MIT Research News' Journal
[Most Recent Entries]
[Calendar View]
Tuesday, July 24th, 2018
| Time |
Event |
| 12:00a |
Environmental regulation in a polarized culture With an affinity for environmental issues and a knack for analysis, MIT doctoral student Parrish Bergquist aims to clarify the ways in which changing political landscapes influence environmental policy outcomes.
Bergquist’s path to doctoral research in the departments of Political Science and Urban Studies and Planning began well before she joined MIT. After graduating from the University of Virginia with a degree in American studies and English, the Birmingham, Alabama, native volunteered for two years with the U.S. Peace Corps in Honduras to study international development and policy. There, she gained a firsthand perspective on the impacts of global climate change.
“People in Honduras lived so much closer to environmental damage than we do in the U.S.,” Bergquist says. “Carbon emissions from developed countries were already starting to have an effect on [climate in that region]. ... It affects everybody.” During conversations with women and children, Bergquist learned that those who were tasked with fetching water had to walk even further with each trip to find clean water sources.
“I was struck by the extent to which industrialization had caused problems that we in the United States have buffers against feeling every single day,” she says. Her experiences inspired her to pursue a career in environmental policy, which led her to earn a master’s degree in urban planning and environmental policy from the University of Michigan. “While I was there, I decided that research was what I was really excited about,” Bergquist says; her next move was to pursue a PhD. She hopes to “gain some traction on understanding the politics behind how environmental policy decisions are made.”
Bergquist was attracted to MIT for her doctoral studies because the Department of Urban Studies and Planning integrated the study of environmental problems with urban studies, and because of the Institute’s strong political science department. “My degree is interdepartmental,” Bergquist says. “I knew when I came in that I wanted to study politics and decision making, so I knew I wanted a school that had strong political science and planning departments.”
Challenging environments
For her dissertation, Bergquist studies the implications of political polarization on environmental politics in the United States. To do this, she uses a mixed-methods approach to examine different federal- and state-level policies.
“One paper looks at whether or not elected officials from the different parties influence the way that environmental agencies enforce federal environmental laws like the Clean Air Act," Bergquist says. She examines other laws as well, such as the Clean Water Act, and the Resource Conservation and Recovery Act, through a similar lens.
She also studies how environmental public opinions change on the state level over time, and whether they have an impact on actual policy decisions. To guide her research, Bergquist starts with a question: “Do legislators from different states vote in favor of environmental legislation based on what their constituents think?”
Today’s increasing political polarization introduces not only a new challenge, but another set of questions for Bergquist's research.
“Scholars have argued that economic factors are more important than political parties and ideology in terms of shaping what states are doing for the environment,” Bergquist says. “But increasingly, every issue is really polarized across the parties — so are there places that are not as polarized for the environment now, and if so, why?”
Part of Bergquist’s research approach has been informed by courses she took early in her MIT career, including 17.150 (The American Political Economy in Comparative Perspective), taught by Kathleen Thelen, the Ford Professor of Political Science, and Devin Caughey, the Silverman Family Career Development Associate Professor of Political Science.
“The readings we did were really great, and it was a chance to think through big ideas, like how politics is structured, how politics and the economy interact, and the way that political systems develop over time,” Bergquist says. “The course really shaped the way I think about my research.”
Mentorship has also been crucial to Bergquist’s development as a scholar. “I’m grateful to have had the opportunity to take courses, teach, and collaborate with some fantastic faculty members,” she notes. Describing her work with Chris Warshaw, one of her advisors, she says: “Collaborating with Chris on a research project has been a ton of fun. Also working with him on revising, submitting, and responding to reviews on our paper has been incredibly instructive.”
Lending support
Bergquist also serves as a graduate resident tutor (GRT) at Simmons Hall, an undergraduate dorm at MIT.
“When I started at MIT, I did not expect to be living in an undergraduate dorm again. But this will be my fifth year doing it,” she says with a laugh. “It's just a really awesome community, and it's been a great way for me to feel like so much more a part of the MIT community than I otherwise would have.”
Through her GRT program, Bergquist plans frequent events for her undergraduate cohorts to foster community and lend support. “I do try to make sure that [undergraduates] feel like I'm approachable and that they could come to me if they have something going on that they need to talk about,” she says.
“I just love everything about it. I love the job and getting to know the students,” Bergquist says.
When she's not at her desk or in the dorm, Bergquist is usually exploring the environment in yet another way, by spending time outside, running, climbing, or biking.
Educating others
In the future, Bergquist hopes to continue her pursuit of academia by becoming a professor and continuing research. “I had always thought about teaching,” Bergquist says. “Part of the reason I majored in English was because I loved my English teacher in high school.”
Bergquist says that her educational journey was strongly shaped by her teachers and professors, who eventually led her to political science and planning. “Discovering those disciplines was very important to my decision to pursue an academic career,” she explains.
Through the course of her master’s degree program, her resolve to teach grew stronger: “I wanted to pursue my own creative and intellectual projects. You know who pursue their creative and intellectual projects and also teach? Professors!”
Bergquist’s ultimate goal involves a combination of scholarship, teaching, relationship-building, and the outdoors.
“I would love to get an academic job where I get to do impactful research with great colleagues and teach fantastic students,” she says. “But I recharge and refresh by spending time with people and staying active. My work is better and I’m happier when I have time to spend with the people that I care about and pursue the activities that I love to do. That’s the dream.” | | 10:00a |
Helping computers perceive human emotions MIT Media Lab researchers have developed a machine-learning model that takes computers a step closer to interpreting our emotions as naturally as humans do.
In the growing field of “affective computing,” robots and computers are being developed to analyze facial expressions, interpret our emotions, and respond accordingly. Applications include, for instance, monitoring an individual’s health and well-being, gauging student interest in classrooms, helping diagnose signs of certain diseases, and developing helpful robot companions.
A challenge, however, is people express emotions quite differently, depending on many factors. General differences can be seen among cultures, genders, and age groups. But other differences are even more fine-grained: The time of day, how much you slept, or even your level of familiarity with a conversation partner leads to subtle variations in the way you express, say, happiness or sadness in a given moment.
Human brains instinctively catch these deviations, but machines struggle. Deep-learning techniques were developed in recent years to help catch the subtleties, but they’re still not as accurate or as adaptable across different populations as they could be.
The Media Lab researchers have developed a machine-learning model that outperforms traditional systems in capturing these small facial expression variations, to better gauge mood while training on thousands of images of faces. Moreover, by using a little extra training data, the model can be adapted to an entirely new group of people, with the same efficacy. The aim is to improve existing affective-computing technologies.
“This is an unobtrusive way to monitor our moods,” says Oggi Rudovic, a Media Lab researcher and co-author on a paper describing the model, which was presented last week at the Conference on Machine Learning and Data Mining. “If you want robots with social intelligence, you have to make them intelligently and naturally respond to our moods and emotions, more like humans.”
Co-authors on the paper are: first author Michael Feffer, an undergraduate student in electrical engineering and computer science; and Rosalind Picard, a professor of media arts and sciences and founding director of the Affective Computing research group.
Personalized experts
Traditional affective-computing models use a “one-size-fits-all” concept. They train on one set of images depicting various facial expressions, optimizing features — such as how a lip curls when smiling — and mapping those general feature optimizations across an entire set of new images.
The researchers, instead, combined a technique, called “mixture of experts” (MoE), with model personalization techniques, which helped mine more fine-grained facial-expression data from individuals. This is the first time these two techniques have been combined for affective computing, Rudovic says.
In MoEs, a number of neural network models, called “experts,” are each trained to specialize in a separate processing task and produce one output. The researchers also incorporated a “gating network,” which calculates probabilities of which expert will best detect moods of unseen subjects. “Basically the network can discern between individuals and say, ‘This is the right expert for the given image,’” Feffer says.
For their model, the researchers personalized the MoEs by matching each expert to one of 18 individual video recordings in the RECOLA database, a public database of people conversing on a video-chat platform designed for affective-computing applications. They trained the model using nine subjects and evaluated them on the other nine, with all videos broken down into individual frames.
Each expert, and the gating network, tracked facial expressions of each individual, with the help of a residual network (“ResNet”), a neural network used for object classification. In doing so, the model scored each frame based on level of valence (pleasant or unpleasant) and arousal (excitement) — commonly used metrics to encode different emotional states. Separately, six human experts labeled each frame for valence and arousal, based on a scale of -1 (low levels) to 1 (high levels), which the model also used to train.
The researchers then performed further model personalization, where they fed the trained model data from some frames of the remaining videos of subjects, and then tested the model on all unseen frames from those videos. Results showed that, with just 5 to 10 percent of data from the new population, the model outperformed traditional models by a large margin — meaning it scored valence and arousal on unseen images much closer to the interpretations of human experts.
This shows the potential of the models to adapt from population to population, or individual to individual, with very few data, Rudovic says. “That’s key,” he says. “When you have a new population, you have to have a way to account for shifting of data distribution [subtle facial variations]. Imagine a model set to analyze facial expressions in one culture that needs to be adapted for a different culture. Without accounting for this data shift, those models will underperform. But if you just sample a bit from a new culture to adapt our model, these models can do much better, especially on the individual level. This is where the importance of the model personalization can best be seen.”
Currently available data for such affective-computing research isn’t very diverse in skin colors, so the researchers’ training data were limited. But when such data become available, the model can be trained for use on more diverse populations. The next step, Feffer says, is to train the model on “a much bigger dataset with more diverse cultures.”
Better machine-human interactions
Another goal is to train the model to help computers and robots automatically learn from small amounts of changing data to more naturally detect how we feel and better serve human needs, the researchers say.
It could, for example, run in the background of a computer or mobile device to track a user’s video-based conversations and learn subtle facial expression changes under different contexts. “You can have things like smartphone apps or websites be able to tell how people are feeling and recommend ways to cope with stress or pain, and other things that are impacting their lives negatively,” Feffer says.
This could also be helpful in monitoring, say, depression or dementia, as people’s facial expressions tend to subtly change due to those conditions. “Being able to passively monitor our facial expressions,” Rudovic says, “we could over time be able to personalize these models to users and monitor how much deviations they have on daily basis — deviating from the average level of facial expressiveness — and use it for indicators of well-being and health.”
A promising application, Rudovic says, is human-robotic interactions, such as for personal robotics or robots used for educational purposes, where the robots need to adapt to assess the emotional states of many different people. One version, for instance, has been used in helping robots better interpret the moods of children with autism.
Roddy Cowie, professor emeritus of psychology at the Queen’s University Belfast and an affective computing scholar, says the MIT work “illustrates where we really are” in the field. “We are edging toward systems that can roughly place, from pictures of people’s faces, where they lie on scales from very positive to very negative, and very active to very passive,” he says. “It seems intuitive that the emotional signs one person gives are not the same as the signs another gives, and so it makes a lot of sense that emotion recognition works better when it is personalized. The method of personalizing reflects another intriguing point, that it is more effective to train multiple ‘experts,’ and aggregate their judgments, than to train a single super-expert. The two together make a satisfying package.” | | 2:52p |
3Q: Richard Milner on a new U.S. particle accelerator The case for an ambitious new particle accelerator to be built in the United States has just gotten a major boost.
Today, the National Academies of Sciences, Engineering, and Medicine have endorsed the development of the Electron Ion Collider, or EIC. The proposed facility, consisting of two intersecting accelerators, would smash together beams of protons and electrons traveling at nearly the speed of light. In the aftermath of each collision, scientists should see “snapshots” of the particles’ inner structures, much like a CT scan for atoms. From these images, scientists hope to piece together a multidimensional picture, with unprecedented depth and clarity, of the quarks and gluons that bind together protons and all the visible matter in the universe.
The EIC, if built, would significantly advance the field of quantum chromodynamics, which seeks to answer fundamental questions in physics, such as how quarks and gluons produce the strong force — the “glue” that holds all matter together. If constructed, the EIC would be the largest accelerator facility in the U.S. and, worldwide, second only to the Large Hadron Collider at CERN. MIT physicists, including Richard Milner, professor of physics at MIT, have been involved from the beginning in making the case for the EIC.
MIT News checked in with Milner, a member of the Laboratory for Nuclear Science, about the need for a new particle collider and its prospects going forward.
Q: Tell us a bit about the history of this design. What has it taken to make the case for this new particle accelerator?
A: The development of both the scientific and technical case for the EIC has been in progress for about two decades. With the development of quantum chromodynamics (QCD) in the 1970s by MIT physics Professor Frank Wilczek and others, nuclear physicists have long sought to bridge the gap between QCD and the successful theory of nuclei based on experimentally observable particles, where the fundamental constituents are the undetectable quarks and gluons.
A high-energy collider with the ability to collide electrons with the full range of nuclei at high rates and to have the electrons and nucleons polarized was identified as the essential tool to construct this bridge. High-energy electron scattering from the proton was how quarks were experimentally discovered at SLAC in the late 1960s (by MIT physics faculty Henry Kendall and Jerome Friedman and colleagues), and it is the accepted technique to directly probe the fundamental quark and gluon structure of matter.
Significant initial impetus for the EIC came from nuclear physicists at the university user-facilities at the University of Indiana and MIT as well as from physicists seeking to understand the origin of the proton’s spin, at laboratories and universities in the U.S. and Europe. Over the last three long-range planning exercises by U.S. nuclear physicists in 2002, 2007, and 2015, the case for the EIC has matured and strengthened. After the 2007 exercise, the two U.S. flagship nuclear facilities, namely the Relativistic Heavy Ion Collider at Brookhaven National Laboratory and the Continuous Electron Beam Accelerator Facility at Jefferson Laboratory, took a leadership role in coordinating EIC activities across the broad U.S. QCD community. This led to the production in 2012 of a succinct summary of the science case, “Electron-Ion Collider: The Next QCD Frontier (Understanding the glue that binds us all).”
The 2015 planning exercise established the EIC as the highest priority for new facility construction in U.S. nuclear physics after present commitments are fulfilled. This led to the formation of a committee by the U.S. National Academy of Sciences (NAS) to assess the EIC science case. The NAS committee deliberated for about a year and the report has been publicly released this month.
Q: Give us an idea of how powerful this new collider will be and what kind of new interactions it will produce. What kinds of phenomena will it help to explain?
A: The EIC will be a powerful and unique new accelerator that will offer an unprecedented window into the fundamental structure of matter. The electron-ion collision rate at the EIC will be high, more than two orders of magnitude greater than was possible at the only previous electron-proton collider, namely HERA, which operated at the DESY laboratory in Hamburg, Germany, from 1992 to 2007. With the EIC, physicists will be able to image the virtual quarks and gluons that make up protons, neutrons, and nuclei, with unprecedented spatial resolution and shutter speed. A goal is to provide images of the fundamental structure of the microcosm that can be appreciated broadly by humanity: to answer questions such as, what does a proton look like? And what does a nucleus look like?
There are three central scientific issues that can be addressed by an electron-ion collider. The first goal is to understand in detail the mechanisms within QCD by which the mass of protons and neutrons, and thus the mass of all the visible matter in the universe, is generated. The problem is that while gluons have no mass, and quarks are nearly massless, the protons and neutrons that contain them are heavy, making up most of the visible mass of the universe. The total mass of a nucleon is some 100 times greater than the mass of the various quarks it contains.
The second issue is to understand the origin of the intrinsic angular momentum, or spin, of nucleons, a fundamental property that underlies many practical applications, including magnetic resonance imaging (MRI). How the angular momentum, both intrinsic as well as orbital, of the internal quarks and gluons gives rise to the known nucleon spin is not understood. And thirdly, the nature of gluons in matter — that is, their arrangements or states — and the details of how they hold matter together, is not well-known. Gluons in matter are a little like dark matter in the universe: unseen but playing a crucial role. An electron-ion collider would potentially reveal new states resulting from the close packing of many gluons within nucleons and nuclei. These issues are fundamental to our understanding of the matter in the universe.
Q: What role will MIT have in this project going forward?
A: At present, more than a dozen MIT physics department faculty lead research groups in the Laboratory for Nuclear Science that work directly on understanding the fundamental structure of matter as described by QCD. It is the largest university-based group in the U.S. working on QCD. Theoretical research is focused at the Center for Theoretical Physics, and experimentalists rely heavily on the Bates Research and Engineering Center for technical support.
MIT theorists are carrying out important calculations using the world’s most powerful computers to understand fundamental aspects of QCD. MIT experimental physicists are conducting experiments at existing facilities, such as BNL, CERN, and Jefferson Laboratory, to reach new insight and to develop new techniques that will be used at the EIC. Further, R&D into new polarized sources, detectors, and innovative data-acquisition schemes by MIT scientists and engineers is in progress. It is anticipated that these efforts will ramp up as the realization of the EIC approaches.
It is anticipated that the U.S. Department of Energy Office of Science will initiate in the near future the official process for EIC by which the U.S. government approves, funds, and constructs new, large scientific facilities. Critical issues are the selection of the site for EIC and the participation of international users. An EIC user group has formed with the participation of more than 700 PhD scientists from over 160 laboratories and universities around the world. If the realization of EIC follows a schedule comparable to that of past large facilities, it should be doing science by about 2030. MIT has a long history of providing leadership in U.S. nuclear physics and will continue to play a significant role as we proceed along the path to EIC. |
|