MIT Latest News

Subscribe to MIT Latest News feed
MIT News is dedicated to communicating to the media and the public the news and achievements of the students, faculty, staff and the greater MIT community.
Updated: 22 hours 22 min ago

Machines that learn language more like kids do

Wed, 10/31/2018 - 11:52am

Children learn language by observing their environment, listening to the people around them, and connecting the dots between what they see and hear. Among other things, this helps children establish their language’s word order, such as where subjects and verbs fall in a sentence.

In computing, learning language is the task of syntactic and semantic parsers. These systems are trained on sentences annotated by humans that describe the structure and meaning behind words. Parsers are becoming increasingly important for web searches, natural-language database querying, and voice-recognition systems such as Alexa and Siri. Soon, they may also be used for home robotics.

But gathering the annotation data can be time-consuming and difficult for less common languages. Additionally, humans don’t always agree on the annotations, and the annotations themselves may not accurately reflect how people naturally speak.

In a paper being presented at this week’s Empirical Methods in Natural Language Processing conference, MIT researchers describe a parser that learns through observation to more closely mimic a child’s language-acquisition process, which could greatly extend the parser’s capabilities. To learn the structure of language, the parser observes captioned videos, with no other information, and associates the words with recorded objects and actions. Given a new sentence, the parser can then use what it’s learned about the structure of the language to accurately predict a sentence’s meaning, without the video.

This “weakly supervised” approach — meaning it requires limited training data — mimics how children can observe the world around them and learn language, without anyone providing direct context. The approach could expand the types of data and reduce the effort needed for training parsers, according to the researchers. A few directly annotated sentences, for instance, could be combined with many captioned videos, which are easier to come by, to improve performance.

In the future, the parser could be used to improve natural interaction between humans and personal robots. A robot equipped with the parser, for instance, could constantly observe its environment to reinforce its understanding of spoken commands, including when the spoken sentences aren’t fully grammatical or clear. “People talk to each other in partial sentences, run-on thoughts, and jumbled language. You want a robot in your home that will adapt to their particular way of speaking … and still figure out what they mean,” says co-author Andrei Barbu, a researcher in the Computer Science and Artificial Intelligence Laboratory (CSAIL) and the Center for Brains, Minds, and Machines (CBMM) within MIT’s McGovern Institute.

The parser could also help researchers better understand how young children learn language. “A child has access to redundant, complementary information from different modalities, including hearing parents and siblings talk about the world, as well as tactile information and visual information, [which help him or her] to understand the world,” says co-author Boris Katz, a principal research scientist and head of the InfoLab Group at CSAIL. “It’s an amazing puzzle, to process all this simultaneous sensory input. This work is part of bigger piece to understand how this kind of learning happens in the world.”

Co-authors on the paper are: first author Candace Ross, a graduate student in the Department of Electrical Engineering and Computer Science and CSAIL, and a researcher in CBMM; Yevgeni Berzak PhD ’17, a postdoc in the Computational Psycholinguistics Group in the Department of Brain and Cognitive Sciences; and CSAIL graduate student Battushig Myanganbayar.

Visual learner

For their work, the researchers combined a semantic parser with a computer-vision component trained in object, human, and activity recognition in video. Semantic parsers are generally trained on sentences annotated with code that ascribes meaning to each word and the relationships between the words. Some have been trained on still images or computer simulations.

The new parser is the first to be trained using video, Ross says. In part, videos are more useful in reducing ambiguity. If the parser is unsure about, say, an action or object in a sentence, it can reference the video to clear things up. “There are temporal components — objects interacting with each other and with people — and high-level properties you wouldn’t see in a still image or just in language,” Ross says.

The researchers compiled a dataset of about 400 videos depicting people carrying out a number of actions, including picking up an object or putting it down, and walking toward an object. Participants on the crowdsourcing platform Mechanical Turk then provided 1,200 captions for those videos. They set aside 840 video-caption examples for training and tuning, and used 360 for testing. One advantage of using vision-based parsing is “you don’t need nearly as much data — although if you had [the data], you could scale up to huge datasets,” Barbu says.

In training, the researchers gave the parser the objective of determining whether a sentence accurately describes a given video. They fed the parser a video and matching caption. The parser extracts possible meanings of the caption as logical mathematical expressions. The sentence, “The woman is picking up an apple,” for instance, may be expressed as: λxy. woman x, pick_up x y, apple y.

Those expressions and the video are inputted to the computer-vision algorithm, called “Sentence Tracker,” developed by Barbu and other researchers. The algorithm looks at each video frame to track how objects and people transform over time, to determine if actions are playing out as described. In this way, it determines if the meaning is possibly true of the video.

Connecting the dots

The expression with the most closely matching representations for objects, humans, and actions becomes the most likely meaning of the caption. The expression, initially, may refer to many different objects and actions in the video, but the set of possible meanings serves as a training signal that helps the parser continuously winnow down possibilities. “By assuming that all of the sentences must follow the same rules, that they all come from the same language, and seeing many captioned videos, you can narrow down the meanings further,” Barbu says.

In short, the parser learns through passive observation: To determine if a caption is true of a video, the parser by necessity must identify the highest probability meaning of the caption. “The only way to figure out if the sentence is true of a video [is] to go through this intermediate step of, ‘What does the sentence mean?’ Otherwise, you have no idea how to connect the two,” Barbu explains. “We don’t give the system the meaning for the sentence. We say, ‘There’s a sentence and a video. The sentence has to be true of the video. Figure out some intermediate representation that makes it true of the video.’”

The training produces a syntactic and semantic grammar for the words it’s learned. Given a new sentence, the parser no longer requires videos, but leverages its grammar and lexicon to determine sentence structure and meaning.

Ultimately, this process is learning “as if you’re a kid,” Barbu says. “You see world around you and hear people speaking to learn meaning. One day, I can give you a sentence and ask what it means and, even without a visual, you know the meaning.”

“This research is exactly the right direction for natural language processing,” says Stefanie Tellex, a professor of computer science at Brown University who focuses on helping robots use natural language to communicate with humans. “To interpret grounded language, we need semantic representations, but it is not practicable to make it available at training time. Instead, this work captures representations of compositional structure using context from captioned videos. This is the paper I have been waiting for!”

In future work, the researchers are interested in modeling interactions, not just passive observations. “Children interact with the environment as they’re learning. Our idea is to have a model that would also use perception to learn,” Ross says.

This work was supported, in part, by the CBMM, the National Science Foundation, a Ford Foundation Graduate Research Fellowship, the Toyota Research Institute, and the MIT-IBM Brain-Inspired Multimedia Comprehension project.

Scene at MIT: Happy Nanoween

Wed, 10/31/2018 - 12:00am

As part of her research on nanomaterials, PhD student Ashley Kaiser recently grew millions of carbon nanotubes — each incredibly strong and only 1/10,000 the width of a human hair — and immersed them in a guiding liquid. Upon drying, the resulting nanotube "forest" created a recognizable spooky pattern.

"The initial motivation behind this work was to densify carbon nanotube forests into predictable, cellular patterns by gently wetting them with a liquid, a process that can help enable scalable nanomaterial manufacturing," says Kaiser, who studies in the lab of Professor Brian Wardle. "The pattern was not precisely planned. While I knew that the carbon nanotubes would form cell-like shapes, I didn't know that these three particular sections would spell out 'Boo' so nicely, so it was a pretty special find."

The image was captured using a scanning electron microscope, which produces images in greyscale; the orange color was added later as a special effect. "It was exciting to find this under the microscope, and I thought that it would be great for Halloween the moment I saw it!" Kaiser says.

Submitted by: William Litant/Department of Aeronautics and Astronautics | Image by: Ashley Kaiser

Have a creative photo of campus life you'd like to share? Submit it to Scene at MIT.

Letter from President Reif: Consoling each other and helping to heal the world

Tue, 10/30/2018 - 12:03pm

The following email was sent today to the MIT community by President L. Rafael Reif.

To the members of the MIT community,

As our nation once again confronts heartbreaking mass violence, sending this annual reminder of MIT’s policies against harassment may feel to some as inconsequential and almost irrelevant.

Yet these policies could not have greater consequence, because they embody our conviction that the ultimate measure of our community is how we treat one another.

By reminding us that violence, racism, harassment and bullying are out of bounds – period – our policies can help lead us from error. Yet they cannot lead us towards the light: the essential duty to treat each other with respect, sympathy, decency, humility and kindness; the responsibility each of us has to make sure that everyone at MIT can truly feel at home; the challenge of finding a way to repair our fractured nation. This work we must do for ourselves.

Our policies also demonstrate that official statements matter – for good or ill. For instance, a recent draft of a government policy would redefine gender in a way that would erase the dignity and lived reality of well over a million transgender Americans, including many members of our MIT community. And next week in Massachusetts, the civil rights of these Americans are up for a vote.

Let me be clear: No matter how government policy may change, it will not change or weaken MIT’s commitment to protecting the rights and safety of every member of the MIT community.

Ultimately, nothing we do or say at MIT can reverse the fact that, from Pittsburgh to Jeffersontown, Charleston to Orlando, a baseball field in Maryland to the Boston Marathon, fellow human beings have been targeted and killed for being themselves.

Against the backdrop of our daily lives, such hatred and violence are much too frequent now. But we can and must fight the numb helplessness that might allow these acts to ever feel “normal.” We must keep ourselves alive to the shock and the pain, and stay focused on finding a better path for our society.

Tomorrow our community will come together to honor those killed or injured and those who helped them, and to console each other.

Vigil for Hope in the Face of Hate
Wednesday, October 31
12:30 PM
Steps of the Student Center (W20)

I am grateful for the way we live and work together at MIT. I am proud that we do not fear each other or the world. As one can see any day in the Infinite Corridor, our openness to talent from every faith, culture, nation and background is central to our success, and central to our humanity. We should never forget the value and strength of this deeply American idea.

In this difficult time, we must use the strength, ingenuity and optimism of our community to help heal the world.

Sincerely,

L. Rafael Reif

Analyzing the 2018 election: Insights from MIT scholars

Tue, 10/30/2018 - 12:00pm

For the 2018 version of the Election Insights series, MIT humanities, arts, and social science faculty members are offering research-based perspectives on issues of importance to the country — ranging from the future of work to national security to civic discourse and the role that, as the Constitution states, "we, the people" have in the defense of democracy itself.

In addition to commentaries, the series also includes "Music for the Midterms," a lively playlist created by our music faculty, and an annotated election book list consisting of nine works selected by MIT humanities scholars for their value illuminating this moment in American history.

Please, remember to vote on or before Nov. 6.

Commentary: On civil society and the defense of democracy

"What is written in a constitution can take a nation only so far unless society is willing to act to protect it. Every constitutional design has its loopholes, and every age brings its new challenges, which even farsighted constitutional designers cannot anticipate. We have to keep reminding ourselves that the future of our much-cherished institutions depends not on others but on ourselves, and that we are all individually responsible for our institutions." —Daron Acemoglu, the Elizabeth and James Killian Professor of Economics  Read more >>

Commentary: On partisan politics

"Partisan polarization is one of most important political developments of the past half-century. Of course, Democrats and Republicans have always taken divergent positions on issues ranging from slavery to internal improvements. Nevertheless, contemporary polarization differs from that of earlier eras, if only because the U.S. government directly shapes the lives of so many more people, in the U.S. and around the world." —Devin Caughey, associate professor of political science  Read more >>

Commentary: On media technology and immigration policy

"Widespread access to social media lowers the barrier for communities that have been marginalized by mass media and makes it easier for them to gain visibility and adherents. How might any of this affect the midterm elections? Here are three brief hypotheses, based on my ongoing research into the relationship between media technologies and social movements." —Sasha Costanza-Chock, associate professor of civic media Read more >>

Commentary: On democracy and civic discourse

"Elections are helpful reminders (as if we needed any) that we do not all agree. Yet, we must somehow figure out how to get along despite our disagreements. In particular, we may wonder whether, and to what extent, we should tolerate views we disagree with. In some cases, a well-functioning discursive market — a public forum of diverse views — may require us to respond to certain views with 'discursive intolerance." —Justin Khoo, associate professor of philosophy  Read more >>

Commentary: On female candidates of color

“A record number of women have filed as candidates this year, and a record number have won primaries in House and Senate races. Women of color make up one-third of the women candidates for the House, and three of four female gubernatorial nominees are women of color." —Helen Elaine Lee, professor of writing  Read more >>

Commentary: On social media and youth political engagement

"Although discussions about youth and new media tend to assume that something about the technology itself is responsible for political and social changes, in fact, the political possibilities associated with contemporary media are highly contingent upon societal power structures.” —Jennifer Light, the Bern Dibner Professor of the History of Science and Technology  Read more >>

Commentary: On the U.S.-North Korea relationship

"The North Korean nuclear program is not something to be 'solved' — that window has closed — it is an issue to be managed. The good news is that the United States has a lot of experience managing the emergence of new nuclear weapons powers." —Vipin Narang, associate professor of political science  Read more >>

Commentary: On reducing gun violence

"America’s gun culture is a resilient fact of political life. Attempts to reverse the country’s appetite for firearms have largely failed, even as gun violence persists at an astonishing pace. Lately, however, a social movement to challenge gun culture has rocked politics for the first time in a generation." —John Tirman, executive director and principal research scientist in the Center for International Studies  Read more >>

Commentary: On American identity

"The stories and interpretations that different groups of Americans offer of economic changes, including the loss of manufacturing jobs and growing inequality, are central to how they understand their own social positions as well as the kinds of economic and political futures they can envision. Many Americans are now struggling for a way to understand and talk about these economic changes — changes that are also apparent in other wealthy countries but more extreme in the United States.” —Christine Walley, professor of anthropology  Read more >>

Playlist: Music for the Midterms

As America heads toward the 2018 midterm elections on Nov. 6, MIT Music faculty offer a wide-ranging playlist — from Verdi to Gershwin to Lin-Manuel Miranda — along with notes on why each work resonates with this election season. Access the playlist >>

Annotated election book list: Reading for the Midterms

As the 2018 midterms approach, MIT writers and scholars in the humanities offer a selection of nine books — along with notes on why each work is illuminating for this moment in American political history. Browse the book list >>

Model paves way for faster, more efficient translations of more languages

Tue, 10/30/2018 - 10:59am

MIT researchers have developed a novel “unsupervised” language translation model — meaning it runs without the need for human annotations and guidance — that could lead to faster, more efficient computer-based translations of far more languages.

Translation systems from Google, Facebook, and Amazon require training models to look for patterns in millions of documents — such as legal and political documents, or news articles — that have been translated into various languages by humans. Given new words in one language, they can then find the matching words and phrases in the other language.

But this translational data is time consuming and difficult to gather, and simply may not exist for many of the 7,000 languages spoken worldwide. Recently, researchers have been developing “monolingual” models that make translations between texts in two languages, but without direct translational information between the two.

In a paper being presented this week at the Conference on Empirical Methods in Natural Language Processing, researchers from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) describe a model that runs faster and more efficiently than these monolingual models.

The model leverages a metric in statistics, called Gromov-Wasserstein distance, that essentially measures distances between points in one computational space and matches them to similarly distanced points in another space. They apply that technique to “word embeddings” of two languages, which are words represented as vectors — basically, arrays of numbers — with words of similar meanings clustered closer together. In doing so, the model quickly aligns the words, or vectors, in both embeddings that are most closely correlated by relative distances, meaning they’re likely to be direct translations.

In experiments, the researchers’ model performed as accurately as state-of-the-art monolingual models — and sometimes more accurately — but much more quickly and using only a fraction of the computation power.

“The model sees the words in the two languages as sets of vectors, and maps [those vectors] from one set to the other by essentially preserving relationships,” says the paper’s co-author Tommi Jaakkola, a CSAIL researcher and the Thomas Siebel Professor in the Department of Electrical Engineering and Computer Science and the Institute for Data, Systems, and Society. “The approach could help translate low-resource languages or dialects, so long as they come with enough monolingual content.”

The model represents a step toward one of the major goals of machine translation, which is fully unsupervised word alignment, says first author David Alvarez-Melis, a CSAIL PhD student: “If you don’t have any data that matches two languages … you can map two languages and, using these distance measurements, align them.”

Relationships matter most

Aligning word embeddings for unsupervised machine translation isn’t a new concept. Recent work trains neural networks to match vectors directly in word embeddings, or matrices, from two languages together. But these methods require a lot of tweaking during training to get the alignments exactly right, which is inefficient and time consuming.

Measuring and matching vectors based on relational distances, on the other hand, is a far more efficient method that doesn’t require much fine-tuning. No matter where word vectors fall in a given matrix, the relationship between the words, meaning their distances, will remain the same. For instance, the vector for “father” may fall in completely different areas in two matrices. But vectors for “father” and “mother” will most likely always be close together.

“Those distances are invariant,” Alvarez-Melis says. “By looking at distance, and not the absolute positions of vectors, then you can skip the alignment and go directly to matching the correspondences between vectors.”

That’s where Gromov-Wasserstein comes in handy. The technique has been used in computer science for, say, helping align image pixels in graphic design. But the metric seemed “tailor made” for word alignment, Alvarez-Melis says: “If there are points, or words, that are close together in one space, Gromov-Wasserstein is automatically going to try to find the corresponding cluster of points in the other space.”

For training and testing, the researchers used a dataset of publicly available word embeddings, called FASTTEXT, with 110 language pairs. In these embeddings, and others, words that appear more and more frequently in similar contexts have closely matching vectors. “Mother” and “father” will usually be close together but both farther away from, say, “house.”

Providing a “soft translation”

The model notes vectors that are closely related yet different from the others, and assigns a probability that similarly distanced vectors in the other embedding will correspond. It’s kind of like a “soft translation,” Alvarez-Melis says, “because instead of just returning a single word translation, it tells you ‘this vector, or word, has a strong correspondence with this word, or words, in the other language.’”

An example would be in the months of the year, which appear closely together in many languages. The model will see a cluster of 12 vectors that are clustered in one embedding and a remarkably similar cluster in the other embedding. “The model doesn’t know these are months,” Alvarez-Melis says. “It just knows there is a cluster of 12 points that aligns with a cluster of 12 points in the other language, but they’re different to the rest of the words, so they probably go together well. By finding these correspondences for each word, it then aligns the whole space simultaneously.”

The researchers hope the work serves as a “feasibility check,” Jaakkola says, to apply Gromov-Wasserstein method to machine-translation systems to run faster, more efficiently, and gain access to many more languages.

Additionally, a possible perk of the model is that it automatically produces a value that can be interpreted as quantifying, on a numerical scale, the similarity between languages. This may be useful for linguistics studies, the researchers say. The model calculates how distant all vectors are from one another in two embeddings, which depends on sentence structure and other factors. If vectors are all really close, they’ll score closer to 0, and the farther apart they are, the higher the score. Similar Romance languages such as French and Italian, for instance, score close to 1, while classic Chinese scores between 6 and 9 with other major languages.

“This gives you a nice, simple number for how similar languages are … and can be used to draw insights about the relationships between languages,” Alvarez-Melis says.

The BSU at 50

Tue, 10/30/2018 - 10:56am

In 1968, the black student community at MIT was small and needed a way to amplify its voice. Formed during that tumultuous year in political and racial history in the U.S., the MIT Black Students’ Union (BSU) launched a journey of advocacy and community that now continues 50 years later.

In the late 1960s, about 11 percent of Americans were black, but each 1,000-member class at MIT had perhaps half a dozen black students. Galvanized by the assassination of Martin Luther King Jr., black student groups were forming at overwhelmingly white college campuses across the country, and MIT was no exception. The students who started MIT BSU had two goals in mind: to support each other and to bring more black students to the Institute. “Surely there were more than three blacks in the high school class of 1965 who could belong to the MIT tribe,” says Linda C. Sharpe ’69, one of the BSU founders, who is a past president of the MIT Alumni Association and a former MIT Corporation member.

In fall of 1968, the new group drew up and presented a list of recommendations to the MIT administration: increasing the number of black students, creating a pre-­enrollment summer program for minority students, and hiring more minority faculty members. In response, MIT established the Task Force on Educational Opportunity (TFEO), which was made up of a group of BSU representatives and MIT administrators and chaired by associate provost (and future MIT president) Paul Gray ’54, SM ’55, ScD ’60. Through a series of often intense discussions, the TFEO designed the summer program, called Project Interphase, and came up with more inclusive approaches to things like recruitment, admissions, and financial aid.

“The Institute rolled up its sleeves and attacked [the recommendations] in the MIT way — that is, being very analytical about what the challenges and problems were, and then trying to figure out solutions to those challenges,” says founding BSU co-chair Shirley Ann Jackson ’68, PhD ’73, who went on to become the first black woman to earn a PhD from MIT and is now the president of Rensselaer Polytechnic Institute and a life member of the MIT Corporation. “That doesn’t mean there wasn’t great emotion around it, because there really, really was on all sides.”

Other key players in the birth of the BSU were founding cochair James Turner, PhD ’71, Jennifer Rudd ’68, Charles Kidwell ’69, Nathan Seely ’70, Sekazi Mtingwa ’71, Fred Johnson ’72, and Ronald Mickens, who was a postdoctoral associate in physics.

Gray, who died in 2017, would ultimately recall that being part of the Task Force was eye-opening: “I came away with an understanding I had none of two years before, as best a white person can understand what it was like to be black in the United States in the era before and during the civil rights revolution. It was a powerful experience.”

Through the efforts of the Admissions Office and BSU members who began recruiting black applicants from all over the country, the number of African-American students jumped to about 50 in the Class of 1973 and continued to rise, as did the numbers of women and other members of underrepresented minorities. Meanwhile, in an event modeled after political takeovers of buildings on other college campuses, a group of black students disrupted a meeting of the MIT Corporation in 1970 to advocate for the BSU demands and support kitchen workers involved in a labor dispute.

“The BSU has always played a major role in helping the Institute to not fall back from the goals of commitment and participation of black students, faculty, and administration. It’s been a key agent in helping MIT look at itself,” says adjunct professor emeritus of urban studies Clarence Williams, HM ’09, who joined the MIT administration in 1972 as assistant dean of the graduate school and has since served in multiple positions, including as acting director of the Office of Minority Education, special assistant to the president and chancellor, and Institute ombudsperson. Williams, who started the Black History Project in 1995, is the author of Technology and the Dream: Reflections on the Black Experience at MIT, 1941–1999 (MIT Press, 2001) and co-produced the 1996 video “It’s Intuitively Obvious,” which documented the experience of black students at MIT.

In addition to working to increase the number of black students on campus, the BSU advocated for recruitment and retention of black faculty and staff. “We also sought to broaden the dialogue on campus around issues pertinent to our community,” says Michelle Harton ’83, the outgoing chair of Black Alumni of MIT (BAMIT). Over the years, the BSU has organized discussion panels and cultural events, hosted prospective minority students, and played a central role in MIT’s annual Black History Month observance. Among the speakers brought to campus by the BSU are Benjamin L. Hooks, then executive director of the NAACP, and Ivan Van Sertima, author of They Came Before Columbus: The African Presence in Ancient America.

Five decades after the formation of the BSU, black students now make up 6.2 percent of MIT’s undergraduate population (as of fall 2017), up from 0.6 percent in 1968. And Sharpe notes that today, “the number of black women in the freshman class is nearly equal to the number of all women in my class.” She adds, “Times do change, if a lot more slowly than we would like.”

And the work continues today. In a parallel to the 1968 BSU proposals, the BSU and the Black Graduate Student Association (BGSA) met with President L. Rafael Reif in 2015, following several racially charged incidents across the country. The two groups issued a set of recommendations that included diversity orientation and training for all students, a diversity representative within each department, an MIT Medical clinician specializing in psychological issues affecting African-Americans, and a requirement that all undergraduates take an “immersion studies” elective focusing on multiculturalism or diversity. BAMIT and other groups also made recommendations. Many have already been completely or partially implemented, and conversations on how to advance other recommendations on the departmental level are ongoing.

“The work that needs to be done at MIT is similar to what needs to take place across the country — greater cultural understanding and value for the differences that people bring, plus mechanisms for civil discourse,” says Elaine Harris ’78, a BAMIT board member and cosponsor of what’s now called the Hack for Inclusion, an annual hackathon to tackle issues of bias, diversity, and inclusion. Outcomes from the hackathon include projects to create a more welcoming Boston for the black community and to address bias in machine learning. “I wish that the problem-­solving skills we apply to technical challenges and the metrics we develop to assess progress could be used in the domain of diversity, equity, and inclusion,” she says.

The BSU held an on-campus event in February to celebrate its 50-year legacy of advocating for black students and all minorities at MIT. In June, Jackson and Rudd — the first two black women to earn undergraduate degrees at MIT — became the first black women to earn their red jackets at their 50th reunion, where Jackson also served as a class speaker. In November, she is slated to speak at the BAMIT capstone event, “Road to 50: The Power of Community,” which will feature historical recollections, discussions, and a look forward.

Kelvin Green ’21, current cochair of the BSU, believes the organization is still playing an integral part to ensure equality within the MIT community, and that’s one of the reasons why he chose the Institute.

“Diversity is but a stepping-stone toward a higher goal,” he says. “Inclusion — truly valuing the people brought to this campus in all of the identities they bring — is where we must look. Let us not stop at the stepping-stone of diversity and ponder why it cannot support our weight; we must transition to the rock of inclusion, which is by definition created to support us all.”

This article originally appeared in the MIT News section of the November/December 2018 issue of MIT Technology Review magazine.

Inside these fibers, droplets are on the move

Mon, 10/29/2018 - 3:00pm

Microfluidics devices are tiny systems with microscopic channels that can be used for chemical or biomedical testing and research. In a potentially game-changing advance, MIT researchers have now incorporated microfluidics systems into individual fibers, making it possible to process much larger volumes of fluid, in more complex ways. In a sense, the advance opens up a new “macro” era of microfluidics.

Traditional microfluidics devices, developed and used extensively over the last couple of decades, are manufactured onto microchip-like structures and provide ways of mixing, separating, and testing fluids in microscopic volumes. Medical tests that only require a tiny droplet of blood, for example, often rely on microfluidics. But the diminutive scale of these devices also poses limitations; for example, they generally aren’t useful for procedures that need larger volumes of liquid to detect substances present in minute amounts.

A team of MIT researchers found a way around that, by making microfluidic channels inside fibers. The fibers can be made as long as needed to accommodate larger throughput, and they offer great control and flexibility over the shapes and dimensions of the channels. The new concept is described in a paper appearing this week in the journal Proceedings of the National Academy of Sciences, written by MIT graduate student Rodger Yuan, professors Joel Voldman and Yoel Fink, and four others.

A multidisciplinary approach

The project came about as a result of a “speedstorming” event (an amalgam of brainstorming and speed dating, an idea initiated by Professor Jeffrey Grossman) that was instigated by Fink when he was director of MIT’s Research Laboratory of Electronics. The events are intended to help researchers develop new collaborative projects, by having pairs of students and postdocs brainstorm for six minutes at a time and come up with hundreds of ideas in an hour, which are ranked and evaluated by a panel. In this particular speedstorming session, students in electrical engineering worked with others in materials science and microsystems technology to develop a novel approach to cell sorting using a new class of multimaterial fibers.

Yuan explains that, although microfluidic technology has been extensively developed and widely used for processing small amounts of liquid, it suffers from three inherent limitations related to the devices’ overall size, their channel profiles, and the difficulty of incorporating additional materials such as electrodes.

Because they are typically made using chip-manufacturing methods, microfluidic devices are limited to the size of the silicon wafers used in such systems, which are no more than about 8 inches across. And the photolithography methods used to make such chips limit the shapes of the channels; they can only have square or rectangular cross sections. Finally, any additional materials, such as electrodes for sensing or manipulating the channels’ contents, must be individually placed in position in a separate process, severely limiting their complexity.

“Silicon chip technology is really good at making rectangular profiles, but anything beyond that requires really specialized techniques,” says Yuan, who carried out the work as part of his doctoral research. “They can make triangles, but only with certain specific angles.” With the new fiber-based method he and his team developed, a variety of cross-sectional shapes for the channels can be implemented, including star, cross, or bowtie shapes that may be useful for particular applications, such as automatically sorting different types of cells in a biological sample.

In addition, for conventional microfluidics, elements such as sensing or heating wires, or piezoelectric devices to induce vibrations in the sampled fluids, must be added at a later processing stage. But they can be completely integrated into the channels in the new fiber-based system.

A shrinking profile

Like other complex fiber systems developed over the years in the laboratory of co-author Yoel Fink, professor of materials science and engineering and head of the Advanced Functional Fabrics of America (AFFOA) consortium, these fibers are made by starting with an oversized polymer cylinder called a preform. These preforms contain the exact shape and materials desired for the final fiber, but in much larger form — which makes them much easier to make in very precise configurations. Then, the preform is heated and loaded into a drop tower, where it is slowly pulled through a nozzle that constricts it to a narrow fiber that’s one-fortieth the diameter of the preform, while preserving all the internal shapes and arrangements.

In the process, the material is also elongated by a factor of 1,600, so that a 100-millimeter-long (4-inch-long) preform, for example, becomes a fiber 160 meters long (about 525 feet), thus dramatically overcoming the length limitations inherent in present microfluidic devices. This can be crucial for some applications, such as detecting microscopic objects that exist in very small concentrations in the fluid — for example, a small number of cancerous cells among millions of normal cells.

“Sometimes you need to process a lot of material because what you’re looking for is rare,” says Voldman, a professor of electrical engineering who specializes in biological microtechnology. That makes this new fiber-based microfluidics technology especially appropriate for such uses, he says, because “the fibers can be made arbitrarily long,” allowing more time for the liquid to remain inside the channel and interact with it.

While traditional microfluidics devices can make long channels by looping back and forth on a small chip, the resulting twists and turns change the profile of the channel and affect the way the liquid flows, whereas in the fiber version these can be made as long as needed, with no changes in shape or direction, allowing uninterrupted flow, Yuan says.

The system also allows electrical components such as conductive wires to be incorporated into the fiber. These can be used for example to manipulate cells, using a method called dielectrophoresis, in which cells are affected differently by an electric field produced between two conductive wires on the sides of the channel.

With these conductive wires in the microchannel, one can control the voltage so the forces are “pushing and pulling on the cells, and you can do it at high flow rates,” Voldman says.

As a demonstration, the team made a version of the long-channel fiber device designed to separate cells, sorting dead cells from living ones, and proved its efficiency in accomplishing this task. With further development, they expect to be able to perform more subtle discrimination between cell types, Yuan says.

“For me this was a wonderful example of how proximity between research groups at an interdisciplinary lab like RLE leads to groundbreaking research, initiated and led by a graduate student. We the faculty were essentially dragged in by our students,” Fink says.

The researchers emphasize that they do not see the new method as a substitute for present microfluidics, which work very well for many applications. “It’s not meant to replace; it’s meant to augment” present methods, Voldman says, allowing some new functions for particular uses that have not previously been possible.

“Exemplifying the power of interdisciplinary collaboration, a new understanding arises here from unexpected combinations of manufacturing, materials science, biological flow physics, and microsystems design,” says Amy Herr, a professor of bioengineering at the University of California at Berkeley, who was not involved in this research. She adds that this work “adds important degrees of freedom — regarding geometry of fiber cross-section and material properties — to emerging fiber-based microfluidic design strategies.”

The team included graduate student Jaemyon Lee, Hao Wei Su PhD ’16, and postdocs Etgar Levy and Tural Khudryev. The work was supported by the National Science Foundation, the National Institutes of Health, the Defense Advanced Research Projects Agency, the U.S. Army Research Laboratory and the U.S. Army Research Office through the Institute for Soldier Nanotechnologies at MIT, and the Center for Materials Science and Engineering.

3Q: Alyce Johnson on upholding the rights of MIT’s transgender community

Mon, 10/29/2018 - 10:33am

When Massachusetts voters head to the polls on Nov. 6 for the 2018 midterm election, one of the items they’ll find on the ballot is Question 3, about whether to uphold a 2016 state law barring discrimination against transgender people in public spaces such as stores and restaurants.

With the election drawing near, MIT News spoke with interim Institute Community and Equity Officer Alyce Johnson, who leads the ICEO in its mission to advance a respectful and caring community that embraces diversity and empowers everyone to learn and do their best at MIT.

Johnson spoke about the Institute’s commitment to protecting the rights of transgender members of the MIT community, regardless of the ballot initiative’s outcome, and why respectful, open dialogue is an essential component of MIT’s campus life.

Q: What would be the effect on MIT policies if Question 3 does not pass, and the law protecting transgender people from discrimination in public places is overturned?

A: MIT’s senior administration and I want our community to know that our support for MIT’s transgender community is steadfast, and that the Institute’s nondiscrimination policy, which expressly prohibits discrimination based on gender identity, will remain in place if the state law is repealed following the Nov. 6 election. The same will be true if the federal government more narrowly defines gender under federal law, a change that is reportedly under consideration. I want to reassure our community that MIT is permitted to offer antidiscrimination protections that are broader than what is protected by state and federal law. The Institute’s policies will continue to prohibit discrimination or harassment at MIT based on sex, sexual orientation, and gender identity.

In addition to the policies we have in place, our unwavering commitment to the rights of transgender students, staff, and faculty will also continue regardless of the outcome of the election or of any potential federal policy changes. As just one example, the pilot project to designate certain all-gender restrooms on campus will keep moving forward.

We are working very hard to provide resources and support for those who feel under attack or not safe. The chancellor’s office is doing a great job, with events planned through the Division of Student Life and the Office of Minority Education (OME), of providing welcoming spaces for students during and after the election. The Rainbow Lounge and SPXCE Intercultural Center will both have open houses on Nov. 6 and Nov. 7. And, on Nov. 7, students can gather in the Student Center (PDR 1 and 2) from 3 to 5 p.m. for cookies, community, and conversation. OME is holding several events, including Let’s Chat in the OME with Student Mental Health and Counseling on Nov. 6 from 5:20 to 7 p.m. and on Nov. 7 from 3:00 to 5 p.m. 

For faculty, staff, postdocs, and family members, My Life Services offers expert counseling support and resources. The LBGTQ Employee Resource Group is also available to support our employees.

Q: For many, anticipation about the outcome of the November elections is high. Can you talk about why being active, engaged citizens, as well as promoting respectful dialogue among people with different political beliefs, is important to the MIT community and a priority for the ICEO?

A: This moment offers an opportunity to recommit to our values around mutual respect, openness, and integrity at MIT. Regardless of differences in opinion, we are committed to looking after each other, whatever the issue is, in the best of times and the worst of times.

The MIT culture rests with its citizens. We are a community that includes differences, and acknowledges and leverages those differences through studying and working together, and by having conversations that bring greater understanding of each other. It makes us a better organization when we can honor our differences and learn to have civil discourse in conversations about difficult topics. I learn from you because the lens through which you see things is different from my own lens. That’s what promotes understanding.

Q: One of the ways the ICEO operates is to partner and coordinate with other offices and organizations at MIT working to further diversity, inclusion, and civic engagement. From this vantage point, can you describe some of the activities on campus in the days leading up to the election?

A: There has been a surge of activities to help raise awareness and get out the vote at MIT. The ICEO recently partnered with the LBGTQ Employee Resource Group for an event we called “ICEO Community Town Hall: Trans Rights.” The session focused on a discussion about the current state of transgender rights in Massachusetts, and featured MIT students, alumni, staff, and faculty members.

Last year, a group of undergraduate and graduate students, the chancellor’s office, registrar, and Priscilla King Gray Public (PKG) Service Center implemented TurboVote, a nonpartisan, nonprofit platform, to increase voting and voter registration, and it’s been exciting to watch the initiative expand this fall.

A new nonpartisan student group called MITvote 2018, which works closely with the PKG Service Center, has been focused on voter registration and education. Through their work this fall, they’ve registered 1,056 people, which they estimate to be 13 percent of the eligible student voter population. On Oct. 30, they will be hosting a nonpartisan explanation of Massachusetts ballot races and initiatives with student political organizations in Room 4-370 from 7 to 8:30 p.m.

And, in response to Question 3, a group of students formed “Yes on 3.” Through information sessions and study breaks, this group has been raising awareness among students that transgender protection is an issue in the Massachusetts 2018 election, and they’ve been working to register students to vote.

Pages