More stories

  • in

    A comprehensive map of the SARS-CoV-2 genome

    In early 2020, a few months after the Covid-19 pandemic began, scientists were able to sequence the full genome of SARS-CoV-2, the virus that causes the Covid-19 infection. While many of its genes were already known at that point, the full complement of protein-coding genes was unresolved.

    Now, after performing an extensive comparative genomics study, MIT researchers have generated what they describe as the most accurate and complete gene annotation of the SARS-CoV-2 genome. In their study, which appears today in Nature Communications, they confirmed several protein-coding genes and found that a few others that had been suggested as genes do not code for any proteins.

    “We were able to use this powerful comparative genomics approach for evolutionary signatures to discover the true functional protein-coding content of this enormously important genome,” says Manolis Kellis, who is the senior author of the study and a professor of computer science in MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) as well as a member of the Broad Institute of MIT and Harvard.

    The research team also analyzed nearly 2,000 mutations that have arisen in different SARS-CoV-2 isolates since it began infecting humans, allowing them to rate how important those mutations may be in changing the virus’ ability to evade the immune system or become more infectious.

    Comparative genomics

    The SARS-CoV-2 genome consists of nearly 30,000 RNA bases. Scientists have identified several regions known to encode protein-coding genes, based on their similarity to protein-coding genes found in related viruses. A few other regions were suspected to encode proteins, but they had not been definitively classified as protein-coding genes.

    To nail down which parts of the SARS-CoV-2 genome actually contain genes, the researchers performed a type of study known as comparative genomics, in which they compare the genomes of similar viruses. The SARS-CoV-2 virus belongs to a subgenus of viruses called Sarbecovirus, most of which infect bats. The researchers performed their analysis on SARS-CoV-2, SARS-CoV (which caused the 2003 SARS outbreak), and 42 strains of bat sarbecoviruses.

    Kellis has previously developed computational techniques for doing this type of analysis, which his team has also used to compare the human genome with genomes of other mammals. The techniques are based on analyzing whether certain DNA or RNA bases are conserved between species, and comparing their patterns of evolution over time.

    Using these techniques, the researchers confirmed six protein-coding genes in the SARS-CoV-2 genome in addition to the five that are well established in all coronaviruses. They also determined that the region that encodes a gene called ORF3a also encodes an additional gene, which they name ORF3c. The gene has RNA bases that overlap with ORF3a but occur in a different reading frame. This gene-within-a-gene is rare in large genomes, but common in many viruses, whose genomes are under selective pressure to stay compact. The role for this new gene, as well as several other SARS-CoV-2 genes, is not known yet.

    The researchers also showed that five other regions that had been proposed as possible genes do not encode functional proteins, and they also ruled out the possibility that there are any more conserved protein-coding genes yet to be discovered.

    “We analyzed the entire genome and are very confident that there are no other conserved protein-coding genes,” says Irwin Jungreis, lead author of the study and a CSAIL research scientist. “Experimental studies are needed to figure out the functions of the uncharacterized genes, and by determining which ones are real, we allow other researchers to focus their attention on those genes rather than spend their time on something that doesn’t even get translated into protein.”

    The researchers also recognized that many previous papers used not only incorrect gene sets, but sometimes also conflicting gene names. To remedy the situation, they brought together the SARS-CoV-2 community and presented a set of recommendations for naming SARS-CoV-2 genes, in a separate paper published a few weeks ago in Virology.

    Fast evolution

    In the new study, the researchers also analyzed more than 1,800 mutations that have arisen in SARS-CoV-2 since it was first identified. For each gene, they compared how rapidly that particular gene has evolved in the past with how much it has evolved since the current pandemic began.

    They found that in most cases, genes that evolved rapidly for long periods of time before the current pandemic have continued to do so, and those that tended to evolve slowly have maintained that trend. However, the researchers also identified exceptions to these patterns, which may shed light on how the virus has evolved as it has adapted to its new human host, Kellis says.

    In one example, the researchers identified a region of the nucleocapsid protein, which surrounds the viral genetic material, that had many more mutations than expected from its historical evolution patterns. This protein region is also classified as a target of human B cells. Therefore, mutations in that region may help the virus evade the human immune system, Kellis says.

    “The most accelerated region in the entire genome of SARS-CoV-2 is sitting smack in the middle of this nucleocapsid protein,” he says. “We speculate that those variants that don’t mutate that region get recognized by the human immune system and eliminated, whereas those variants that randomly accumulate mutations in that region are in fact better able to evade the human immune system and remain in circulation.”

    The researchers also analyzed mutations that have arisen in variants of concern, such as the B.1.1.7 strain from England, the P.1 strain from Brazil, and the B.1.351 strain from South Africa. Many of the mutations that make those variants more dangerous are found in the spike protein, and help the virus spread faster and avoid the immune system. However, each of those variants carries other mutations as well.

    “Each of those variants has more than 20 other mutations, and it’s important to know which of those are likely to be doing something and which aren’t,” Jungreis says. “So, we used our comparative genomics evidence to get a first-pass guess at which of these are likely to be important based on which ones were in conserved positions.”

    This data could help other scientists focus their attention on the mutations that appear most likely to have significant effects on the virus’ infectivity, the researchers say. They have made the annotated gene set and their mutation classifications available in the University of California at Santa Cruz Genome Browser for other researchers who wish to use it.

    “We can now go and actually study the evolutionary context of these variants and understand how the current pandemic fits in that larger history,” Kellis says. “For strains that have many mutations, we can see which of these mutations are likely to be host-specific adaptations, and which mutations are perhaps nothing to write home about.”

    The research was funded by the National Human Genome Research Institute and the National Institutes of Health. Rachel Sealfon, a research scientist at the Flatiron Institute Center for Computational Biology, is also an author of the paper. More

  • in

    A robot that can help you untangle your hair

    With rapidly growing demands on health care systems, nurses typically spend 18 to 40 percent of their time performing direct patient care tasks, oftentimes for many patients and with little time to spare. Personal care robots that brush hair could provide substantial help and relief. 

    This may seem like a truly radical form of “self-care,” but crafty robots for things like shaving, hair-washing, and makeup are not new. In 2011, the tech giant Panasonic developed a robot that could wash, massage, and even blow-dry hair, explicitly designed to help support “safe and comfortable living of the elderly and people with limited mobility, while reducing the burden of caregivers.” 

    Hair-combing bots, however, proved to be less explored, leading scientists from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) and the Soft Math Lab at Harvard University to develop a robotic arm setup with a sensorized soft brush. The robot is equipped with a camera that helps it “see” and assess curliness, so it can plan a delicate and time-efficient brush-out.  

    Play video

    Robotic Hair Brushing

    The team’s control strategy is adaptive to the degree of tangling in the fiber bunch, and they put “RoboWig” to the test by brushing wigs ranging from straight to very curly hair.

    While the hardware setup of RoboWig looks futuristic and shiny, the underlying model of the hair fibers is what makes it tick. CSAIL postdoc Josie Hughes and her team opted to represent the entangled hair as sets of entwined double helices —  think classic DNA strands. This level of granularity provided key insights into mathematical models and control systems for manipulating bundles of soft fibers, with a wide range of applications in the textile industry, animal care, and other fibrous systems. 

    “By developing a model of tangled fibers, we understand from a model-based perspective how hairs must be entangled: starting from the bottom and slowly working the way up to prevent ‘jamming’ of the fibers,” says Hughes, the lead author on a paper about RoboWig. “This is something everyone who has brushed hair has learned from experience, but is now something we can demonstrate through a model, and use to inform a robot.”  

    This task at hand is a tangled one. Every head of hair is different, and the intricate interplay between hairs when combing can easily lead to knots. What’s more, if the incorrect brushing strategy is used, the process can be very painful and damaging to the hair. 

    Previous research in the brushing domain has mostly been on the mechanical, dynamic, and visual properties of hair, as opposed to RoboWig’s refined focus on tangling and combing behavior. 

    To brush and manipulate the hair, the researchers added a soft-bristled sensorized brush to the robot arm, to allow forces during brushing to be measured. They combined this setup with something called a “closed-loop control system,” which takes feedback from an output and automatically performs an action without human intervention. This created “force feedback” from the brush — a control method that lets the user feel what the device is doing — so the length of the stroke could be optimized to take into account both the potential “pain,” and time taken to brush. 

    Initial tests preserved the human head — for now — and instead were done on a number of wigs of various hair styles and types. The model provided insight into the behaviors of the combing, related to the number of entanglements, and how those could be efficiently and effectively brushed out by choosing appropriate brushing lengths. For example, for curlier hair, the pain cost would dominate, so shorter brush lengths were optimal. 

    The team wants to eventually perform more realistic experiments on humans, to better understand the performance of the robot with respect to their experience of pain — a metric that is obviously highly subjective, as one person’s “two” could be another’s “eight.”

    “To allow robots to extend their task-solving abilities to more complex tasks such as hair brushing, we need not only novel safe hardware, but also an understanding of the complex behavior of the soft hair and tangled fibers,” says Hughes. “In addition to hair brushing, the insights provided by our approach could be applied to brushing of fibers for textiles, or animal fibers.” 

    Hughes wrote the paper alongside Harvard University School of Engineering and Applied Sciences PhD students Thomas Bolton Plumb-Reyes and Nicholas Charles; Professor L. Mahadevan of Harvard’s School of Engineering and Applied Sciences, Department of Physics, and Organismic and Evolutionary Biology; and MIT professor and CSAIL Director Daniela Rus. They presented the paper virtually at the IEEE Conference on Soft Robotics (RoboSoft) earlier this month. 

    The project was supported, in part, by the National Science Foundation’s Emerging Frontiers in Research and Innovation program between MIT CSAIL and the Soft Math Lab at Harvard.  More

  • in

    Media Advisory — MIT researchers: AI policy needed to manage impacts, build more equitable systems

    On Thursday, May 6 and Friday, May 7, the AI Policy Forum — a global effort convened by researchers from MIT — will present their initial policy recommendations aimed at managing the effects of artificial intelligence and building AI systems that better reflect society’s values. Recognizing that there is unlikely to be any singular national AI policy, but rather public policies for the distinct ways in which we encounter AI in our lives, forum leaders will preview their preliminary findings and policy recommendations in three key areas: finance, mobility, and health care.

    The inaugural AI Policy Forum Symposium, a virtual event hosted by the MIT Schwarzman College of Computing, will bring together AI and public policy leaders, government officials from around the world, regulators, and advocates to investigate some of the pressing questions posed by AI in our economies and societies. The symposium’s program will feature remarks from public policymakers helping shape governments’ approaches to AI; state and federal regulators on the front lines of these issues; designers of self-driving cars and cancer-diagnosing algorithms; faculty examining the systems used in emerging finance companies and associated concerns; and researchers pushing the boundaries of AI.

    WHAT: AI Policy Forum (AIPF) Symposium

    WHO:MIT speakers: 

    Martin A. Schmidt, MIT provost
    Daniel Huttenlocher, AIPF chair and dean of the MIT Schwarzman College of Computing
    Regina Barzilay, MIT School of Engineering Distinguished Professor of AI and Health; AI faculty lead of the Jameel Clinic at MIT
    Daniel Weitzner, founding director of the MIT Internet Policy Research Initiative; former U.S. deputy chief technology officer in the Office of Science and Technology Policy
    Luis Videgaray, senior lecturer in the MIT Sloan School of Management; former foreign minister and minister of finance of Mexico
    Aleksander Madry, professor of computer science in the MIT Department of Electrical Engineering and Computer Science
    R. David Edelman, director of public policy for the MIT Internet Policy Research Initiative; former special assistant to U.S. President Barack Obama for economic and technology policy
    Julie Shah, MIT associate professor of aeronautics and astronautics; associate dean of social and ethical responsibilities of computing in the MIT Schwarzman College of Computing
    Andrew Lo, professor of finance in the MIT Sloan School of Management

    Guest speakers and participants: 

    Julie Bishop, chancellor of the Australian National University; former minister of foreign affairs and member of the Parliament of Australia
    Andrew Wyckoff, director for science, technology and innovation at the Organization for Economic Cooperation and Development (OECD)
    Martha Minow, 300th Anniversary University Professor at Harvard Law School; former dean of the Harvard Law School
    Alejandro Poiré, dean of the School of Public Policy at Monterrey Tec; former secretary of the interior of Mexico
    Ngaire Woods, dean of the Blavatnik School of Government at the University of Oxford
    Darran Anderson, director of strategy and innovation at the Texas Department of Transportation
    Nat Beuse, vice president of security at Aurora; former head safety regulator for autonomous vehicles at the U.S. Department of Transportation
    Laura Major, chief technology officer of Motional
    Manuela Veloso, head of AI research at JP Morgan Chase
    Stephanie Lee, managing director of BlackRock Systematic Active Equities Emerging Markets

    WHEN: Thursday and Friday, May 6 and 7

    Media RSVP:Reporters interested in attending can register here. More information on the AI Policy Forum can be found here.  More

  • in

    Nano flashlight enables new applications of light

    In work that could someday turn cell phones into sensors capable of detecting viruses and other minuscule objects, MIT researchers have built a powerful nanoscale flashlight on a chip.

    Their approach to designing the tiny light beam on a chip could also be used to create a variety of other nano flashlights with different beam characteristics for different applications. Think of a wide spotlight versus a beam of light focused on a single point.

    For many decades, scientists have used light to identify a material by observing how that light interacts with the material. They do so by essentially shining a beam of light on the material, then analyzing that light after it passes through the material. Because all materials interact with light differently, an analysis of the light that passes through the material provides a kind of “fingerprint” for that material. Imagine doing this for several colors — i.e., several wavelengths of light — and capturing the interaction of light with the material for each color. That would lead to a fingerprint that is even more detailed.

    Most instruments for doing this, known as spectrometers, are relatively large. Making them much smaller would have a number of advantages. For example, they could be portable and have additional applications (imagine a futuristic cell phone loaded with a self-contained sensor for a specific gas). However, while researchers have made great strides toward miniaturizing the sensor for detecting and analyzing the light that has passed through a given material, a miniaturized and appropriately shaped light beam—or flashlight—remains a challenge. Today that light beam is most often provided by macroscale equipment like a laser system that is not built into the chip itself as the sensors are.

    Complete sensor

    Enter the MIT work. In two recent papers in Nature Scientific Reports, researchers describe not only their approach for designing on-chip flashlights with a variety of beam characteristics, they also report building and successfully testing a prototype. Importantly, they created the device using existing fabrication technologies familiar to the microelectronics industry, so they are confident that the approach could be deployable at a mass scale with the lower cost that implies.

    Overall, this could enable industry to create a complete sensor on a chip with both light source and detector. As a result, the work represents a significant advance in the use of silicon photonics for the manipulation of light waves on microchips for sensor applications.

    “Silicon photonics has so much potential to improve and miniaturize the existing bench-scale biosensing schemes. We just need smarter design strategies to tap its full potential. This work shows one such approach,” says PhD candidate Robin Singh SM ’18, who is lead author of both papers.

    “This work is significant, and represents a new paradigm of photonic device design, enabling enhancements in the manipulation of optical beams,” says Dawn Tan, an associate professor at the Singapore University of Technology and Design who was not involved in the research.

    The senior coauthors on the first paper are Anuradha “Anu” Murthy Agarwal, a principal research scientist in MIT’s Materials Research Laboratory, Microphotonics Center, and Initiative for Knowledge and Innovation in Manufacturing; and Brian W. Anthony, a principal research scientist in MIT’s Department of Mechanical Engineering. Singh’s coauthors on the second paper are Agarwal; Anthony; Yuqi Nie, now at Princeton University; and Mingye Gao, a graduate student in MIT’s Department of Electrical Engineering and Computer Science.

    How they did it

    Singh and colleagues created their overall design using multiple computer modeling tools. These included conventional approaches based on the physics involved in the propagation and manipulation of light, and more cutting-edge machine-learning techniques in which the computer is taught to predict potential solutions using huge amounts of data. “If we show the computer many examples of nano flashlights, it can learn how to make better flashlights,” says Anthony. Ultimately, “we can then tell the computer the pattern of light that we want, and it will tell us what the design of the flashlight needs to be.”

    All of these modeling tools have advantages and disadvantages; together they resulted in a final, optimal design that can be adapted to create flashlights with different kinds of light beams.

    The researchers went on to use that design to create a specific flashlight with a collimated beam, or one in which the rays of light are perfectly parallel to each other. Collimated beams are key to some types of sensors. The overall flashlight that the researchers made involved some 500 rectangular nanoscale structures of different dimensions that the team’s modeling predicted would enable a collimated beam. Nanostructures of different dimensions would lead to different kinds of beams that in turn are key to other applications.

    The tiny flashlight with a collimated beam worked. Not only that, it provided a beam that was five times more powerful than is possible with conventional structures. That’s partly because “being able to control the light better means that less is scattered and lost,” says Agarwal.

    Singh describes the excitement he felt upon creating that first flashlight. “It was great to see through a microscope what I had designed on a computer. Then we tested it, and it worked!”

    This research was supported, in part, by the MIT Skoltech Initiative.

    Additional MIT facilities and departments that made this work possible are the Department of Materials Science and Engineering, the Materials Research Laboratory, the Institute for Medical Engineering and Science, and MIT.nano. More

  • in

    Climate solutions depend on technology, policy, and businesses working together

    “The challenge for humanity now is how to decarbonize the global economy by 2050. To do that, we need a supercharged decade of energy innovation,” said Ernest J. Moniz, the Cecil and Ida Green Professor of Physics and Engineering Systems Emeritus, founding director of the MIT Energy Initiative, and a former U.S. secretary of energy, as he opened the MIT Forefront virtual event on April 21. “But we also need practical visionaries, in every economic sector, to develop new business models that allow them to remain profitable while achieving the zero-carbon emissions.”

    The event, “Addressing Climate and Sustainability through Technology, Policy, and Business Models,” was the third in the MIT Forefront series, which invites top minds from the worlds of science, industry, and policy to propose bold new answers to urgent global problems. Moniz moderated the event, and more than 12,000 people tuned in online.

    MIT and other universities play an important role in preparing the world’s best minds to take on big climate challenges and develop the technology needed to advance sustainability efforts, a point illustrated in the main session with a video about Via Separations, a company supported by MIT’s The Engine. Co-founded by Shreya Dave ’09, SM ’12, PhD ’16, Via Separations customizes filtration technology to reduce waste and save money across multiple industries. “By next year, we are going to be eliminating carbon dioxide emissions from our customers’ facilities,” Dave said.

    Via Separations is one of many innovative companies born of MIT’s energy and climate initiatives — the work of which, as the panel went on to discuss, is critical to achieving net-zero emissions and deploying successful environmental sustainability efforts. As Moniz put it, the company embodies “the spirit of science and technology in action for the good of humankind” and exemplifies how universities and businesses, as well as technology and policy, must work together to make the best environmental choices.

    How businesses confront climate change

    Innovation in sustainable practices can be met with substantial challenges when proposed or applied to business models, particularly on the policy side, the panelists noted. But they shared some key ways that their respective organizations have employed current technologies and the challenges they face in reaching their sustainability goals. Despite each business’s different products and services, a common thread of needing new technologies to achieve their sustainability goals emerged. 

    Although 2050 is the long-term goal for net-zero emissions put forth by the Paris Agreement, the businesses represented by the panelists are thinking about the shorter term. “IBM has committed to net-zero emissions by 2030 ― without carbon offsets,” said Arvind Krishna, chairman and chief executive officer of IBM. “We believe that some carbon taxes would be a good policy tool. But policy alone is insufficient. We need advanced technological tools to reach our goal.” 

    Jeff Wilke SM ’93, who retired as Amazon’s chief executive officer of Worldwide Consumer in February 2021, outlined a number of initiatives that the online retail giant is undertaking to curb emissions. Transportation is one of their biggest hurdles to reaching zero emissions, leading to a significant investment in Class 8 electric trucks. “Another objective is to remove the need for plane shipments by getting inventory closer to urban areas, and that has been happening steadily over the years,” he said.

    Jim Fitterling, chair and chief executive officer of Dow, explained that Dow has reduced its carbon emissions by 15 percent in the past decade and is poised to reduce it further in the next. Future goals include working toward electrifying ethylene production. “If we can electrify that, it will allow us to make major strides toward carbon-dioxide reduction,” he said. “But we need more reliable and stable power to get to that point.” 

    Collaboration is key to advancing climate solutions

    Maria T. Zuber, MIT’s vice president for research, who was recently appointed by U.S. President Joe Biden as co-chair of the President’s Council of Advisors on Science and Technology, stressed that MIT innovators and industry leaders must work together to implement climate solutions. 

    “Innovation is a team sport,” said Zuber, who is also the E. A. Griswold Professor of Geophysics. “Even if MIT researchers make a huge discovery, deploying it requires cooperation on a policy level and often industry support. Policymakers need to solve problems and seize opportunities in ways that are popular. It’s not just solving technical problems ― there is a human behavior component.”

    But businesses, Zuber said, can play a huge role in advancing innovation. “If a company becomes convinced of the potential of a new technology, they can be the best advocates with policymakers,” she said.

    The question of “sustainability vs. shareholders” 

    During the Q&A session, an audience member pointed out that environmentalists are often distrustful of companies’ sustainability policies when their focus is on shareholders and profit.

    “Companies have to show that they’re part of the solution,” Fitterling said. “Investors will be afraid of high costs up front, so, say, completely electrifying a plant overnight is off the table. You have to make a plan to get there, and then incentivize that plan through policy. Carbon taxes are one way, but miss the market leverage.”

    Krishna also pushed back on the idea that companies have to choose between sustainability and profit. “It’s a false dichotomy,” he said. “If companies were only interested in short-term profits, they wouldn’t last for long.”

    “A belief I’ve heard from some environmental groups is that ‘anything a company does is greenwashing,’ and that they’ll abandon those efforts if the economy tanks,” Zuber said, referring to a practice wherein organizations spend more time marketing themselves as environmentally sustainable than on maximizing their sustainability efforts. “The economy tanked in 2020, though, and we saw companies double down on their sustainability plans. They see that it’s good for business.”

    The role of universities and businesses in sustainability innovation

    “Amazon and all corporations are adapting to the effects of climate change, like extreme weather patterns, and will need to adapt more — but I’m not ready to throw in the towel for decarbonization,” Wilke said. “Either way, companies will have to invest in decarbonization. There is no way we are going to make the progress we have to make without it.” 

    Another component is the implications of artificial intelligence (AI) and quantum computing. Krishna noted multiple ways that AI and quantum computing will play a role at IBM, including finding the most environmentally sustainable and cost-efficient ways to advance carbon separation in exhaust gases and lithium battery life in electric cars. 

    AI, quantum computing, and alternate energy sources such as fusion energy that have the potential to achieve net-zero energy, are key areas that students, researchers, and faculty members are pursuing at MIT.

    “Universities like MIT need to go as fast as we can as far as we can with the science and technology we have now,” Zuber said. “In parallel, we need to invest in and deploy a suite of new tools in science and technology breakthroughs that we need to reach the 2050 goal of decarbonizing. Finally, we need to continue to train the next generation of students and researchers who are solving these issues and deploy them to these companies to figure it out.” More

  • in

    With a zap of light, system switches objects’ colors and patterns

    When was the last time you repainted your car? Redesigned your coffee mug collection? Gave your shoes a colorful facelift?

    You likely answered: never, never, and never. You might consider these arduous tasks not worth the effort. But a new color-shifting “programmable matter” system could change that with a zap of light.

    MIT researchers have developed a way to rapidly update imagery on object surfaces. The system, dubbed “ChromoUpdate” pairs an ultraviolet (UV) light projector with items coated in light-activated dye. The projected light alters the reflective properties of the dye, creating colorful new images in just a few minutes. The advance could accelerate product development, enabling product designers to churn through prototypes without getting bogged down with painting or printing.

    An ultraviolet (UV) light projector is used on a cell-phone case coated in light-activated dye. The projected light alters the reflective properties of the dye, creating images in just a few minutes.

    ChromoUpdate “takes advantage of fast programming cycles — things that wouldn’t have been possible before,” says Michael Wessley, the study’s lead author and a postdoc in MIT’s Computer Science and Artificial Intelligence Laboratory.

    The research will be presented at the ACM Conference on Human Factors in Computing Systems this month. Wessely’s co-authors include his advisor, Professor Stefanie Mueller, as well as postdoc Yuhua Jin, recent graduate Cattalyya Nuengsigkapian ’19, MNG ’20, visiting master’s student Aleksei Kashapov, postdoc Isabel Qamar, and Professor Dzmitry Tsetserukou of the Skolkovo Institute of Science and Technology.

    ChromoUpdate builds on the researchers’ previous programmable matter system, called PhotoChromeleon. That method was “the first to show that we can have high-resolution, multicolor textures that we can just reprogram over and over again,” says Wessely. PhotoChromeleon used a lacquer-like ink comprising cyan, magenta, and yellow dyes. The user covered an object with a layer of the ink, which could then be reprogrammed using light. First, UV light from an LED was shone on the ink, fully saturating the dyes. Next, the dyes were selectively desaturated with a visible light projector, bringing each pixel to its desired color and leaving behind the final image. PhotoChromeleon was innovative, but it was sluggish. It took about 20 minutes to update an image. “We can accelerate the process,” says Wessely.

    They achieved that with ChromoUpdate, by fine-tuning the UV saturation process. Rather than using an LED, which uniformly blasts the entire surface, ChromoUpdate uses a UV projector that can vary light levels across the surface. So, the operator has pixel-level control over saturation levels. “We can saturate the material locally in the exact pattern we want,” says Wessely. That saves time — someone designing a car’s exterior might simply want to add racing stripes to an otherwise completed design. ChromoUpdate lets them do just that, without erasing and reprojecting the entire exterior.

    This selective saturation procedure allows designers to create a black-and-white preview of a design in seconds, or a full-color prototype in minutes. That means they could try out dozens of designs in a single work session, a previously unattainable feat. “You can actually have a physical prototype to see if your design really works,” says Wessely. “You can see how it looks when sunlight shines on it or when shadows are cast. It’s not enough just to do this on a computer.”

    Play video

    That speed also means ChromoUpdate could be used for providing real-time notifications without relying on screens. “One example is your coffee mug,” says Wessely. “You put your mug in our projector system and program it to show your daily schedule. And it updates itself directly when a new meeting comes in for that day, or it shows you the weather forecast.”

    Wessely hopes to keep improving the technology. At present, the light-activated ink is specialized for smooth, rigid surfaces like mugs, phone cases, or cars. But the researchers are working toward flexible, programmable textiles. “We’re looking at methods to dye fabrics and potentially use light-emitting fibers,” says Wessely. “So, we could have clothing — t-shirts and shoes and all that stuff — that can reprogram itself.”

    The researchers have partnered with a group of textile makers in Paris to see how ChomoUpdate can be incorporated into the design process.

    This research was funded, in part, by Ford. More

  • in

    Undergraduates explore practical applications of artificial intelligence

    Deep neural networks excel at finding patterns in datasets too vast for the human brain to pick apart. That ability has made deep learning indispensable to just about anyone who deals with data. This year, the MIT Quest for Intelligence and the MIT-IBM Watson AI Lab sponsored 17 undergraduates to work with faculty on yearlong research projects through MIT’s Advanced Undergraduate Research Opportunities Program (SuperUROP).

    Students got to explore AI applications in climate science, finance, cybersecurity, and natural language processing, among other fields. And faculty got to work with students from outside their departments, an experience they describe in glowing terms. “Adeline is a shining testament of the value of the UROP program,” says Raffaele Ferrari, a professor in MIT’s Department of Earth and Planetary Sciences, of his advisee. “Without UROP, an oceanography professor might have never had the opportunity to collaborate with a student in computer science.”

    Highlighted below are four SuperUROP projects from this past year.

    A faster algorithm to manage cloud-computing jobs

    The shift from desktop computing to far-flung data centers in the “cloud” has created bottlenecks for companies selling computing services. Faced with a constant flux of orders and cancellations, their profits depend heavily on efficiently pairing machines with customers.

    Approximation algorithms are used to carry out this feat of optimization. Among all the possible ways of assigning machines to customers by price and other criteria, they find a schedule that achieves near-optimal profit.​ For the last year, junior Spencer Compton worked on a virtual whiteboard with MIT Professor Ronitt Rubinfeld and postdoc Slobodan Mitrović to find a faster scheduling method.

    “We didn’t write any code,” he says. “We wrote proofs and used mathematical ideas to find a more efficient way to solve this optimization problem. The same ideas that improve cloud-computing scheduling can be used to assign flight crews to planes, among other tasks.”

    In a pre-print paper on arXiv, Compton and his co-authors show how to speed up an approximation algorithm under dynamic conditions. They also show how to locate machines assigned to individual customers without computing the entire schedule.

    A big challenge was finding the crux of the project, he says. “There’s a lot of literature out there, and a lot of people who have thought about related problems. It was fun to look at everything that’s been done and brainstorm to see where we could make an impact.”​

    How much heat and carbon can the oceans absorb?

    Earth’s oceans regulate climate by drawing down excess heat and carbon dioxide from the air. But as the oceans warm, it’s unclear if they will soak up as much carbon as they do now. A slowed uptake could bring about more warming than what today’s climate models predict. It’s one of the big questions facing climate modelers as they try to refine their predictions for the future.

    The biggest obstacle in their way is the complexity of the problem: today’s global climate models lack the computing power to get a high-resolution view of the dynamics influencing key variables like sea-surface temperatures. To compensate for the lost accuracy, researchers are building surrogate models to approximate the missing dynamics without explicitly solving for them.

    In a project with MIT Professor Raffaele Ferrari and research scientist Andre Souza, MIT junior Adeline Hillier is exploring how deep learning solutions can be used to improve or replace physical models of the uppermost layer of ocean, which drives the rate of heat and carbon uptake. “If the model has a small footprint and succeeds under many of the physical conditions encountered in the real world, it could be incorporated into a global climate model and hopefully improve climate projections,” she says.

    In the course of the project, Hillier learned how to code in the programming language Julia. She also got a crash course in fluid dynamics. “You’re trying to model the effects of turbulent dynamics in the ocean,” she says. “It helps to know what the processes and physics behind them look like.”

    In search of more efficient deep learning models

    There are thousands of ways to design a deep learning model to solve a given task. Automating the design process promises to narrow the options and make these tools more accessible. But finding the optimal architecture is anything but simple. Most automated searches pick the model that maximizes validation accuracy without considering the structure of the underlying data, which may suggest a simpler, more robust solution. As a result, more reliable or data-efficient architectures are passed over.

    “Instead of looking at the accuracy of the model alone, we should focus on the structure of the data,” says MIT senior Kristian Georgiev. In a project with MIT Professor Asu Ozdaglar and graduate student Alireza Fallah, Georgiev is looking at ways to automatically query the data to find the model that best suits its constraints. “If you choose your architecture based on the data, you’re more likely to get a good and robust solution from a learning theory perspective,” he says.

    The hardest part of the project was the exploratory phase at the start, he says. To find a good research question he read through papers ranging from topics in autoML to representation theory. But it was worth it, he says, to be able to work at the intersection of optimization and generalization. “To make good progress in machine learning you need to combine both of these fields.”

    What makes humans so good at recognizing faces?

    Face recognition comes easily to humans. Picking out familiar faces in a blurred or distorted photo is a cinch. But we don’t really understand why or how to replicate this superpower in machines. To home in on the principles important to recognizing faces, researchers have shown headshots to human subjects that are progressively degraded to see where recognition starts to break down. They are now performing similar experiments on computers to see if deeper insights can be gained

    In a project with MIT Professor Pawan Sinha and the MIT Quest for Intelligence, junior Ashika Verma applied a set of filters to a dataset of celebrity photos. She blurred their faces, distorted them, and changed their color to see if a face-recognition model could pick out photos of the same face. She found that the model did best when the photos were either natural color or grayscale, consistent with the human studies. Accuracy slipped when a color filter was added, but not as much as it did for the human subjects — a wrinkle that Verma plans to investigate further.

    The work is part of a broader effort to understand what makes humans so good at recognizing faces, and how machine vision might be improved as a result. It also ties in with Project Prakash, a nonprofit in India that treats blind children and tracks their recovery to learn more about the visual system and brain plasticity. “Running human experiments takes more time and resources than running computational experiments,” says Verma’s advisor, Kyle Keane, a researcher with MIT Quest. “We’re trying to make AI as human-like as possible so we can run a lot of computational experiments to identify the most promising experiments to run on humans.”

    Degrading the images to use in the experiments, and then running them through the deep nets, was a challenge, says Verma. “It’s very slow,” she says. “You work 20 minutes at a time and then you wait.” But working in a lab with an advisor made it worth it, she says. “It was fun to dip my toes into neuroscience.”

    SuperUROP projects were funded, in part, by the MIT-IBM Watson AI Lab, MIT Quest Corporate, and by Eric Schmidt, technical advisor to Alphabet Inc., and his wife, Wendy. More

  • in

    Q&A: Vivienne Sze on crossing the hardware-software divide for efficient artificial intelligence

    Not so long ago, watching a movie on a smartphone seemed impossible. Vivienne Sze was a graduate student at MIT at the time, in the mid 2000s, and she was drawn to the challenge of compressing video to keep image quality high without draining the phone’s battery. The solution she hit upon called for co-designing energy-efficient circuits with energy-efficient algorithms.

    Sze would go on to be part of the team that won an Engineering Emmy Award for developing the video compression standards still in use today. Now an associate professor in MIT’s Department of Electrical Engineering and Computer Science, Sze has set her sights on a new milestone: bringing artificial intelligence applications to smartphones and tiny robots.

    Her research focuses on designing more-efficient deep neural networks to process video, and more-efficient hardware to run those applications. She recently co-published a book on the topic, and will teach a professional education course on how to design efficient deep learning systems in June.

    On April 29, Sze will join Assistant Professor Song Han for an MIT Quest AI Roundtable on the co-design of efficient hardware and software moderated by Aude Oliva, director of MIT Quest Corporate and the MIT director of the MIT-IBM Watson AI Lab. Here, Sze discusses her recent work.

    Q: Why do we need low-power AI now?

    A: AI applications are moving to smartphones, tiny robots, and internet-connected appliances and other devices with limited power and processing capabilities. The challenge is that AI has high computing requirements. Analyzing sensor and camera data from a self-driving car consumes about 2,500 watts, but the computing budget of a smartphone is just about a single watt. Closing this gap requires rethinking the entire stack, a trend that will define the next decade of AI.

    Q: What’s the big deal about running AI on a smartphone?

    A: It means that the data processing no longer has to take place in the “cloud,” on racks of warehouse servers. Untethering compute from the cloud allows us to broaden AI’s reach. It gives people in developing countries with limited communication infrastructure access to AI. It also speeds up response time by reducing the lag caused by communicating with distant servers. This is crucial for interactive applications like autonomous navigation and augmented reality, which need to respond instantaneously to changing conditions. Processing data on the device can also protect medical and other sensitive records. Data can be processed right where they’re collected.

    Q: What makes modern AI so inefficient?

    A: The cornerstone of modern AI — deep neural networks — can require hundreds of millions to billions of calculations — orders of magnitude greater than compressing video on a smartphone. But it’s not just number crunching that makes deep networks energy-intensive — it’s the cost of shuffling data to and from memory to perform these computations. The farther the data have to travel, and the more data there are, the greater the bottleneck.

    Q: How are you redesigning AI hardware for greater energy efficiency?

    A: We focus on reducing data movement and the amount of data needed for computation. In some deep networks, the same data are used multiple times for different computations. We design specialized hardware to reuse data locally rather than send them off-chip. Storing reused data on-chip makes the process extremely energy-efficient.  

    We also optimize the order in which data are processed to maximize their reuse. That’s the key property of the Eyeriss chip that I co-designed with Joel Emer. In our followup work, Eyeriss v2, we made the chip flexible enough to reuse data across a wider range of deep networks. The Eyeriss chip also uses compression to reduce data movement, a common tactic among AI chips. The low-power Navion chip that I co-designed with Sertac Karaman for mapping and navigation applications in robotics uses two to three orders of magnitude less energy than a CPU, in part by using optimizations that reduce the amount of data processed and stored on-chip. 

    Q: What changes have you made on the software side to boost efficiency?

    A: The more that software aligns with hardware-related performance metrics like energy efficiency, the better we can do. Pruning, for example, is a popular way to remove weights from a deep network to reduce computation costs. But rather than remove weights based on their magnitude, our work on energy-aware pruning suggests you can remove the more energy-intensive weights to improve overall energy consumption. Another method we’ve developed, NetAdapt, automates the process of adapting and optimizing a deep network for a smartphone or other hardware platforms. Our recent followup work, NetAdaptv2, accelerates the optimization process to further boost efficiency.

    Q: What low-power AI applications are you working on?

    A: I’m exploring autonomous navigation for low-energy robots with Sertac Karaman. I’m also working with Thomas Heldt to develop a low-cost and potentially more effective way of diagnosing and monitoring people with neurodegenerative disorders like Alzheimer’s and Parkinson’s by tracking their eye movements. Eye-movement properties like reaction time could potentially serve as biomarkers for brain function. In the past, eye-movement tracking took place in clinics because of the expensive equipment required. We’ve shown that an ordinary smartphone camera can take measurements from a patient’s home, making data collection easier and less costly. This could help to monitor disease progression and track improvements in clinical drug trials.

    Q: Where is low-power AI headed next?

    A: Reducing AI’s energy requirements will extend AI to a wider range of embedded devices, extending its reach into tiny robots, smart homes, and medical devices. A key challenge is that efficiency often requires a tradeoff in performance. For wide adoption, it will be important to dig deeper into these different applications to establish the right balance between efficiency and accuracy. More