Musco, an MIT graduate student in electrical engineering and computer science.
For years, Lynch’s group has studied communication and resource allocation in ad hoc networks — networks whose members are continually leaving and rejoining. But recently, the team has begun using the tools of network analysis to investigate biological phenomena.
“There’s a close correspondence between the behavior of networks of computers or other devices like mobile phones and that of biological systems,” Lynch says. “We’re trying to find problems that can benefit from this distributed-computing perspective, focusing on algorithms for which we can prove mathematical properties.”
In recent years, artificial neural networks — computer models roughly based on the structure of the brain — have been responsible for some of the most rapid improvement in artificial-intelligence systems, from speech transcription to face recognition software.
An artificial neural network consists of “nodes” that, like individual neurons, have limited information-processing power but are densely interconnected. Data are fed into the first layer of nodes. If the data received by a given node meet some threshold criterion — for instance, if it exceeds a particular value — the node “fires,” or sends signals along all of its outgoing connections.
Each of those outgoing connections, however, has an associated “weight,” which can augment or diminish a signal. Each node in the next layer of the network receives weighted signals from multiple nodes in the first layer; it adds them together, and again, if their sum exceeds some threshold, it fires. Its outgoing signals pass to the next layer, and so on.
In artificial-intelligence applications, a neural network is “trained” on sample data, constantly adjusting its weights and firing thresholds until the output of its final layer consistently represents the solution to some computational problem.
Machines that predict the future, robots that patch wounds, and wireless emotion-detectors are just a few of the exciting projects that came out of MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) this year. Here’s a sampling of 16 highlights from 2016 that span the many computer science disciplines that make up CSAIL.
Robots for exploring Mars — and your stomach
- A team led by CSAIL director Daniela Rus developed an ingestible origami robot that unfolds in the stomach to patch wounds and remove swallowed batteries.
- Researchers are working on NASA’s humanoid robot, “Valkyrie,” who will be programmed for trips into outer space and to autonomously perform tasks.
- A 3-D printed robot was made of both solids and liquids and printed in one single step, with no assembly required.
Keeping data safe and secure
- CSAIL hosted a cyber summit that convened members of academia, industry, and government, including featured speakers Admiral Michael Rogers, director of the National Security Agency; and Andrew McCabe, deputy director of the Federal Bureau of Investigation.
- Researchers came up with a system for staying anonymous online that uses less bandwidth to transfer large files between anonymous users.
- A deep-learning system called AI2 was shown to be able to predict 85 percent of cyberattacks with the help of some human input.
Advancements in computer vision
- A new imaging technique called Interactive Dynamic Video lets you reach in and “touch” objects in videos using a normal camera.
- Researchers from CSAIL and Israel’s Weizmann Institute of Science produced a movie display called Cinema 3D that uses special lenses and mirrors to allow viewers to watch 3-D movies in a theater without having to wear those clunky 3-D glasses.
- A new deep-learning algorithm can predict human interactions more accurately than ever before, by training itself on footage from TV shows like “Desperate Housewives” and “The Office.”
- A group from MIT and Harvard University developed an algorithm that may help astronomers produce the first image of a black hole, stitching together telescope data to essentially turn the planet into one large telescope dish.
The new TX-Green computing system at the MIT Lincoln Laboratory Supercomputing Center (LLSC) has been named the most powerful supercomputer in New England, 43rd most powerful in the U.S., and 106th most powerful in the world. A team of experts at TOP500 ranks the world’s 500 most powerful supercomputers biannually. The systems are ranked based on a LINPACK Benchmark, which is a measure of a system’s floating-point computing power, i.e., how fast a computer solves a dense system of linear equations.
Established in early 2016, the LLSC was developed to enhance computing power and accessibility for more than 1,000 researchers across the laboratory. The LLSC uses interactive supercomputing to augment the processing power of desktop systems to process large sets of sensor data, create high-fidelity simulations, and develop new algorithms. Located in Holyoke, Massachusetts, the new system is the only zero-carbon supercomputer on the TOP500 list; it uses energy from a mixture of hydroelectric, wind, solar, and nuclear sources.
In November, Dell EMC installed a new petaflop-scale system, which consists of 41,472 Intel processor cores and can compute 1015 operations per second. Compared to LLSC’s previous technology, the new system provides 6 times more processing power and 20 times more bandwidth. This technology enables research in several laboratory research areas, such as space observation, robotic vehicles, communications, cybersecurity, machine learning, sensor processing, electronic devices, bioinformatics, and air traffic control.
The LLSC mission is to address supercomputing needs, develop new supercomputing capabilities and technologies, and collaborate with MIT campus supercomputing initiatives. “The LLSC vision is to enable the brilliant scientists and engineers at Lincoln Laboratory to analyze and process enormous amounts of information with complex algorithms,” says Jeremy Kepner, Lincoln Laboratory Fellow and head of the LLSC. “Our new system is one of the largest on the East Coast and is specifically focused on enabling new research in machine learning, advanced physical devices, and autonomous systems.”
Compilers are programs that convert computer code written in high-level languages intelligible to humans into low-level instructions executable by machines.
But there’s more than one way to implement a given computation, and modern compilers extensively analyze the code they process, trying to deduce the implementations that will maximize the efficiency of the resulting software.
Code explicitly written to take advantage of parallel computing, however, usually loses the benefit of compilers’ optimization strategies. That’s because managing parallel execution requires a lot of extra code, and existing compilers add it before the optimizations occur. The optimizers aren’t sure how to interpret the new code, so they don’t try to improve its performance.
At the Association for Computing Machinery’s Symposium on Principles and Practice of Parallel Programming next week, researchers from MIT’s Computer Science and Artificial Intelligence Laboratory will present a new variation on a popular open-source compiler that optimizes before adding the code necessary for parallel execution.
As a consequence, says Charles E. Leiserson, the Edwin Sibley Webster Professor in Electrical Engineering and Computer Science at MIT and a coauthor on the new paper, the compiler “now optimizes parallel code better than any commercial or open-source compiler, and it also compiles where some of these other compilers don’t.”
That improvement comes purely from optimization strategies that were already part of the compiler the researchers modified, which was designed to compile conventional, serial programs. The researchers’ approach should also make it much more straightforward to add optimizations specifically tailored to parallel programs. And that will be crucial as computer chips add more and more “cores,” or parallel processing units, in the years ahead.
The idea of optimizing before adding the extra code required by parallel processing has been around for decades. But “compiler developers were skeptical that this could be done,” Leiserson says.
“Everybody said it was going to be too hard, that you’d have to change the whole compiler. And these guys,” he says, referring to Tao B. Schardl, a postdoc in Leiserson’s group, and William S. Moses, an undergraduate double major in electrical engineering and computer science and physics, “basically showed that conventional wisdom to be flat-out wrong. The big surprise was that this didn’t require rewriting the 80-plus compiler passes that do either analysis or optimization. T.B. and Billy did it by modifying 6,000 lines of a 4-million-line code base.”
One way to handle big data is to shrink it. If you can identify a small subset of your data set that preserves its salient mathematical relationships, you may be able to perform useful analyses on it that would be prohibitively time consuming on the full set.
The methods for creating such “coresets” vary according to application, however. Last week, at the Annual Conference on Neural Information Processing Systems, researchers from MIT’s Computer Science and Artificial Intelligence Laboratory and the University of Haifa in Israel presented a new coreset-generation technique that’s tailored to a whole family of data analysis tools with applications in natural-language processing, computer vision, signal processing, recommendation systems, weather prediction, finance, and neuroscience, among many others.
“These are all very general algorithms that are used in so many applications,” says Daniela Rus, the Andrew and Erna Viterbi Professor of Electrical Engineering and Computer Science at MIT and senior author on the new paper. “They’re fundamental to so many problems. By figuring out the coreset for a huge matrix for one of these tools, you can enable computations that at the moment are simply not possible.”
As an example, in their paper the researchers apply their technique to a matrix — that is, a table — that maps every article on the English version of Wikipedia against every word that appears on the site. That’s 1.4 million articles, or matrix rows, and 4.4 million words, or matrix columns.
That matrix would be much too large to analyze using low-rank approximation, an algorithm that can deduce the topics of free-form texts. But with their coreset, the researchers were able to use low-rank approximation to extract clusters of words that denote the 100 most common topics on Wikipedia. The cluster that contains “dress,” “brides,” “bridesmaids,” and “wedding,” for instance, appears to denote the topic of weddings; the cluster that contains “gun,” “fired,” “jammed,” “pistol,” and “shootings” appears to designate the topic of shootings.
Joining Rus on the paper are Mikhail Volkov, an MIT postdoc in electrical engineering and computer science, and Dan Feldman, director of the University of Haifa’s Robotics and Big Data Lab and a former postdoc in Rus’s group.
The researchers’ new coreset technique is useful for a range of tools with names like singular-value decomposition, principal-component analysis, and latent semantic analysis. But what they all have in common is dimension reduction: They take data sets with large numbers of variables and find approximations of them with far fewer variables.
In this, these tools are similar to coresets. But coresets are application-specific, while dimension-reduction tools are general-purpose. That generality makes them much more computationally intensive than coreset generation — too computationally intensive for practical application to large data sets.
People generally associate graphic processing units (GPUs) with imaging processing. Developed for video games in the 1990s, modern GPUs are specialized circuits with thousands of small, efficient processing units, or “cores,” that work simultaneously to rapidly render graphics on screen.
But for the better part of a decade, GPUs have also found general computing applications. Because of their incredible parallel-computing speeds and high-performance memory, GPUs are today used for advanced lab simulations and deep-learning programming, among other things.
Now, Todd Mostak, a former researcher at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL), is using GPUs to develop an analytic database and visualization platform called MapD, which is the fastest of its kind in the world, according to Mostak.
MapD is essentially a form of a commonly used database-management system that’s modified to run on GPUs instead of the central processing units (CPUs) that power most traditional database-management systems. By doing so, MapD can process billions of data points in milliseconds, making it 100 times faster than traditional systems. Moreover, MapD visualizes all processed data points nearly instantaneously — such as, say, plotting tweets on a world map — and parameters can be modified on the fly to adjust the visualized display.
With its first product launched last March, MapD’s clients already include Verizon and other big-name telecommunications companies, a social media giant, and financial and advertising firms. In October, the investment arm of the U.S. Central Intelligence Agency, In-Q-Tel, announced that it had invested in MapD’s latest funding round to accelerate the development of certain features for the U.S. intelligence community.
“[The CIA has] a lot of geospatial data, and they need to be able to form, visualize, and query that data in real-time. It’s a real need across the intelligence community,” Mostak says.
MIT researchers and their colleagues have developed a new computational model of the human brain’s face-recognition mechanism that seems to capture aspects of human neurology that previous models have missed.
The researchers designed a machine-learning system that implemented their model, and they trained it to recognize particular faces by feeding it a battery of sample images. They found that the trained system included an intermediate processing step that represented a face’s degree of rotation — say, 45 degrees from center — but not the direction — left or right.
This property wasn’t built into the system; it emerged spontaneously from the training process. But it duplicates an experimentally observed feature of the primate face-processing mechanism. The researchers consider this an indication that their system and the brain are doing something similar.
“This is not a proof that we understand what’s going on,” says Tomaso Poggio, a professor of brain and cognitive sciences at MIT and director of the Center for Brains, Minds, and Machines (CBMM), a multi-institution research consortium funded by the National Science Foundation and headquartered at MIT. “Models are kind of cartoons of reality, especially in biology. So I would be surprised if things turn out to be this simple. But I think it’s strong evidence that we are on the right track.”
Indeed, the researchers’ new paper includes a mathematical proof that the particular type of machine-learning system they use, which was intended to offer what Poggio calls a “biologically plausible” model of the nervous system, will inevitably yield intermediary representations that are indifferent to angle of rotation.
Poggio, who is also a primary investigator at MIT’s McGovern Institute for Brain Research, is the senior author on a paper describing the new work, which appeared today in the journal Computational Biology. He’s joined on the paper by several other members of both the CBMM and the McGovern Institute: first author Joel Leibo, a researcher at Google DeepMind, who earned his PhD in brain and cognitive sciences from MIT with Poggio as his advisor; Qianli Liao, an MIT graduate student in electrical engineering and computer science; Fabio Anselmi, a postdoc in the IIT@MIT Laboratory for Computational and Statistical Learning, a joint venture of MIT and the Italian Institute of Technology; and Winrich Freiwald, an associate professor at the Rockefeller University.
During January of her junior year at MIT, Caroline Colbert chose to do a winter externship at Massachusetts General Hospital (MGH). Her job was to shadow the radiation oncology staff, including the doctors that care for patients and medical physicists that design radiation treatment plans.
Colbert, now a senior in the Department of Nuclear Science and Engineering (NSE), had expected to pursue a career in nuclear power. But after working in a medical environment, she changed her plans.
She stayed at MGH to work on building a model to automate the generation of treatment plans for patients who will undergo a form of radiation therapy called volumetric-modulated arc therapy (VMAT). The work was so interesting that she is still involved with it and has now decided to pursue a doctoral degree in medical physics, a field that allows her to blend her training in nuclear science and engineering with her interest in medical technologies.
She’s even zoomed in on schools with programs that have accreditation from the Commission on Accreditation of Medical Physics Graduate Programs so she’ll have the option of having a more direct impact on patients. “I don’t know yet if I’ll be more interested in clinical work, research, or both,” she says. “But my hope is to work in a hospital setting.”
Many NSE students and faculty focus on nuclear energy technologies. But, says Colbert, “the department is really supportive of students who want to go into other industries.”
It was as a middle school student that Colbert first became interested in engineering. Later, in a chemistry class, a lesson about nuclear decay set her on a path towards nuclear science and engineering. “I thought it was so cool that one element can turn into another,” she says. “You think of elements as the fundamental building blocks of the physical world.”
Colbert’s parents, both from the Boston area, had encouraged her to apply to MIT. They also encouraged her towards the medical field. “They loved the idea of me being a doctor, and then when I decided on nuclear engineering, they wanted me to look into medical physics,” she says. “I was trying to make my own way. But when I did look seriously into medical physics, I had to admit that my parents were right.”
At MGH, Colbert’s work began with searching for practical ways to improve the generation of VMAT treatment plans. As with another form of radiation therapy called intensity-modulated radiation therapy (IMRT), the technology focuses radiation doses on the tumor and away from the healthy tissue surrounding it. The more accurate the dosing, the fewer side effects patients have after therapy.
It’s a fact of nature that a single conversation can be interpreted in very different ways. For people with anxiety or conditions such as Asperger’s, this can make social situations extremely stressful. But what if there was a more objective way to measure and understand our interactions?
Researchers from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) and Institute of Medical Engineering and Science (IMES) say that they’ve gotten closer to a potential solution: an artificially intelligent, wearable system that can predict if a conversation is happy, sad, or neutral based on a person’s speech patterns and vitals.
“Imagine if, at the end of a conversation, you could rewind it and see the moments when the people around you felt the most anxious,” says graduate student Tuka Alhanai, who co-authored a related paper with PhD candidate Mohammad Ghassemi that they will present at next week’s Association for the Advancement of Artificial Intelligence (AAAI) conference in San Francisco. “Our work is a step in this direction, suggesting that we may not be that far away from a world where people can have an AI social coach right in their pocket.”
As a participant tells a story, the system can analyze audio, text transcriptions, and physiological signals to determine the overall tone of the story with 83 percent accuracy. Using deep-learning techniques, the system can also provide a “sentiment score” for specific five-second intervals within a conversation.
“As far as we know, this is the first experiment that collects both physical data and speech data in a passive but robust way, even while subjects are having natural, unstructured interactions,” says Ghassemi. “Our results show that it’s possible to classify the emotional tone of conversations in real-time.”
The researchers say that the system’s performance would be further improved by having multiple people in a conversation use it on their smartwatches, creating more data to be analyzed by their algorithms. The team is keen to point out that they developed the system with privacy strongly in mind: The algorithm runs locally on a user’s device as a way of protecting personal information. (Alhanai says that a consumer version would obviously need clear protocols for getting consent from the people involved in the conversations.)
How it works
Many emotion-detection studies show participants “happy” and “sad” videos, or ask them to artificially act out specific emotive states. But in an effort to elicit more organic emotions, the team instead asked subjects to tell a happy or sad story of their own choosing.
Subjects wore a Samsung Simband, a research device that captures high-resolution physiological waveforms to measure features such as movement, heart rate, blood pressure, blood flow, and skin temperature. The system also captured audio data and text transcripts to analyze the speaker’s tone, pitch, energy, and vocabulary.
Daniela Rus loves Singapore. As the MIT professor sits down in her Frank Gehry-designed office in Cambridge, Massachusetts, to talk about her research conducted in Singapore, her face starts to relax in a big smile.
Her story with Singapore started in the summer of 2010, when she made her first visit to one of the most futuristic and forward-looking cities in the world. “It was love at first sight,” says the Andrew (1956) and Erna Viterbi Professor of Electrical Engineering and Computer Science and the director of MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL). That summer, she came to Singapore to join the Singapore-MIT Alliance for Research and Technology (SMART) as the first principal investigator in residence for the Future of Urban Mobility Research Program.
“In 2010, nobody was talking about autonomous driving. We were pioneers in developing and deploying the first mobility on demand for people with self-driving golf buggies,” says Rus. “And look where we stand today! Every single car maker is investing millions of dollars to advance autonomous driving. Singapore did not hesitate to provide us, at an early stage, with all the financial, logistical, and transportation resources to facilitate our work.”
Since her first visit, Rus has returned each year to follow up on the research, and has been involved in leading revolutionary projects for the future of urban mobility. “Our team worked tremendously hard on self-driving technologies, and we are now presenting a wide range of different devices that allow autonomous and secure mobility,” she says. “Our objective today is to make taking a driverless car for a spin as easy as programming a smartphone. A simple interaction between the human and machine will provide a transportation butler.”
The first mobility devices her team worked on were self-driving golf buggies. Two years ago, these buggies advanced to a point where the group decided to open them to the public in a trial that lasted one week at the Chinese Gardens, an idea facilitated by Singapore’s Land and Transportation Agency (LTA). Over the course of a week, more than 500 people booked rides from the comfort of their homes, and came to the Chinese Gardens at the designated time and spot to experience mobility-on-demand with robots.
The test was conducted around winding paths trafficked by pedestrians, bicyclists, and the occasional monitor lizard. The experiments also tested an online booking system that enabled visitors to schedule pickups and drop-offs around the garden, automatically routing and redeploying the vehicles to accommodate all the requests. The public’s response was joyful and positive, and this brought the team renewed enthusiasm to take the technology to the next level.
Since the Chinese Gardens public trial, the autonomous car group has introduced a few other self-driving vehicles: a self-driving city car, and two personal mobility robots, a self-driving scooter and a self-driving wheelchair. Each of these vehicles was created in three phases: In the first phase, the vehicle was converted to drive-by-wire control, which allows a computer to control acceleration, braking, and steering of the car. In the second phase, the vehicle drives on each of the pathways in its operation environment and makes a map using features detected by the sensors. In the third phase, the vehicle uses the map to compute a path from the customer’s pick-up point to the customer’s drop-off point and proceeds to drive along the path, localizing continuously and avoiding any other cars, people, and unexpected obstacles. The devices also used traffic data from LTA to model traffic patterns and to study the benefits of ride sharing systems.
Today, loading a web page on a big website usually involves a database query — to retrieve the latest contributions to a discussion you’re participating in, a list of news stories related to the one you’re reading, links targeted to your geographic location, or the like.
But database queries are time consuming, so many websites store — or “cache” — the results of common queries on web servers for faster delivery.
If a site user changes a value in the database, however, the cache needs to be updated, too. The complex task of analyzing a website’s code to identify which operations necessitate updates to which cached values generally falls to the web programmer. Missing one such operation can result in an unusable site.
This week, at the Association for Computing Machinery’s Symposium on Principles of Programming Languages, researchers from MIT’s Computer Science and Artificial Intelligence Laboratory presented a new system that automatically handles caching of database queries for web applications written in the web-programming language Ur/Web.
Although a website may be fielding many requests in parallel — sending different users different cached data, or even data cached on different servers — the system guarantees that, to the user, every transaction will look exactly as it would if requests were handled in sequence. So a user won’t, for instance, click on a link showing that tickets to an event are available, only to find that they’ve been snatched up when it comes time to pay.
In experiments involving two websites that had been built using Ur/Web, the new system’s automatic caching offered twofold and 30-fold speedups.
“Most very popular websites backed by databases don’t actually ask the database over and over again for each request,” says Adam Chlipala, an associate professor of electrical engineering and computer science at MIT and senior author on the conference paper. “They notice that, ‘Oh, I seem to have asked this question quite recently, and I saved the result, so I’ll just pull that out of memory.’”