Lighting the path

When she was an MIT undergraduate studying electrical engineering, Jeannette Wing ’78, SM ’79, PhD ’83 took a required computer science class and began thinking about changing her major. But before making the decision, she called her father, a professor of electrical engineering at Columbia University, to ask one big question: Is computer science just a fad?

“I literally remember asking him that question,” Wing said, drawing chuckles from an audience of MIT students and faculty. Wing’s father assured her that computer science was here to stay. “So I switched,” said Wing, who is herself now the Avanessians Director of the Data Science Institute and professor of computer science at Columbia. “And I’ve never looked back.”

Patti Maes’ career path also began in college, when she couldn’t decide between two majors. Because of an economic downturn at the time, she also worried about the employment prospects in both fields. “I got interested in computer science as a way of choosing between biology and architecture” — and ensuring that she could find a job after graduation, she said. Later, as a visiting professor and research scientist at MIT, Maes began working with robots and artificial intelligence (AI), but eventually moved to the MIT Media Lab, where she is now a professor media of arts and sciences. She said she’s more interesting in human intelligence, focusing on, for instance, how to help people improve their memories, become more creative, and listen better — “these soft skills that we really desperately need to do well in life.”

Wing and Maes were among five academic and industry leaders who participated in “Perspectives from Luminaries: A Panel on Computing and Cognition” in Huntington Hall on Tuesday. The two were joined by MIT Institute Professor and computer scientist Barbara Liskov; Laura Schulz, MIT professor of cognitive science; and Jaime Teevan SM ’01, PhD ’07, chief scientist for Microsoft’s Experiences and Devices product team. The panel was moderated by Stefanie Mueller, the X-Window Consortium Career Development Professor in the Department of Electrical Engineering and Computer Science (EECS), and Vivienne Sze, an associate professor in EECS.

In introducing the panel, MIT Chancellor Cynthia Barnhart noted that the event capped the opening day of this week’s campus-wide celebration of the new MIT Stephen A. Schwarzman College of Computing. “The theme for today was ‘explore,’” Barnhart said. “Tonight, we’re here to listen to and learn from true luminaries.” 

Liskov noted that the event organizers had asked panelists to send a photo of themselves with their first computers. “When I was growing up, there were computers — some,” she said, describing the room-sized machines of the time. However, computer science wasn’t yet an academic discipline, and it didn’t occur to her to study engineering because, in the late 1950s, that was “not something girls did.” So she earned a bachelor’s degree in mathematics from the University of California at Berkeley, then looked for a job. “That’s when I discovered computers,” she said.

At the time, companies needed programmers. But with no computer-science graduates yet available, Liskov said, employers would hire anyone with expertise that would let them quickly pick up programming skills. So despite knowing nothing about programming, Liskov landed a job at MITRE Corp. On her first day, someone handed her a FORTRAN manual and told her to write a program to solve a problem, and that began her long and distinguished career in computer science. "I was in the right place at right time," she said, adding later: "I was lucky to get into computer science very early, when there were huge problems just waiting to be  worked on."

After a year, she left MITRE for a programming job at Harvard University, working on computer translation of human language, before returning to graduate school. When she received a PhD in computer science from Stanford University in 1968, she was one of the first women in the United States to earn a doctorate in the field. She returned to MITRE for a few years, then joined the MIT faculty in 1972. She was named an Institute Professor, MIT’s highest faculty honor, in 2008. A year later, she received the Association for Computing Machinery’s A.M. Turing Award, sometimes described as “the Nobel Prize of computing,” in recognition of her contributions to programming language and system design.

Liskov’s definition for computing hasn’t changed much since that first job at MITRE. “One thing that stuck with me all these years is [viewing] computing as a way of solving problems,” Liskov said, drawing nods from many in the audience. “That’s what’s meant by computational thinking.” But she added a warning: “I hope that we as a society learn to tame the technology that we have with us now and make good choices about the technology that’s coming.”

Teevan summed up her story this way: “It’s about being the wrong place at the right time.” After receiving a bachelor’s degree in computer science from Yale University, Teevan worked for a couple of years as a software engineer, then headed to MIT for graduate study. At the time, she said, MIT didn’t offer much in the way of information retrieval or human-computer interaction, so initially, she wondered whether she’d made a mistake. Fortunately, she said, those research in those fields evolved rapidly during her time at the Institute (as did her family; she gave birth to her first three children while in graduate school and later had a fourth).

Teevan has been with Microsoft since 2006, first with Microsoft Research, then as technical advisor to CEO Satya Nadella, and now as chief scientist for Microsoft’s Experience and Devices product team. Her advice to the audience: Keep your entire career in perspective. “Take the time not just to look forward, but to look back, to reflect on what you’re doing,” she said.

Schulz described her computing experience as comparatively brief. “I got a TRS-80 when I was, I guess, 11,” she said, referring to the early desktop microcomputer, and the last time she did any programming was shortly after that. Instead, she focused on human intelligence while studying for a bachelor’s degree in philosophy at the University of Michigan. “I was interested in how human learners engage with the world,” she said. After receiving a master’s degree and PhD in developmental psychology from the University of California at Berkeley, she joined the faculty of MIT’s Department of Brain and Cognitive Sciences. Given her nontechnical background, she recalled, friends jokingly asked whether she knew what the “T” in “MIT” stands for. But she added that there’s an obvious need to study both kinds of intelligence: “It’s very clear that children learn in different ways than powerful machines are learning.”

Some panelists offered the audience a peek at research in progress. Maes is exploring how ubiquitous smart devices might positively impact human mental health and psychology. She described how these devices — which already know a great deal about their users — could potentially “intervene in the moment and maybe talk to us when we’re about to engage in a behavior we want to change, or help calm us down when we’re stressed out or anxious, or remind us to be attentive when we’re sitting in a lecture.” Wing, meanwhile, is trying to build a community around what she calls “trustworthy AI.” She noted that researchers understand AI’s exciting potential, “but we also recognize the danger of models that are unpredictable, or unexplainable, or not fair, in the sense of discrimination,” she said. “We need to find ways for people and society to actually trust these systems.”

The panelists said that opportunities for women have improved in both academia and industry, but much work remains to be done. “We just have to keep on going, and pushing,” Liskov said. “One thing that helps is women looking out for women.”

And, she said, women must also look out for themselves. “Sometimes a door will open and you have to decide whether to step through it or not,” she said. “You have to come to grips with what you want, and not what somebody wants for you — something you’re good at and that makes you happy.”

de MIT News https://ift.tt/2EouMNl

Reading the heartbeat of the road

Smartphones can set your thermostat, control your lights, and even monitor your heart rate. But thanks to MIT Concrete Sustainability Hub research assistant Jake Roxon, they will also soon be able to measure pavement quality and reduce vehicle emissions.

In collaboration with Harvard University student Shahd Nara, Roxon has applied his love of cars with his engineering expertise to create Carbin, a crowdsourcing app that measures road quality and eventually will guide drivers on the most fuel-efficient route. The app utilizes uses research conducted by the MIT Concrete Sustainability Hub to cut a vehicle’s emissions by an estimated 5 to 10 percent.

The key to Carbin is a phone’s accelerometer. Ubiquitous in smartphones, the accelerometer measures orientation and local acceleration, allowing the device to accomplish tasks like changing the position of a screen or measuring footsteps. But most phones don’t fully utilize their accelerometers; smartphone accelerometers generally take measurements far below their practical potential of 100 hertz, a frequency of 100 times per second. By tapping into this reserve, Roxon can increase the sensitivity of a smartphone to detect the slightest defects in pavements, all from within a moving car’s cabin.

After recording these pavement defects with the accelerometer, Roxon can then quantify a road’s roughness, meaning he has essentially created is heartbeat monitor for the road.

Roughness is expressed in IRI, or International Roughness Index, which is the metric for measuring road quality. In addition to ride quality, it also contributes to fuel consumption. 

“If you are interested in a road’s roughness, the app can display IRI; if you are interested in the fuel consumption the app can show you if, say, 20 to 25 percent of your fuel consumption comes from a poor-quality road — which is totally possible in a city,” Roxon says.

One issue arises, though, when measuring the roughness of the ride: How can one determine what readings are from the road, the car, or even the occupants? According to Roxon, only time can tell.

“For us to gather any information about the vehicle, we need at the very minimum three minutes of data,” he says. “If you think about how often the accelerometer takes measurements, that can give you up to 18,000 data points.”

With this robust dataset, Roxon and his team say they can cut out the “noise,” or the extraneous data from the car or its cargo.

The next step for Roxon and his team is to use the data to determine excess fuel consumption and pavement quality in real-time. Traditional methods that calculate pavement data en masse would completely swamp a phone’s processing ability, so instead, Roxon determines fuel consumption on a step by step basis. By selecting data points, for example, every minute and comparing them to the prior minute, he can calculate the difference in pavement quality, and therefore a pavement’s roughness.

Eventually, by leveraging crowdsourcing, he hopes to overlay this roughness data onto road maps and then use machine learning algorithms like those of Apple Maps or Google maps to route drivers on the path of least resistance. As one might expect, this feature could prove highly useful for fleet drivers who drive regularly.

“On average, a semi-truck in the U.S. consumes around 20,000 gallons of diesel per year and the average diesel price in 2018 was $3.19. If you multiply one by the other, you are looking at $63,600 per year per year in fuel costs,” says Roxon, “Now, with our app, we could safely identify 5 percent to 10 percent savings for these drivers. That becomes over $6,000 dollars for just one truck.”

With some fleets in the U.S. totaling in the thousands, even tens of thousands, this could save companies large sums of money.

Truck drivers may also find Carbin useful for a different reason. It also has the potential to simply and accurately monitor tire pressure.

Here’s how it works: When a tire’s pressure is low, a car uses more energy to turn it; this can have a slight influence on the motion of the car. Carbin allows a smartphone’s accelerometer to read this often-unnoticeable motion and determine which tire needs inflating and by how much.

Since improperly inflated tires increase expenses and, at worst, can lead to dangerous events like blowouts, this could prove valuable to truckers. This is particularly the case since many trucks lack tire pressure sensors — which are both expensive and often unreliable.

State and city agencies may also find a use for the app. While federal agencies have the funds to monitor the quality of their pavements using sophisticated technology, the method is expensive for state agencies and beyond the budget of most cities. With limited funding, states must survey their pavements in sections over several years which prevents accurate monitoring of the performance of the entire road network. Cities have even less resources, so they tend to rely on reporting by citizens and visual examination by inspectors to determine road maintenance. The Carbin app would allow cities and states to gain a better understanding of their pavement quality at a far more reasonable price.

When its features roll out later this month Carbin will offer users a suite of options to save money, track excess emissions, and understand the carbon footprint of their infrastructure. 

de MIT News https://ift.tt/2H7gy6r

Yazmin Guzman named 2019 Gates Cambridge Scholar

Yazmin Guzman, a master's student in the Department of Urban Studies and Planning, has won the prestigious Gates Cambridge scholarship, which offers students an opportunity to pursue graduate study in the field of their choice at Cambridge University. 

Guzman, from Wichita, Kansas, recently completed her bachelor's in DUSP with a minor in economics, and this spring will complete a Master in City Planning. At Cambridge University she plans on reading for a degree in educational policy. Guzman aims for a career in educational reform with a focus on increasing educational opportunities for low-income students and improving policies related to community colleges and vocational training. 

The daughter of immigrants from Oaxaca, Mexico, Guzman is the first person in her family to pursue a college degree. At MIT, she has researched deindustrialization, immigration law, and residential mobility, and conducted interviews with Latino residents in Lawrence, Massachusetts. Justin Steil, the Charles H. and Ann E. Spaulding Career Development Assistant Professor in the Department of Urban Studies and Planning, has advised and mentored Guzman throughout her time at MIT.

“Yazmin is a problem solver at heart and is constantly inspiring and building bridges with others," Steil says. "A first-generation college student, Yazmin is passionate about improving access to high-quality education for all. Over her years at MIT, Yazmin discovered a love of statistical analysis and enthusiasm for applying those skills to meaningful public problems, such as educational access and quality. It has been an honor to have Yazmin as a student and an advisee and to work with her on research.”

Guzman has made strategic use of MIT’s many resources for community engagement and experiential learning to prepare herself for a career addressing inequalities. With support from the PKG Center, she spent an Independent Activities Period working with the KIPP school in Washington Heights, New York, where she developed community-building resources and supported classroom teachers, while seeking to understand how education policy affects the classroom experience of teachers and pupils. The following summer, she investigated how education equality initiatives use their data to improve services, as a research initiative intern at Questbridge, an organization that helps students from low-income backgrounds access leading institutions of higher education. Back at MIT, Guzman volunteered as a tutor and mentor for Amphibious Achievement, a dual athletic-academic program for underprivileged students in the Boston area, and directed the organization’s academic programming. She also tutored in mathematics through MIT SHINE for Girls, a mentorship program for middle school girls that combines dance and mathematics.

Alison Hynd, assistant dean at the PKG Center says, “Yazmin brings to her work a drive to make real change in the world, combined with a focus on deeply understanding the problems she wants to address. She’s an extraordinary woman and a future leader in education equity!"

Guzman has been president of La Union Chicana Por Aztlan (LUChA) and is the co-founder of Hermanas Unidas, an initiative that seeks to empower Latinas across MIT’s campus. She has held leadership positions with MIT organizations that mentor local high school and middle school students, including Amphibious Achievement and SHINE. She has been a student representative for the Committee on Academic Performance, vice president of educational outreach for Latinos in Engineering and Science, and a tutor at correctional facilities through the Petey Greene program. She is also a member of the Sakata Afrique dance team. 

Guzman was advised in her application by Kim Benard in the Office of Distinguished Fellowships. Established by the Bill and Melinda Gates Foundation in 2000, the Gates Cambridge Scholarship provides full funding for talented students from outside the United Kingdom to pursue postgraduate study in any subject at Cambridge University. The 2019 awards process was extremely competitive, with 34 ultimately chosen. Since the program’s inception in 2001, there have been 28 Gates Cambridge Scholars from MIT.

de MIT News https://ift.tt/2StBRBe

National competition veterans pay it forward at regional science contest

On Saturday, Feb. 23, more than 100 middle school students gathered at MIT to compete in the annual Northeast Regional Middle School Science Bowl.

The event, now in its fourth year, was coordinated and executed by Kathleen Schwind, a senior in the five-year program in the MIT Department of Urban Studies and Planning, and Songela Chen, a senior in the MIT Department of Chemistry. Many of the organizers and volunteers, including Schwind and Chen, are veterans of middle and high school National Science Bowl (NSB) competitions.

“The Northeast Middle School Science Bowl really is something special,” says Schwind. “[It’s] an event run by young people for young people, and an opportunity to not only celebrate the youth in our community, but also inspire them to continue being a part of NSB and to give back to other young people one day, too.”

The first several rounds were a round-robin style warmup for the 21 teams of four or five middle school students representing 10 schools from Maine, New Hampshire, and Massachusetts. Correctly answered questions in fields such as life science, physical science, earth and space science, and math won a team points and the chance at a bonus question. An incorrect answer passed the question over to the other team, who could then attempt an answer.

After a lunch break and group photo, elimination rounds began. Those knocked out switched their attention to fun engineering challenges such as building a tower out of supplied paper bags, aluminum foil, cups, and straws. At the end of the day, eliminated participants watched the tight race for third place, followed by a championship round for the title.

This year, that title went to Jonas Clarke Middle School Team One from Lexington, Massachusetts. William Diamond Middle School Team One, also from Lexington, took second. The winning team received a coveted trophy and the opportunity to represent the Northeast in the National Science Bowl hosted by the U.S. Department of Energy in Washington in April.

However, the competition was not about winning, said several participants, all of whom wore matching green shirts stating, “Keep calm and science bowl on.” Instead, it was about the fun and comradery of being part of a team. “It gives you purpose,” said a seventh-grade student from William Diamond Middle School. Being on MIT's campus was an opportunity to interact with an even larger scientific community. “It’s fun and confusing and kind of scary,” said an eighth grader from Jonas Clarke, who wants to be a marine biologist. “Scary because of the number of people and how big MIT is,” she clarified with a laugh.

As a student at MIT, Schwind founded the Northeast Regional Middle School Science Bowl when she learned the region lacked a local chapter. She used experience gained from founding and coordinating such events since age 16 — the youngest coordinator to date.

“The science bowl is extremely valuable for promoting science and the broader appreciation of science, so I am delighted to continue my support through MIT’s School of Science for this year’s event,” School of Science Dean Michael Sipser says of his recurring interest in sponsorship of the event.

Schwind also recruited the help of fellow National Science Bowl alumni, such as Chen. Although both will graduate in the spring, Schwind and Chen plan to continue running this event next year, remotely if necessary.

As a seven-year alumna of middle and high school science bowls, Chen says it was a motivator for her career in science and she hopes to pay it forward, “to show middle school students how valuable and rewarding science can be.”

“There is nothing like seeing a competitor have that sparkle in their eyes after the event and tell you that they now want to be a scientist or mathematician and go to MIT one day,” Schwind says.

de MIT News https://ift.tt/2Eiec1y

The future of education in a world of pervasive computing

Computers have become so pervasive in today’s world that preparing students to work and assume leadership roles in this shifting landscape requires giving them a better understanding of how computers work, how to use them, and how they affect every aspect of society. That’s the reasoning behind the creation of the new MIT Stephen A. Schwarzman College of Computing, and it was the theme of many of the presentations and panel discussions in this week’s three-day celebration of the new college.

“We’re in the midst of a global transformation that’s catalyzed by the rapid acceleration of digital technologies, including unprecedented access to computation and data,” said Farnam Jahanian, president of Carnegie Mellon University, in a keynote address on Wednesday. “The scale and scope and pace of these advances are truly unprecedented in human history.”

“The impact of these technologies is ubiquitous,” he said, “with a wide range of applications from health care to transportation, finance, energy, manufacturing, and far beyond. … The pace of innovation is accelerating dramatically.”

These changes require a profound rethinking of the role of education in this rapidly changing environment, Jahanian said. “Imagine a day when by integrating emerging technologies, such as AI-enabled learning techniques and inverted classrooms, we can achieve personalized, outcome-based education,” he said.

MIT Provost Martin Schmidt, in a discussion with reporters, said that in creating the MIT Schwarzman College of Computing, “one of the things that’s really critical to us is that not only should this advance computation, but it should really link to all the disciplines across the campus.” The college will “strengthen those disciplines in their use of these new tools,” he said, “but also when we learn things about how we apply those tools to the disciplines, that knowledge flows back … and informs the next generation” of computing research.

Schmidt added that in planning the new college, a key question was how MIT will deliver on its promise of making sure that the college “has in its DNA” an awareness of the societal impact of current and future advances in computing. This appreciation “should inform our educational agenda, what our undergraduates and graduates learn in the classroom, and it should inform our research agenda,” he said. “It should shape how the research is performed, and the kind of content we produce that informs policies and informs governments on how they should respond to the deployment of these technologies.”

The new college was founded partly in response to the fact that “there really was a transformation occurring across the campus,” with computation increasingly forming a key part of the work in amost all disciplines, Schmidt said. While about 40 percent of MIT students major in computer science, there was a clear need for an even greater integration of computation and data science early and deeply into every aspect of education.

Melissa Nobles, dean of MIT’s School of Humanities, Arts and Social Sciences, who also participated in the discussion, told reporters that students in that school were very excited to take part in this increased integration of their disciplines with computation. She cited examples of classes where mixed groups of computer science students and those majoring in arts, economics, or literature worked on problems that combined their different kinds of expertise. In one class, for example, the students studied in exhaustive detail the way writers of 19th century novels used male and female pronouns and how that related to the genders of the author and the main characters. The project required both computer expertise to analyze thousands of texts, and a knowledge of the literature in order to provide context for their findings.

Also during the discussion, Maria Klawe, president of Harvey Mudd College in California and another keynote speaker, pointed out that a deep understanding of computers and their impact is increasingly needed in a rapidly changing world where it is estimated that many of the jobs people perform today “are just going to disappear” within the next few decades. That makes interdisciplinary education more important than ever, she said.

Regarding the creation of the new college, she said, “I see this as an incredibly important step for MIT, and I think it’s going to influence other institutions to do similar things.”

The goals of the college reach far beyond just helping people in other disciplines to use computers more effectively, Nobles and others emphasized. It’s also important, they said, to make sure that the skills and knowledge from other fields flow back into computer science, influencing the ethical, political, and social implications of the work in that field — not just as an afterthought but as a fundamental part of thinking and planning.

For example, while it is tempting to make use of massive sets of data collected by social media, the use of such datasets can raise serious concerns about privacy and informed consent. Such issues may be relatively new territory for computer scientists, but they are longstanding issues that have been dealt with extensively by social scientists and philosophers whose expertise can help inform the data collection and analysis procedures.

The speakers at Wednesday’s symposium, representing many different fields and institutions, shared a sense of excitement about the potential for the MIT Schwarzman College of Computing to bring about significant innovations. “MIT continues to be a world-class institution that offers a distinctive education and research, of course,” Jahanian said in his keynote, “and this latest development will certainly increase its impact in this changing world.”

de MIT News https://ift.tt/2ED5EDT

Making family business our business

Did you know that more than half of publicly traded companies in the United States are family companies? This is low compared to other stock exchanges around the world, like Mexico and the Philippines, where family-controlled companies dominate the listed businesses. In fact, family-owned firms account for two-thirds of all businesses around the world, with matching influence on global GDP and job creation. The Family Capital top 750 ranking of the world’s largest family businesses illustrates just how large these companies can get and suggests how much they contribute to the world economy. The companies on this list, including household names like Walmart, BMW, Dell Technologies, and LG Electronics, have combined revenues of more than $9 trillion and directly employ around 30 million people.

Although the ownership, governance, management, and leadership of family firms is critical to the economic success of nations, they are less talked about than anonymously-owned public corporations like IBM, American Express, or ExxonMobil. They also face a unique set of challenges over the long term.

“Family companies perform significantly better in terms of sustainability, profitability, and growth of all kinds when compared to non-family businesses,” says John Davis, a globally recognized authority and pioneer in the field of family enterprise. “They are even shown to be more innovative than non-family companies. But despite their success, many family-owned businesses face serious challenges with sustaining their success over multiple generations.” Only about 30 percent of family-owned businesses survive into the second generation, according to the Family Business Institute. Twelve percent are still viable into the third generation, and only about 3 percent of all family businesses operate into the fourth generation or beyond.

John Davis is one of the best-known voices in the world of family enterprise and family office. A strategic advisor and professor on the topic for decades, Davis has now joined MIT Sloan School of Management as a senior lecturer and is behind a series of new Sloan Executive Education programs designed to help family owners achieve multigenerational success and prepare for the future.

These programs can’t come soon enough, as digitalization, innovation, and technology have companies of all types on edge. For family-owned enterprises with a long tradition of doing things a certain way, the breakneck speed of technological change and industry disruptions can feel especially disorienting. Disruption and change are shortening time horizons for all businesses, but family businesses have extra layers of complexity — the future success of the business is inseparable from the well-being of the family — and all the more reason to be better educated about the future.

“I have always said that family businesses are slower but better,” says Davis. “But these days, can you still be slower and better? Where MIT really has done a lot of work and can add a lot to the thinking around family enterprise is the area of technological change and disruption and how it’s influencing industries, business models, and the way work is done. There is no faculty out there that is stronger in these ways than my colleagues at MIT.”

Davis’s new executive education programs help families manage disruption, think about where their industries are going, and get ahead of change — all while managing the evergreen issues that families need to stay on top of, such as preparing the next generation for skills they’ll need in the future, having the right governance, and keeping family relationships strong.

• Leading and Transforming Family Businesses - China guides leaders of Chinese family enterprises through a three-week journey on the campuses of MIT, Oxford University, and Peking University to develop sound growth strategies, strengthen their organizations, and build a firm foundation for long-term success.
• Future Family Enterprise: Sustaining Multigenerational Success leads multigenerational families through a stimulating week-long conversation that produces clarity on the path ahead for participating families and their enterprises.
• Founder to Family: Sustaining Family Business Success helps founders, or first-generation families in business, build a bridge to the second generation.

In all three of these courses, participants engage in interactive classroom work and exchanges among families from around the world. In the two Cambridge, Massachusetts, programs, families attend as teams and have daily private, facilitated discussions with an experienced family-enterprise coach to focus on their interests and agenda. And each family team leaves the program with a tailored action plan built together over the duration of the course.

“I’ve been doing this for 40 years, and my work now is about the future of family enterprise,” says Davis. “Where business is going, where family offices are going, how ownership is changing, how capital is being raised, and how families are changing. All of these trends influence the nature of family enterprise, and I am mapping these changes and thinking about how this influences not just the management of a company, but the role of owners of these companies.”

The dynamic global family business ecosystem is set to grow even more important and influential in the years ahead. By virtue of this ecosystem’s impact on the economic health of countries and the well-being of their citizens, these new programs are well aligned with MIT Sloan’s mission to make a difference in the world.

de MIT News https://ift.tt/2NBejJD

Taking the lead in shaping the future of computing and artificial intelligence

With a box of popcorn in one hand, Hal Abelson, a renowned computer scientist, strolled through the first floor of the Ray and Maria Stata Center studying the machine learning exhibits that surrounded him on the afternoon of Feb. 26. Everywhere he looked he saw evidence of the remarkable things MIT students can do when given access to computing resources.

“Computing tools and infrastructure have gotten to a place where students can outperform professional researchers. You are constrained mostly by your imagination. It’s just an amazing time,” said Abelson, the Class of 1922 Professor of Computer Science and Engineering.

Abelson, and a crowd of hundreds, was witnessing the kickoff of a three-day celebration of the MIT Stephen A. Schwarzman College of Computing. The afternoon event was an exposition of projects that transformed the student street lobby area of the Stata Center into a computing fairground of sorts, replete with courtesy popcorn, bubble tea, lemon squares, brownies, celebratory stickers, and a host of student exhibits that crossed disciplines, broke barriers, and inspired new thinking.

For Kadeem Khan, a graduate student in urban studies and planning and an expo participant, the day was special. “I wanted to do a project focused on machine learning and the developing world,” he said. Khan applied machine learning to generate useful insights on poverty in Nairobi by analyzing data from multiple sources, including census, satellite imagery, and data from a geographic information system.

“The poverty exhibit is an example of what I was just saying,” said Abelson. “Somehow the resources are here now to allow students to bring things to the next level.” Abeslson and Nicholas Roy, a professor of aeronautics and astronautics, CSAIL researcher, and director of the Bridge in the Quest for Intelligence, helped judge the teams during the monthlong student computing challenges leading up to yesterday.

Like Khan, MIT electrical engineering and computer science graduate student Natalie Lao embarked on a winning project with the potential to make transformative change in the world. “My background is in AI — but I'm also very interested in ethics and fairness and the risks involved when applying AI to the real world,” she said. Her team’s project uses network propagation and analysis to automatically discover and potentially halt the spread of fake news across a variety of media platforms. “We’re talking to the Department of Defense and various companies and trying to see how we can get the solution out in the world,” she said.

The MIT Schwarzman College of Computing, which represents a $1 billion commitment to addressing the global opportunities and challenges presented by the prevalence of computing and the rise of artificial intelligence, will provide students with unprecedented computing resources, including access to large data sets and the tools to learn from them. Yesterday, top entrants spoke in excited tones about the data sets they accessed during the Machine Learning Across Disciplines Challenge, which, along with the Connect Arts, Community, and Computing Challenge, was funded by the MIT-IBM Watson AI Lab.

Graduate students Agni Orfanoudaki and Antonin Dauvin, who are both studying operations research at the MIT Sloan School of Management, applied machine learning and techniques developed at MIT Operations Research Center to patient data from Boston Medical Center spanning two decades. They are developing an analytic approach to understanding the impact of different anti-hypertensive drugs.

Senior Sarah Wooders, an undergraduate in math and computer science, has collected a dataset of over 4 million product images and descriptions scraped from online sources. She then trained models to collectively label over 90 important clothing attributes and is is now building a system that can automatically label new clothing products. “It’s really exciting to see all the applications of AI,” said Wooders, also a top entrant. “My project feels like such an obvious idea but this type of system hasn’t been created yet. It seems the same thing is true for a lot of things in AI right now. And so someone like me can come along and do it.”

de MIT News https://ift.tt/2IIs0Yz

Engineers make clear droplets produce iridescent colors

Engineers at MIT and Penn State University have found that under the right conditions, ordinary clear water droplets on a transparent surface can produce brilliant colors, without the addition of inks or dyes.

In a paper published today in Nature, the team reports that a surface covered in a fine mist of transparent droplets and lit with a single lamp should produce a bright color if each tiny droplet is precisely the same size.

This iridescent effect is due to “structural color,” by which an object generates color simply due to the way light interacts with its geometric structure. The effect may explain certain iridescent phenomena, such as the colorful condensation on a plastic dish or inside a water bottle.

The researchers have developed a model that predicts the color a droplet will produce, given specific structural and optical conditions. The model could be used as a design guide to produce, for example, droplet-based litmus tests, or color-changing powders and inks in makeup products.

“Synthetic dyes used in consumer products to create bright colors might not be as healthy as they should be,” says Mathias Kolle, assistant professor of mechanical engineering at MIT. “As some of these dyes are more strongly regulated, companies are asking, can we use structural colors to replace potentially unhealthy dyes? Thanks to the careful observations by Amy Goodling and Lauren Zarzar at Penn State and to Sara’s modeling, which brought this effect and its physical explanation to light, there might be an answer.”

Sara Nagelberg of MIT, along with lead author Goodling, Zarzar, and others from Penn State, are Kolle’s co-authors on the paper.

Follow the rainbow

Last year, Zarzar and Goodling were studying transparent droplet emulsions made from a mixture of oils of different density. They were observing the droplets’ interactions in a clear Petri dish, when they noticed the drops appeared surprisingly blue. They took a photo and sent it off to Kolle with a question: Why is there color here?

Initially, Kolle thought the color might be due to the effect that causes rainbows, in which sunlight is redirected by rain drops and individual colors are separated into different directions. In physics, Mie scattering theory is used to describe the way spheres such as raindrops scatter a plane of electromagnetic waves, such as incoming sunlight. But the droplets that Zarzar and Goodling observed were not spheres, but rather, hemispheres or domes on a flat surface.

“Initially we followed this rainbow-causing effect,” says Nagelberg, who headed up the modeling effort to try to explain the effect. “But it turned out to be something quite different.”

She noted that the team’s hemispherical droplets broke symmetry, meaning they were not perfect spheres — a seemingly obvious fact but nevertheless an important one, as it meant that light should behave differently in hemispheres versus spheres. Specifically, the concave surface of a hemisphere allows an optical effect that is not possible in perfect spheres: total internal reflection, or TIR.

Total internal reflection is a phenomenon in which light strikes an interface between a high refractive index medium (water, for instance) to a lower refractive index medium (such as air) at a high angle such that 100 percent of that light is reflected. This is the effect that allows optical fibers to carry light for kilometers with low loss. When light enters a single droplet, it is reflected by TIR along its concave interface.

In fact, once light makes its way into a droplet, Nagelberg found that it can take different paths, bouncing two, three, or more times before exiting at another angle. The way light rays add up as they exit determines whether a droplet will produce color or not.

For example, two rays of white light, containing all visible wavelengths of light, entering at the same angle and exiting at the same angle, could take entirely different paths within a droplet. If one ray bounces three times, it has a longer path than a ray that bounces twice, so that it lags behind slightly before exiting the droplet. If this phase lag results in the two rays’ waves being in phase (meaning the waves’ troughs and crests are aligned), the color corresponding to that wavelength will be visible. This interference effect, which ultimately produces color in otherwise clear droplets, is much stronger in small rather than large droplets.

“When there is interference, it’s like kids making waves in a pool,” Kolle says. “If they do whatever they want, there’s no constructive adding up of effort, and just a lot of mess in the pool, or random wave patterns. But if they all push and pull together, you get a big wave. It’s the same here: If you get waves in phase coming out, you get more intensity of color.”

A carpet of color

The color that droplets produce also depends on structural conditions, such as the size and curvature of the droplets, along with the droplet’s refractive indices.

Nagelberg incorporated all these parameters into a mathematical model to predict the colors that droplets would produce under certain structural and optical conditions. Zarzar and Goodling then tested the model’s predictions against actual droplets they produced in the lab.

First, the team optimized their initial experiment, creating droplet emulsions, the sizes of which they could precisely control using a microfluidic device. They produced, as Kolle describes, a “carpet” of droplets of the exact same size, in a clear Petri dish, which they illuminated with a single, fixed white light. They then recorded the droplets with a camera that circled around the dish, and observed that the droplets exhibited brilliant colors that shifted as the camera circled around. This  demonstrated how the angle at which light is seen to enter the droplet affects the droplet’s color.

The team also produced droplets of various sizes on a single film and observed that from a single viewing direction, the color would shift redder as the droplet size increased, and then would loop back to blue and cycle through again. This makes sense according to the model, as larger droplets would give light more room to bounce, creating longer paths and larger phase lags.

To demonstrate the importance of curvature in a droplet’s color, the team produced water condensation on a transparent film that was treated with a hydrophobic (water-repelling) solution, with the droplets forming the shape of an elephant. The hydrophobic parts created more concave droplets, whereas the rest of the film created shallower droplets. Light could more easily bounce around in the concave droplets, compared to the shallow droplets. The result was a very colorful elephant pattern against a black background.

In addition to liquid droplets, the researchers 3-D-printed tiny, solid caps and domes from various transparent, polymer-based materials, and observed a similar colorful effect in these solid particles, that could be predicted by the team’s model.

Kolle expects that the model may be used to design droplets and particles for an array of color-changing applications.

“There’s a complex parameter space you can play with,” Kolle says. “You can tailor a droplet’s size, morphology, and observation conditions to create the color you want.”

This research was supported, in part, by the National Science Foundation and the U. S. Army Research Office through the Institute for Soldier Nanotechnologies at MIT.

de MIT News https://ift.tt/2SvF4jJ

QS World University Rankings rates MIT No. 1 in 11 subjects for 2019

MIT has been honored with 11 No. 1 subject rankings in the QS World University Rankings for 2019.

The Institute received a No. 1 ranking in the following QS subject areas: Chemistry; Computer Science and Information Systems; Chemical Engineering; Civil and Structural Engineering; Electrical and Electronic Engineering; Mechanical, Aeronautical and Manufacturing Engineering; Linguistics; Materials Science; Mathematics; Physics and Astronomy; and Statistics and Operational Research.

MIT also placed second in six subject areas: Accounting and Finance; Architecture/Built Environment; Biological Sciences; Earth and Marine Sciences; Economics and Econometrics; and Environmental Sciences.

Quacquarelli Symonds Limited subject rankings, published annually, are designed to help prospective students find the leading schools in their field of interest. Rankings are based on research quality and accomplishments, academic reputation, and graduate employment.

MIT has been ranked as the No. 1 university in the world by QS World University Rankings for seven straight years.

de MIT News https://ift.tt/2U9c3fp

Mathematics Professor Bonnie Berger honored with ISCB Senior Scientist Award

Bonnie Berger, the Simons Professor of Mathematics at MIT, has been selected as the 2019 recipient of the International Society for Computational Biology Senior Scientist Award.

The annual award recognizes “highly significant, long-term career achievement,” in Berger’s case for visionary, foundational, and deep contributions to the field. ISCB is the premier society in computational biology and bioinformatics with 3,400 members. 

“It’s a tremendous honor to join such a distinguished and accomplished group of scientists,” said Berger, who holds a joint appointment in the Department of Electrical Engineering and Computer Science.

“Dr. Berger made fundamental contributions in diverse areas of bioinformatics, starting from important contributions in protein structure prediction in 1990s to founding the area of compressive genomics a few years ago,” said her nominator, Pavel Pevzner, a professor of computer science and engineering at the University of California at San Diego. “Her work combines algorithmic depth, biological relevance, practical utility, and broad applicability.” 

Berger will receive the award and give the Senior Scientist keynote at the ISMB/ECCB 2019 meeting this July in Basel, Switzerland.

Berger was also named an ISCB Distinguished Fellow in 2012.

de MIT News https://ift.tt/2BUDx1b

Supercomputers can spot cyber threats

Identifying cybersecurity threats from raw internet data can be like locating a needle in a haystack. The amount of internet traffic data generated in a 48-hour period, for example, is too massive for one or even 100 laptops to process into something digestible for human analysts. That's why analysts rely on sampling to search for potential threats, selecting small segments of data to look at in depth, hoping to find suspicious behavior.

While this type of sampling may work for some tasks, such as identifying popular IP addresses, it is inadequate for finding subtler threatening trends.

"If you're trying to detect anomalous behavior, by definition that behavior is rare and unlikely," says Vijay Gadepally, a senior staff member at the Lincoln Laboratory Supercomputing Center (LLSC). "If you're sampling, it makes an already rare thing nearly impossible to find."

Gadepally is part of a research team at the laboratory that believes supercomputing can offer a better method — one that grants analysts access to all pertinent data at once — for identifying these subtle trends. In a recently published paper, the team successfully condensed 96 hours of raw, 1-gigabit network link internet traffic data into a query-ready bundle. They created the bundle by running 30,000 cores of processing (equal to about 1,000 laptops) at the LLSC located in Holyoke, Massachusetts, and it is stored in the MIT SuperCloud, where it can be accessed by anyone with an account.

"[Our research] showed that we could leverage supercomputing resources to bring in a massive quantity of data and put it in a position where a cybersecurity researcher can make use of it," Gadepally explains.

An example of the type of threatening activity that requires analysts to dig in to such a massive amount of data are instructions from command-and-control (C&C) servers. These servers issue commands to devices infected with malware in order to steal or manipulate data.

Gadepally likens their pattern of behavior to that of spam phone callers: While a normal caller might make and receive an equal number of calls, a spammer would make millions more calls than they receive. It's the same idea for a C&C server, and this pattern can be found only by looking at lots of data over a long period of time.

"The current industry standard is to use small windows of data, where you toss out 99.99 percent," Gadepally says. "We were able to keep 100 percent of the data for this analysis."

The team plans to spread the word about their ability to compress such a large quantity of data and they hope analysts will take advantage of this resource to take the next step in cracking down on threats that have so far been elusive. They are also working on ways to better understand what "normal" internet behavior looks like as a whole, so that threats can be more easily identified.

"Detecting cyber threats can be greatly enhanced by having an accurate model of normal background network traffic," says Jeremy Kepner, a Lincoln Laboratory fellow at the LLSC who is spearheading this new research. Analysts could compare the internet traffic data they are investigating with these models to bring anomalous behavior to the surface more readily.

"Using our processing pipeline, we are able to develop new techniques for computing these background models," he says.

As government, business, and personal users increasingly rely on the internet for their daily operations, maintaining cybersecurity will remain an essential task for researchers and the researchers say supercomputing is an untapped resource that can help.

de MIT News https://ift.tt/2Eguvf4

What 126 studies say about education technology

In recent years, there has been widespread excitement around the transformative potential of technology in education. In the United States alone, spending on education technology has now exceeded $13 billion. Programs and policies to promote the use of education technology may expand access to quality education, support students’ learning in innovative ways, and help families navigate complex school systems.

However, the rapid development of education technology in the United States is occurring in a context of deep and persistent inequality. Depending on how programs are designed, how they are used, and who can access them, education technologies could alleviate or aggravate existing disparities. To harness education technology’s full potential, education decision-makers, product developers, and funders need to understand the ways in which technology can help — or in some cases hurt — student learning.

To address this need, J-PAL North America recently released a new publication summarizing 126 rigorous evaluations of different uses of education technology. Drawing primarily from research in developed countries, the publication looks at randomized evaluations and regression discontinuity designs across four broad categories: (1) access to technology, (2) computer-assisted learning or educational software, (3) technology-enabled nudges in education, and (4) online learning.

This growing body of evidence suggests some areas of promise and points to four key lessons on education technology.

First, supplying computers and internet alone generally do not improve students’ academic outcomes from kindergarten to 12th grade, but do increase computer usage and improve computer proficiency. Disparities in access to information and communication technologies can exacerbate existing educational inequalities. Students without access at school or at home may struggle to complete web-based assignments and may have a hard time developing digital literacy skills.

Broadly, programs to expand access to technology have been effective at increasing use of computers and improving computer skills. However, computer distribution and internet subsidy programs generally did not improve grades and test scores and in some cases led to adverse impacts on academic achievement. The limited rigorous evidence suggests that distributing computers may have a more direct impact on learning outcomes at the postsecondary level.

Second, educational software (often called “computer-assisted learning”) programs designed to help students develop particular skills have shown enormous promise in improving learning outcomes, particularly in math. Targeting instruction to meet students’ learning levels has been found to be effective in improving student learning, but large class sizes with a wide range of learning levels can make it hard for teachers to personalize instruction. Software has the potential to overcome traditional classroom constraints by customizing activities for each student. Educational software programs range from light-touch homework support tools to more intensive interventions that re-orient the classroom around the use of software.

Most educational software that have been rigorously evaluated help students practice particular skills through personalized tutoring approaches. Computer-assisted learning programs have shown enormous promise in improving academic achievement, especially in math. Of all 30 studies of computer-assisted learning programs, 20 reported statistically significant positive effects, 15 of which were focused on improving math outcomes.

Third, technology-based nudges — such as text message reminders — can have meaningful, if modest, impacts on a variety of education-related outcomes, often at extremely low costs. Low-cost interventions like text message reminders can successfully support students and families at each stage of schooling. Text messages with reminders, tips, goal-setting tools, and encouragement can increase parental engagement in learning activities, such as reading with their elementary-aged children.

Middle and high schools, meanwhile, can help parents support their children by providing families with information about how well their children are doing in school. Colleges can increase application and enrollment rates by leveraging technology to suggest specific action items, streamline financial aid procedures, and/or provide personalized support to high school students.

Finally, relative to courses with some degree of face-to-face teaching, students taking online-only courses may experience negative learning outcomes.

Online courses are developing a growing presence in education, but the limited experimental evidence suggests that online-only courses lower student academic achievement compared to in-person courses. In four of six studies that directly compared the impact of taking a course online versus in-person only, student performance was lower in the online courses. However, students performed similarly in courses with both in-person and online components compared to traditional face-to-face classes.

The new publication is meant to be a resource for decision-makers interested in learning which uses of education technology go beyond the hype to truly help students learn. At the same time, the publication outlines key open questions about the impacts of education technology, including questions relating to the long-term impacts of education technology and the impacts of education technology on different types of learners.

To help answer these questions, J-PAL North America’s Education, Technology, and Opportunity Initiative is working to build the evidence base on promising uses of education technology by partnering directly with education leaders.

Education leaders are invited to submit letters of interest to partner with J-PAL North America through its Innovation Competition. Anyone interested in learning more about how to apply is encouraged to contact initiative manager Vincent Quan.

de MIT News https://ift.tt/2TcZouE

Student challenges kick off celebration of MIT Stephen A. Schwarzman College of Computing

When MIT graduate student Matthew Claudel learned of the student computing challenges launched to accompany the three-day celebration of the MIT Stephen A. Schwarzman College of Computing that begins today, he eagerly signed on.

“We were excited about extending the definition of computation and exploring how it might relate to the arts,” says Claudel, who is studying in the Department of Urban Studies and Planning.

The Computing Connections Challenges involved monthlong challenges with themes such as connecting arts, community, and computing, or using machine learning to explore cross-disciplinary topics including health care, transportation, privacy, ethics, architecture, design, commerce, finance, poverty, neuroscience, linguistics, and more. Other challenges involved the application of computing to finance and to the world of sports.

Participants came from all five of the Institute’s schools, and a Computing Exposition today will include top entries from the challenges, as well as interactive demonstrations that illustrate the ways in which MIT is advancing computing across disciplines.

Celebrations of the new college began over the weekend with a free film series on computing hosted by the MIT Lecture Series Committee, which has been bringing entertainment to campus since 1944. The events continue Feb. 26-28, with three days of programming with the themes of “explore,” “teach,” and “celebrate.”

For the “Connect Arts, Community, and Computing” challenge, Claudel and Kimberly Smith SM ’17, who studied media arts and sciences, created an augmented reality mobile application called New Gravity, which paints the sphere atop the Green Building with a color scheme derived from the mathematics of gravitation and includes graphics that demonstrate gravitational waves.

“Neither of us are computer scientists, but both of us have collaborated extensively with computer scientists. We see an incredible future where the College of Computing channels the collaborative potential of disciplines such as arts, urban science, education, or astrophysics,” says Claudel.

“New Gravity is a thoroughly MIT kind of monument. It’s beautiful and it’s a bit quirky. It merges art and science, digital and physical. It’s shared among us, the MIT community, and with the city around us. New Gravity invites us all to stand, for a moment in awe as the distance to the Green Building collapses into a phone and expands to 1.3 billion light years,” he says.

Over the past several weeks, MIT students have dedicated their free time to such computing challenges, including the Connect Arts, Community, and Computing Challenge and the Machine Learning Across Disciplines Challenge, which were both funded by the MIT-IBM Watson AI Lab. Host partners for the challenges include the Arts at MIT, Computer Science and Artificial Intelligence Laboratory, Martin Trust Center, School of Engineering, the MIT Quest for Intelligence, and the Undergraduate Association. Many of the student creations are on display today in Memorial Lobby (Building 10) and at the Ray and Maria Stata Center and the Koch Institute.

MIT graduate student Tianyu Su from the Department of Urban Studies and Planning says his team of student researchers and designers from MIT and Harvard University joined the arts and community challenge to learn from each other and from other teams.

“We all have a strong interest in how we could better interact with the emerging technologies and how to apply them into underrepresented communities,” says Su, whose team, Eco-LIVE, applied emerging technology — including augmented reality and computational modeling — to design ecological education that integrates playfulness, cohesiveness, and accessibility in community education.

“We are excited. We can see so many new possibilities in the new college and look forward to the coming activities,” he says.

A workshop in early February reflected the student excitement around the mural augmentation project that was one option within the “Connect Arts, Community, and Computing” challenge. During the workshop, MIT junior Jierui Fang, an art and design major who is minoring in computer science, helped coach student participants.

“The College of Computing isn’t just about computation,” Fang said at the time. “I think it’s really important that people are aware that MIT students are very multifaceted. We have different aspects to our personalities and different passions.”

Fang is a member of the Borderline Student Group, which together with more than 20 artists created a connected, augmented mural with animations in the tunnels under campus. During the challenge, she and other Borderline members offered other students help with augmenting images.

The multifaceted nature of MIT students was clear from the playful top entries, which can be viewed by holding a mobile device in front of murals in the Stata Center while running an application called Artivive that overlays the student animations onto the canvas.

First-year student Heya Lee created a digital animation of a butterfly in flight against the backdrop of an autumn day. Natasha Hirt, also a first-year student, developed animation that swirls and blurs across a mural featuring an image of the Stata Center. Second-year architecture students Jacqueline Chen, Peiling Jiang, and Yi Yang joined with second-year mathematics student Wanyi Xiao to develop an animation of a creature climbing a tree in a mural developed in a collaboration with MIT and the Suffolk County House of Corrections. 

de MIT News https://ift.tt/2Vjceoz

Plasma Science and Fusion Center leads new Center of Excellence

The High-Energy-Density Physics (HEDP) division of MIT’s Plasma Science and Fusion Center (PSFC), along with four other universities, has been awarded a five-year, $10 million grant to establish a Stewardship Science Academic Alliances Center of Excellence. The PSFC will be the lead partner in the center, which includes the University of Iowa; the University of Nevada at Reno; the University of Rochester; and Virginia Polytechnic Institute and State University. 

The U.S. Department of Energy’s (DOE) National Nuclear Security Administration (NNSA) award will support educational and research missions across the partners. The goal of the newly established center is to generate exceptional experimental and theoretical PhDs in HEDP and inertial confinement fusion (ICF), while addressing issues of critical interest to the Department of Energy’s NNSA and national labs.

Officially called the Center for Advanced Nuclear Diagnostics and Platforms for Inertial ICF and HEDP at Omega, NIF and Z, the center will focus on the properties of plasma under extreme conditions of temperature, density and pressure. Center partners will collaborate closely with the Lawrence Livermore National Laboratory, Los Alamos National Laboratory, Sandia National Laboratory, the Laboratory for Laser Energetics, and General Atomics.

MIT's HEDP division has a long and established history of collaboration with these labs, regularly using Laser Energetics’s 30-kilojoule OMEGA laser, Lawrence Livermore’s National Ignition Facility, and Sandia's Z machine to pursue a wide range of research, including inertial confinement fusion, nuclear science, and laboratory astrophysics. The division has used its Accelerator Facility to develop and characterize diagnostics for these machines, and as part of the new center will pursue new diagnostic techniques for advanced research.

HEDP division head and Center of Excellence Director Richard Petrasso acknowledges the importance of this partnership.

“The center is about our work in inertial confinement fusion, and also in laboratory astrophysics, simulating aspects of astrophysical phenomena, such as the jetting in the crab nebula," Petrasso says. "There is lots of interesting physics that students and staff have been observing for years. This new center allows us, with our partners, to really expand our investigations.”

PSFC Director Dennis Whyte observed that the new center is a recognition of the HEDP division’s excellence. Thanking the team for the exceptional work, under the encouragement of the senior leadership, he said, “Your work is one of the gems of the PSFC. This division produces outstanding, unique science, and with a mission that is critical to national security.”

Launched in 2002, the Stewardship Science Academic Alliances Centers of Excellence program emphasizes areas of research that are relevant to NNSA’s stockpile stewardship mission, and promotes the education of the next generation of highly-trained nuclear scientists and engineers.

de MIT News https://ift.tt/2Ewqh4p

Independence problem solved though collaboration

In 2009, when Yufei Zhao was an MIT undergraduate, he was intrigued by a 2001 conjecture by Rutgers University mathematician Jeff Kahn regarding the number of independent sets in a graph. An independent set in a graph is a subset of vertices such that no two of them are joined by an edge.

“Many important structures can be modeled using independent sets,” said Zhao. “For example, if the graph models some kind of incompatibility, then an independent set represents a mutually compatible collection.”

Zhao was participating in a Research Experience for Undergraduates (REU) summer program in Duluth, Minnesota, and while he was researching what would be one of his first math research papers, he came across a combinatorics problem by Kahn. The problem puzzled him. An attempt to solve it came close, as he described in a paper he wrote with David Galvin in 2010 titled, “The number of independent sets in a graph with small maximum degree.”

Yufei Zhao graduated in 2010, received his PhD in 2015, and is now the Class of 1956 Career Development Assistant Professor at MIT. His focus is on combinatorics, discrete mathematics, and graph theory.

Through the years, that conjecture continued to nag at him, so last spring, Zhao decided to pass on the challenge to his “fearless” student mathematicians, sophomore Ashwin Sah and junior Mehtaab Sawhney.

Harvard senior David Stoner, who Sawhney coincidentally befriended at that REU program in Duluth, also takes combinatorics classes at MIT. When he heard about his friends’ project, he asked to join in.    

Zhao recalls that, over the course of a month, they debated their ideas online as well as at many a late-night marathon session in Building 2, “where they tore apart one inequality after another.”

“To my amazement, they came to me with a solution of this old conjecture,” he says. “Lots of experienced mathematicians have worked on this conjecture without success.”  

To cap off their success, they have published a paper — “The number of independent sets in an irregular graph” — in Journal of Combinatorial Theory, Series B, a leading journal in combinatorics.

“The approach in the paper, while not very difficult, ends ups being very technical,” says Sawhney. “Figuring out how to handle all the various terms that appear in our proof and reduce them into a manageable form plays a key role.”

Adds Stoner: “Probably the most challenging part in retrospect was discovering a particular application of Holder's Inequality. This allowed for the inductive inequality to be transformed into something completely local.”

For Sah, the hardest part was “figuring out the right approach and framing, and understanding the theorem in the correct way.”

“Once we believed that the ‘local inequality’ was true, it allowed us to view the problem in a very different way compared to what was already known, and although there are still lots of difficulties beyond this realization, it definitely underpins the whole effort,” Sah says.

Kahn was excited with the results. “It was great to see my old conjecture finally resolved, and even better to see what it led to. Some of the problems settled by Yufei and company had been tried by some excellent, much more experienced people. Of course, having a fresh young mind can also be an advantage.”

Zhao, for his part, is now happy to check that conjecture off his bucket list. But he is even happier to have students like them in his combinatorics class.

“The students are amazing,” he says. “They are constantly asking questions and bombarding me with ideas. I have learned so much from them.”

The techniques that they found to solve that conjecture quickly led to work on several related problems, including for their upcoming paper “A Reverse Sidorenko Inequality,” related to graph colorings and graph homomorphisms. Explains Zhao: “This paper solves several open problems concerning graph colorings and homomorphisms, including one of my favorite problems regarding maximizing the number of q-colorings in a d-regular graph.”   

The three students are prolific. Sawhney has written 12 papers so far, including four others with Stoner, and another with Sah, who is the author or coauthor of six papers in total. Stoner has eight papers so far.

“They are incredible,” said Zhao. “Most graduate students don’t have as many papers.”

All three enjoy working together. “Often we discuss ideas over the phone or in person and tend to communicate ideas quickly even if they are only half-baked,” Sawhney says. “This leads to always feeling as if there is something else to try on a problem we are working on.”

Adds Stoner: “When it comes time to execute these ideas in detail, we generally try to play to each of our individual strengths.”

Ashwin describes their dynamic as “pretty relaxed,” using Slack and other online mediums to discuss ideas and debate the shortcomings of various methods when they can’t meet up in person. “There's definitely always something happening and something to think about,” he says.

Sah says that collaborative atmosphere is what attracted him to MIT. 

“I'm definitely grateful to be able to work with this particular group of people on combinatorics research,” he says.

de MIT News https://ift.tt/2T2dQGs

Low-cost ventilator wins Sloan health care prize

If someone begins struggling to breathe on their own, machine ventilators are the best way to keep them alive. Unfortunately, the high price of the machines forces many hospitals in the poorest regions of the world to rely on a simple solution known as an Ambu Bag that requires hospital staff or even a patient’s own family member to apply constant manual pressure in order to get oxygen to the lungs.

Ambu Bags are imprecise and carry their own risks, which helps explain why respiratory disease is one of the leading causes of death in more than 60 developing countries worldwide.

On Feb. 21, the student team Umbulizer won $20,000 to help address that problem with a device it claims can help 90 percent of patients struggling to breathe, at a fraction of the cost of traditional ventilators. The money was part of the company’s first-place finish in the annual MIT Sloan Healthcare Innovations Prize competition.

The event, which is open to entrepreneurial students from Boston-area colleges and universities, featured eight finalist teams pitching their health care innovations to a group of judges and a packed audience at MIT Sloan’s Wong Auditorium.

Boston University graduate Shaheer Piracha and Harvard Medical School student Sanchay Gupta gave the winning pitch for the Umbulizer team, which also includes MIT alumni Moiz Imam ’18 and Abdurrahman Akkas ’18, MIT mechanical engineering student Wasay Anwer, Boston University student Rohan Jadeja, and Farzan Khan, who recently graduated from New York University Abu Dhabi.

The company’s device looks more like a desktop printer than a traditional bedside ventilator and is capable of running on batteries for added mobility. After an operator connects the device’s single tube to a patient, it rhythmically pumps a safe amount of air into their lungs.

Umbulizer’s device will cost around $2,000 compared to the $15,000 price tag of regular ventilators. The key to the student team’s cost savings is its decision to focus on providing the four most common functions of ventilators with their device. Machine ventilators are typically designed to perform 15 different functions, many of which are rarely needed to save a life, Piracha told the audience.

The team is currently focused on bringing its solution to Pakistan, a country with over 200 million people that Piracha says has less than 2,000 machine ventilators across all of its hospitals. The company says early results from a clinical trial there are promising.

“When we spoke to Pakistani doctors and hospital administrators, they expressed a need for a device that is simple to operate, capable of remote monitoring, portable, and built using locally sourced material. All of those considerations have informed our [first iteration of this machine],” Piracha told the audience. “Our device’s competitive advantage lies in the fact that we’ve balanced the accuracy and consistency of a traditional ventilator with the portability and affordability of an Ambu Bag.”

Piracha says the prize money will be put toward Umbulizer’s current clinical trial. If the company can find success selling to hospitals in Pakistan, it sees a market in countries across South Asia, Africa, and South America, regions where there are a combined 2 million preventable deaths from respiratory diseases each year.

The competition’s $4,000 second-place prize went to OcculAR, which has built a virtual reality headset capable of conducting remote eye exams on patients and sending 3-D video summaries of the exams to doctors for analysis.

“Worldwide, there are 276 million cases of blindness or severe vision loss, and fortunately 223 million of those cases are preventable or treatable. The problem is that, in order to prevent or treat [the problem], you first have to diagnose it,” OcculAR team member Brett Sternfield, an MBA candidate at Sloan, told the audience. “Today, ophthalmologists use a slit lamp. It’s a bulky machine, where the patient and the doctor have to be co-located in the same room, and it really limits the ability to get this technology outside of the clinic and into the rest of the world.”

The company’s prototype uses proprietary illumination and imaging software to offer the same functionality of a traditional slit lamp, with no moving parts. The data it gathers could also be used for future artificial intelligence-driven diagnostics.

The OcculAR team has begun securing strategic partnerships and is working to begin validation trials in the Boston area.

The $1,000 audience choice award went to Precavida, a digital platform that arranges appointments for uninsured patients with health care providers in Brazil. The company has been building its solution over the last five years and now works with over 100,000 patients and 175 health care providers, facilitating around 34,000 appointments each year.

The company is trying to fix some major shortcomings in the current system: Only about 25 percent of Brazilians have health insurance, and people wait an average of six months to see a specialist. Precavida, which charges patients for every appointment scheduled, can get patients in front of specialists in the same week or day, and charges rates 45 percent lower than the current private health system, according to team member and MIT Sloan Fellows MBA candidate Lais Fonseca Alves.

The company can also manage medical records, assist with prescriptions, schedule lab tests, and arrange follow up visits.

“Our goal is to become a one-stop shop platform for health care services,” Alves said.

The annual pitch competition is run by the student-run MIT Sloan Healthcare Club and is part of the Sloan Healthcare and BioInnovations Conference that was held Feb. 22.

The conference brings together members of the local health care ecosystem including academics, investors, government officials, and leaders in the private sector to discuss some of the most pressing issues and promising developments in the health care sector. This year’s theme explored the impact of digitalization on the industry.

A little over 40 teams applied to this year’s competition, but only eight teams earned the right to pitch in the finals. Each team had five minutes to give their presentations, which were judged based on five categories including the solution’s impact, novelty, market opportunity, feasibility, and traction.

The competition marked the end of a full week of support for finalists, which included networking opportunities, office hours with local venture capitalists, and workshops on skills such as delivering a successful pitch.

“A lot of the applicants are early stage, so they may have a great idea, but maybe their business model isn’t fully worked out, or they would like more practice in delivering a successful pitch to a diverse audience,” said Jasmine Kang, an organizer of the competition and a Sloan MBA candidate.

Other finalists of the competition included i^4, which has developed an incubator for infants that tracks four different vital signs to help prevent deaths from infection and hypothermia; Memoir Health, which provides recovering drug addicts with behavioral and mental health care both virtually and in-person; VASERA Male Contraceptives, which is developing a long-lasting, reversible male contraceptive; ATEM, which has developed a stylish, “smart” asthma inhaler that notifies users when they forget it and tracks usage; and 02 Therapies, which is developing a technology to map oxygen levels in tumors to enable personalized and optimized cancer therapies.

de MIT News https://ift.tt/2Ea3PNk

Oxygen-tracking method could improve diabetes treatment

Transplanting pancreatic islet cells into patients with diabetes is a promising alternative to the daily insulin injections that many of these patients now require. These cells could act as a bioartificial pancreas, monitoring blood glucose levels and secreting insulin when needed.

For this kind of transplantation to be successful, scientists need to make sure that the implanted cells receive enough oxygen, which they need in order to produce insulin and to remain viable. MIT engineers have now devised a way to measure oxygen levels of these cells over long periods of time in living animals, which should help them predict which implants will be most effective.

In a paper appearing in the Proceedings of the National Academy of Sciences the week of Feb. 25, the researchers demonstrated that they could use this method, a specialized type of magnetic resonance imaging (MRI), to track how oxygen levels of implanted cells in the intraperitoneal (IP) cavity of mice change as they move through the cavity over a prolonged period of time.

“Our goal is to make living cellular factories that can supply drugs on demand for patients. The ability to track the oxygen supply and the location of implanted cells will help us better understand how to build and use successful therapies,” says Daniel Anderson, an associate professor in MIT’s Department of Chemical Engineering, a member of MIT’s Koch Institute for Integrative Cancer Research and Institute for Medical Engineering and Science (IMES), and the senior author of the study.

Virginia Spanoudaki, the scientific director of the Koch Institute Animal Imaging and Preclinical Testing Core Facility, is the lead author of the study. Other authors include MIT postdocs Joshua Doloff and Shady Farah, research scientist Wei Huang, former research affiliate Samuel Norcross, and David H. Koch Institute Professor Robert Langer.

Better measurements

For the past several years, Anderson, Langer, and their colleagues have been developing implantable islet cells encapsulated in particles made of alginate, a starchy molecule naturally found in algae. Such particles could be used to replace the pancreatic islet cells of people with Type 1 diabetes, which do not function properly.

In an earlier study, the researchers found that larger particles, with a diameter of 1.5 millimeters, maintain their function longer than smaller particles (0.5-millimeter diameter), in part because the smaller particles tend to become surrounded by scar tissue, which blocks their access to oxygen.

However, questions still remained about the role of oxygen in the fate of these implanted cells. The particles can move through the IP space once implanted, which makes tracking them and their oxygen exposure important. Different parts of the IP space contain varying levels of oxygen, and previous studies had shown that the smaller particles tend to cluster in patches of fat, which have less oxygen, contributing to their failure.

Optical microsensors that are typically used for measuring oxygen levels in living tissue are very fragile and invasive, so the MIT team decided to try an alternative approach: fluorine MRI, a previously developed technique that other researchers have used to track living cells. While traditional MRI measures interactions between a magnetic field and hydrogen nuclei, fluorine MRI can measure similar interactions between a magnetic field and fluorine nuclei, as well as how these interactions are affected by the presence of oxygen.

To perform the study, the researchers incorporated a fluorine-containing material called a perfluorocarbon emulsion into the alginate that they normally use to encapsulate their islet cells. They tested particles with diameters of 0.5 and 1.5 millimeters, in both diabetic and nondiabetic mice. The nondiabetic mice received alginate implants with no cells inside, while the diabetic mice received implants with pancreatic islet cells.

The researchers then used fluorine MRI to measure oxygen levels in the IP space over a three-month period. At the same time, they also measured the diabetic mice’s blood glucose levels. To help them analyze the resulting data, the researchers used a machine-learning algorithm to go through all of the images and find associations between the positions of the capsules within the IP space, the oxygen levels, and the blood glucose levels of the mice.

“These kind of imaging studies involve a lot of data, and screening all of these 2-D images and making decisions about how the position of the capsules affects oxygen concentration is extremely challenging and very error prone when it’s done by a human observer,” Spanoudaki says. “So we relied on machine learning to automatically go through the images and find associations between the positions of the capsules and other parameters.”

This analysis revealed that the smaller capsules produce enough insulin to treat diabetic mice during the first 30 days of treatment, but then tend to organize in large clusters and accumulate in the fatty areas of the animals’ extremities. Once the particles become stuck in these oxygen-deprived regions, blood glucose levels rise in the mice.

The larger capsules tended to spread out over a larger area, so that some ended up in low-oxygen areas and others in high-oxygen areas. Overall, the cells secreted enough insulin to keep the diabetic mice’s blood glucose levels stable over several months.

Gordon Weir, the co-head of the Joslin Diabetes Center’s section on islet and regenerative biology, says the study sheds light on important issues regarding the optimal size of the alginate capsules used to deliver islet cells.

“The MIT group has previously shown the better transplant results in mice (and non-human primates) using capsules with a diameter of 1.5 millimeters compared with 0.5 millimeters,” says Weir, who was not involved in the research. “Now with this remarkable technique, we can see what we suspected: that the smaller capsules tend to clump more easily, which results in a more hypoxic environment that leads to impaired insulin secretion and more cell death.”

Toward a bioartificial pancreas

Sigilon Therapeutics, a company started by Langer, Anderson, and others to further develop the bioartificial pancreas, hopes to begin testing implantable islet cells in patients early next year, Anderson says. The new oxygen measurement technique could potentially be adapted for use in larger animals, including humans, which could help guide the development of future versions of the encapsulated islets, the researchers say.

“Based on measurements in larger animals, we would like to understand whether there are different ways to design the bioartificial pancreas, so that this aggregation of capsules that potentially results in reduced oxygen does not happen,” Spanoudaki says. “We are hoping to use this as a guide to make better designs for the bioartificial pancreas.”

The researchers are also hoping to adapt the fluorine MRI technology to study how oxygen levels affect other kinds of cell processes such as metastasis and immune cell activation.

The research was funded by JDRF, the Leona M. and Harry B. Helmsley Charitable Trust Foundation, the Parviz Tayebati Research Fund, and a Koch Institute Support (core) Grant from the National Cancer Institute.

de MIT News https://ift.tt/2UfphXK

Undark magazine wins George Polk Award for Environmental Reporting

In only its third year of existence, Undark magazine, a digital publication of the Knight Science Journalism Fellowship Program at MIT (KSJ), has been awarded a prestigious George Polk Award. The prize, announced at the National Press Club in Washington on Feb. 19, recognized the work of two-time Pulitzer Prize-winning photojournalist Larry C. Price and contributing Undark reporters for a seven-part series on global air pollution, published between August and December 2018, called "Breathtaking." 

Conceived and orchestrated by Undark's editorial team and supported in part by the Pulitzer Center on Crisis Reporting, the Breathtaking project visited seven countries on five continents to document — in text, drone footage, still photography, and innovative 360-degree video — the impacts of fine particulate air pollution, also called PM2.5. Such pollution claims more than 4 million lives annually.

"It's a topic that impacts virtually everyone to some degree, and yet it is far too often overlooked," said Undark's editor in chief, Tom Zeller Jr., a former New York Times staff writer and editor and a 2013-'14 MIT research fellow with KSJ. "We're delighted that the Polk committee has recognized Larry's work and that of our entire team — and we hope that this award will bring more awareness to this pressing issue."

The George Polk Awards, established in 1949 in memory of CBS correspondent George Polk, who was murdered while covering the Greek Civil War, are conferred annually by New York's Long Island University. They are considered to be among the most prestigious in journalism.

In addition to Price and Undark, the Polk committee granted prizes in 16 categories for journalistic work done in 2018 by The New York Times, The Washington Post, The New Yorker, and ProPublica, among other outlets. More than 550 entries were considered — a record year, according to the award's organizers. Previous winners include Bob Woodward, Carl Bernstein, Walter Cronkite, Edward R. Murrow, Christiane Amanpour, Norman Mailer, and Diane Sawyer.

"Few years have been as fruitful as this one," New York Times journalist John Darnton, now the curator of the Polk Awards, said in a statement accompanying the award announcements. “These winners tell us that the best of our journalists remain resilient, courageous, dedicated, and undeterred by attacks on their ability and integrity."

de MIT News https://ift.tt/2Nr9hj2

Twenty-five ways in which MIT has transformed computing

This month MIT is celebrating the launch of the new $1 billion MIT Stephen A. Schwarzman College of Computing. To help commemorate the event, here’s a list of 25 ways in which MIT has already transformed the world of computing technology.

1937: Digital circuits

Master’s student Claude Shannon showed that the principles of true/false logic could be used to represent the on-off states of electric switches — a concept that served as the foundation of the field of digital circuits, and, therefore, the entire industry of digital computing itself.

1944: The digital computer

The first digital computer that could operate in real-time came out of Project Whirlwind, a initiative during World War II in which MIT worked with the U.S. Navy to develop a universal flight simulator. The device’s success led to the creation of MIT Lincoln Laboratory in 1951.

1945: Memex

Professor Vannevar Bush proposed a data system called a “Memex” that would allow a user to “store all his books, records, and communications” and retrieve them at will — a concept that inspired the early hypertext systems that led, decades later, to the World Wide Web.

1958: Functional programming

The first functional programming language was invented at MIT by Professor John McCarthy. Before LISP, programming had difficulty juggling multiple processes at once because it was “procedural” (like cooking a recipe). Functional languages let you describe required behaviors more simply, allowing work on much bigger problems than ever before.

1959: The fax

In trying to understand the words of a strongly-accented colleague over the phone, MIT student Sam Asano was frustrated that they couldn’t just draw pictures and instantly send them to each other — so he created a technology to transmit scanned material through phone lines. His fax machine was licensed to a Japanese telecom company before becoming a worldwide phenomenon.

1962: The multiplayer video game

When a PDP-1 computer arrived at MIT’s Electrical Engineering Department, a group of crafty students — including Steven “Slug” Russell from Marvin Minsky’s artificial intelligence group — went to work creating “SpaceWar!,” a space-combat video game that became very popular among early programmers and is considered the world’s first multiplayer game. (Play it here.)

1963: The password

The average person has 13 passwords — and for that you can thank MIT’s Compatible Time-Sharing System, which by most accounts established the first computer password. “We were setting up multiple terminals which were to be used by multiple persons but with each person having his own private set of files,” Professor Corby Corbato told WIRED. “Putting a password on for each individual user as a lock seemed like a very straightforward solution.”

1963: Graphical user interfaces

Nearly 50 years before the iPad, an MIT PhD student had already come up with the idea of directly interfacing with a computer screen. The “Sketchpad” developed by Ivan Sutherland PhD ’63 allowed users to draw geometric shapes with a touch-pen, pioneering the practice of “computer-assisted drafting” — which has proven vital for architects, planners, and even toddlers.

1964: Multics

MIT spearheaded the time-sharing system that inspired UNIX and laid the groundwork for many aspects of modern computing, from hierarchical file systems to buffer-overflow security. Multics furthered the idea of the computer as a “utility” to be used at any time, like water or electricity.

1969: Moon code

Margaret Hamilton led the MIT team that coded the Apollo 11 navigation system, which landed astronauts Neil Armstrong and Buzz Aldrin ScD ’63 on the moon. The robust software overrode a command to switch the flight computer’s priority system to a radar system, and no software bugs were found on any crewed Apollo missions.

1971: Email

The first email to ever travel across a computer network was sent to two computers that were right next to each other — and it came from MIT alumnus Ray Tomlinson '65 when he was working at spinoff BBN Technologies. (He’s the one you can credit, or blame, for the @ symbol.)

1973: The PC

MIT Professor Butler Lampson founded Xerox’s Palo Alto Research Center (PARC), where his work earned him the title of “father of the modern PC.” The Xerox Alto platform was used to create the first graphical user interface (GUI), the first bitmapped display, and the first “What-You-See-Is-What-You-Get” (WYSIWYG) editor.

1977: Data encryption

E-commerce was first made possible by the MIT team behind the RSA algorithm, a method of data encryption based on the concept of how difficult it is to factor huge prime numbers. Who knew that math would be why you can get your last-minute holiday shopping done?

1979: The spreadsheet

In 1979, Bob Frankston '70 and Dan Brickson '73 worked late into the night on an MIT mainframe to create VisiCalc, the first electronic spreadsheet, which sold more than 100,000 copies in its first year. Three years later, Microsoft got into the game with “Multiplan,” a program that later became Excel.

1980: Ethernet

Before there was Wi-Fi, there was Ethernet — the networking technology that lets you get online with a simple cable plug-in. Co-invented by MIT alumnus Bob Metcalfe '69, who was part of MIT’s Project MAC team and later went on to found 3Com, Ethernet helped make the Internet the fast, convenient platform that it is today.

1980: The optical mouse

Undergrad Steve Kirsch '80 was the first to patent an optical computer mouse — he had wanted to make a “pointing device” with a minimum of precision moving parts — and went on to found Mouse Systems Corp. (He also patented the method of tracking online ad impressions through click-counting.)

1983: The growth of freeware

Early AI Lab programmer Richard Stallman was a major pioneer in hacker culture and the free-software movement through his GNU Project, which set out to develop a free alternative to the Unix OS, and laid the groundwork for Linux and other important computing innovations.

1985: Spanning tree algorithm

Radia Perlman '73, SM '76, PhD '88 hates when people call her “the mother of the Internet,” but her work developing the Spanning Tree Protocol was vital for being able to route data across global computer networks. (She also created LOGO, the first programming language geared toward children.)

1994: The World Wide Web consortium (W3C)

After inventing the web, Tim Berners-Lee joined MIT and launched a consortium devoted to setting global standards for building websites, browsers, and devices. Among other things, W3C standards ensure that sites are accessible, secure, and easily “crawled” for SEO.

1999: The birth of blockchain

MIT Institute Professor Barbara Liskov’s paper on Practical Byzantine Fault Tolerance helped kickstart the field of blockchain, a widely used cryptography system. Her team’s protocol could handle high-transaction throughputs and used concepts that are vital for many of today’s blockchain platforms.

2002: Roomba

While we don’t yet have robots running errands for us, we do have robo-vacuums — and for that, we can thank MIT spinoff iRobot. The company has sold more than 20 million of its Roombas and spawned an entire industry of automated cleaning products.

2007: The mobile personal assistant

Before Siri and Alexa, there was MIT Professor Boris Katz’s StartMobile, an app that allowed users to schedule appointments, get information, and do other tasks using natural language.

2012: EdX

Led by former CSAIL director Anant Agarwal, MIT’s not-for-profit online platform with Harvard University offers free courses that have drawn more than 18 million learners around the globe, all while being open-source and nonprofit.

2013: Boston Dynamics

Professor Marc Raibert’s spinoff Boston Dynamics builds bots like “Big Dog” and “Spot Mini” that can climb, run, jump and even do back-flips. Their humanoid robot Atlas was used in the DARPA Robotics Challenge aimed at developing robots for disaster relief sites.

2016: Robots you can swallow

CSAIL Director Daniela Rus’ ingestible origami robot can unfold itself from a swallowed capsule. Using an external magnetic field, it could one day crawl across your stomach wall to remove swallowed batteries or patch wounds.

de MIT News https://ift.tt/2H30naq