Helping nonexperts build advanced generative AI models

The impact of artificial intelligence will never be equitable if there’s only one company that builds and controls the models (not to mention the data that go into them). Unfortunately, today’s AI models are made up of billions of parameters that must be trained and tuned to maximize performance for each use case, putting the most powerful AI models out of reach for most people and companies.

MosaicML started with a mission to make those models more accessible. The company, which counts Jonathan Frankle PhD ’23 and MIT Associate Professor Michael Carbin as co-founders, developed a platform that let users train, improve, and monitor open-source models using their own data. The company also built its own open-source models using graphical processing units (GPUs) from Nvidia.

The approach made deep learning, a nascent field when MosaicML first began, accessible to far more organizations as excitement around generative AI and large language models (LLMs) exploded following the release of Chat GPT-3.5. It also made MosaicML a powerful complementary tool for data management companies that were also committed to helping organizations make use of their data without giving it to AI companies.

Last year, that reasoning led to the acquisition of MosaicML by Databricks, a global data storage, analytics, and AI company that works with some of the largest organizations in the world. Since the acquisition, the combined companies have released one of the highest performing open-source, general-purpose LLMs yet built. Known as DBRX, this model has set new benchmarks in tasks like reading comprehension, general knowledge questions, and logic puzzles.

Since then, DBRX has gained a reputation for being one of the fastest open-source LLMs available and has proven especially useful at large enterprises.

More than the model, though, Frankle says DBRX is significant because it was built using Databricks tools, meaning any of the company’s customers can achieve similar performance with their own models, which will accelerate the impact of generative AI.

“Honestly, it’s just exciting to see the community doing cool things with it,” Frankle says. “For me as a scientist, that’s the best part. It’s not the model, it’s all the amazing stuff the community is doing on top of it. That's where the magic happens.”

Making algorithms efficient

Frankle earned bachelor’s and master’s degrees in computer science at Princeton University before coming to MIT to pursue his PhD in 2016. Early on at MIT, he wasn't sure what area of computing he wanted to study. His eventual choice would change the course of his life.

Frankle ultimately decided to focus on a form of artificial intelligence known as deep learning. At the time, deep learning and artificial intelligence did not inspire the same broad excitement as they do today. Deep learning was a decades-old area of study that had yet to bear much fruit.

“I don’t think anyone at the time anticipated deep learning was going to blow up in the way that it did,” Frankle says. “People in the know thought it was a really neat area and there were a lot of unsolved problems, but phrases like large language model (LLM) and generative AI weren’t really used at that time. It was early days.”

Things began to get interesting with the 2017 release of a now-infamous paper by Google researchers, in which they showed a new deep-learning architecture known as the transformer was surprisingly effective as language translation and held promise across a number of other applications, including content generation.

In 2020, eventual Mosaic co-founder and tech executive Naveen Rao emailed Frankle and Carbin out of the blue. Rao had read a paper the two had co-authored, in which the researchers showed a way to shrink deep-learning models without sacrificing performance. Rao pitched the pair on starting a company. They were joined by Hanlin Tang, who had worked with Rao on a previous AI startup that had been acquired by Intel.

The founders started by reading up on different techniques used to speed up the training of AI models, eventually combining several of them to show they could train a model to perform image classification four times faster than what had been achieved before.

“The trick was that there was no trick,” Frankle says. “I think we had to make 17 different changes to how we trained the model in order to figure that out. It was just a little bit here and a little bit there, but it turns out that was enough to get incredible speed-ups. That’s really been the story of Mosaic.”

The team showed their techniques could make models more efficient, and they released an open-source large language model in 2023 along with an open-source library of their methods. They also developed visualization tools to let developers map out different experimental options for training and running models.

MIT’s E14 Fund invested in Mosaic’s Series A funding round, and Frankle says E14’s team offered helpful guidance early on. Mosaic’s progress enabled a new class of companies to train their own generative AI models.

“There was a democratization and an open-source angle to Mosaic’s mission,” Frankle says. “That’s something that has always been very close to my heart. Ever since I was a PhD student and had no GPUs because I wasn’t in a machine learning lab and all my friends had GPUs. I still feel that way. Why can’t we all participate? Why can’t we all get to do this stuff and get to do science?”

Open sourcing innovation

Databricks had also been working to give its customers access to AI models. The company finalized its acquisition of MosaicML in 2023 for a reported $1.3 billion.

“At Databricks, we saw a founding team of academics just like us,” Frankle says. “We also saw a team of scientists who understand technology. Databricks has the data, we have the machine learning. You can't do one without the other, and vice versa. It just ended up being a really good match.”

In March, Databricks released DBRX, which gave the open-source community and enterprises building their own LLMs capabilities that were previously limited to closed models.

“The thing that DBRX showed is you can build the best open-source LLM in the world with Databricks,” Frankle says. “If you’re an enterprise, the sky’s the limit today.”

Frankle says Databricks’ team has been encouraged by using DBRX internally across a wide variety of tasks.

“It’s already great, and with a little fine-tuning it’s better than the closed models,” he says. “You’re not going be better than GPT for everything. That’s not how this works. But nobody wants to solve every problem. Everybody wants to solve one problem. And we can customize this model to make it really great for specific scenarios.”

As Databricks continues pushing the frontiers of AI, and as competitors continue to invest huge sums into AI more broadly, Frankle hopes the industry comes to see open source as the best path forward.

“I’m a believer in science and I’m a believer in progress and I’m excited that we’re doing such exciting science as a field right now,” Frankle says. “I’m also a believer in openness, and I hope that everybody else embraces openness the way we have. That's how we got here, through good science and good sharing.”

de MIT News https://ift.tt/Ng2GClB

New Ragon Institute building opens in the heart of Kendall Square

Leaders from MIT, Harvard University, and Mass General Brigham gathered Monday to celebrate an important new chapter in the Ragon Institute’s quest to harness the immune system to prevent and cure human diseases.

The ceremony marked the opening of the new building for the Ragon Institute of Mass General, MIT, and Harvard, located at 600 Main Street in the heart of Cambridge’s Kendall Square, where its multidisciplinary group of researchers will expand on the collaborations that have proven impactful since the Institute’s founding in 2009.

“Fifteen years ago, the Ragon Institute started with transformative philanthropy from Terry and Susan Ragon,” Ragon Institute Director and MIT professor of the practice Bruce Walker said. “Initially, it was an experiment: Could we bring together scientists, engineers, and medical doctors to pool their creative knowledge and cross-disciplinary specialties to make advances against the greatest global health problems of our time? Now, 15 years later, here we are celebrating the success of that experiment and welcoming the next phase of the Ragon Institute.”

The institute’s new building features five floors of cutting-edge, dedicated lab space and more than double the floor area of the previous facilities. The open, centralized layout of the new building is designed to empower cross-disciplinary research and enable discoveries that will lead to new ways to prevent, detect, and cure diseases. The expanded space will also allow the Ragon Institute to bring in more scientists, researchers, biologists, clinicians, postdocs, and operational staff.

“Cross-disciplinary collaboration is a hallmark of the Ragon Institute, and that is really how you do transformational research and breakthrough science at scale — what everyone talks about but few actually achieve,” said Mass General Brigham President and CEO Anne Klibanski. “Partnerships between health care and academia accelerate these breakthroughs and foster innovation. That is the model of scientific discovery this whole area represents, that Boston and Massachusetts represent, and that this institute represents.”

In addition to state-of-the-art lab space, a third of the new building is open for public use. The Ragon Institute’s leaders expressed a commitment to engaging with the local Cambridge community and believe the institute’s success will further strengthen Kendall Square’s innovation ecosystem.

“As a relative newcomer, I see this elegant new building as an inspiring vote of confidence in the future of Kendall Square,” MIT President Sally Kornbluth said. “I gather that over a few decades, thanks in part to many of you here today, Kendall Square was transformed from a declining postindustrial district to the center of a region that is arguably the biotech capital of the world. I believe we now have an opportunity to secure its future, to make sure Kendall Square becomes an infinitely self-renewing source of biomedical progress, a limitless creative pool perpetually refreshed by a stream of new ideas from every corner of the life sciences and engineering to unlock solutions to the most important problems of our time. This building and this institute embody that vision.”

The Ragon Institute is a collaborative effort of Mass General Brigham, MIT, and Harvard. It was founded in 2009 through support from the Phillip T. and Susan M. Ragon Foundation with the initial goal of developing an HIV vaccine. Since then, it has expanded to focus on other global health initiatives — from playing a vital role in Covid-19 vaccine development to exploring the rising health challenges of climate change and preparing for the next pandemic.

The institute strives to break down siloes between scientists, engineers, and clinicians from diverse disciplines to apply all available knowledge to the fight against diseases of global importance.

During the ceremony, Phillip (Terry) Ragon ’72 discussed the origins of the Institute and his vision for accelerating scientific discovery.

“With Bruce [Walker], I began to see how philanthropy could really make a difference and how we could power a different model that we thought could be particularly effective,” Ragon said. “The fundamental idea was to take an approach like the Manhattan Project, bringing the best and brightest people together from different disciplines, with flexible funding, and leave them to be successful. And so here we are today.”

Ragon Institute faculty are engaged in challenges as varied as developing vaccines for tuberculosis and HIV, cures for malaria, treatments for neuroimmunological diseases, a universal flu vaccine, and therapies for cancer and autoimmune disorders — with the potential to impact billions of lives.

The new building’s opening followed additional funding from Terry and Susan Ragon, which came in recognition of the Ragon Institute’s expanding mission.

“[Through this partnership], we’ve accomplished more than we realized we could, and that’s shown in the scientific progress that the Ragon Institute has achieved,” said Harvard University interim president Alan Garber. “To pull this off requires not only scientific brilliance, but true leadership.”

Walker, the Institute’s founding director, has spent his entire career caring for people living with HIV and studying how the body fights back. He has helped establish two cutting-edge research institutes in Africa, which continue to train the next generation of African scientists. The international reach of the Ragon Institute is another aspect that sets it apart in its mission to impact human health.

“Today we launch the next 100 years of the Ragon Institute, and we’re fortunate to work every day on this enormously challenging and consistently inspiring mission,” Walker said. “We’re motivated by the belief that every day matters, that our efforts will ultimately alleviate suffering, that our mission is urgent, and that together, we will succeed.”

de MIT News https://ift.tt/dB1b8DQ

Toward socially and environmentally responsible real estate

The MIT student of popular imagination is a Tony Stark or a Riri Williams working in a lab and building the technology of the future. Not necessarily someone studying real estate.

Peggy Ghasemlou is doing just that, however, and she’s traveled over thousands of miles and jumped through about as many hoops to do it.

A licensed architect in her hometown of Tehran, Iran’s capital, Ghasemlou enrolled at MIT to pursue her interests in sustainability and inclusion in the fields of architecture and real estate development. Now, after managing visa and travel issues that required her own superhero-like determination, she’s halfway through earning a master of science in real estate development (MSRED) from the School of Architecture and Planning’s Center for Real Estate (CRE). This fall, she will be working with lecturer Jacques Gordon, CRE’s former “executive-in-practice,” on a thesis involving portfolio management.

Throughout her time at MIT, Ghasemlou has enjoyed her program’s balance of economics, technology, sustainability, and design. She says the curriculum has supported and challenged her in equal measure, but above all, she appreciates the program’s emphasis on financial, social, and environmental responsibility.

“I’m so grateful that I chose MSRED, because they are not just thinking about how to make more money,” she says. “They are teaching us about how to make a lasting positive impact.”

It hasn’t been an easy journey. Visa issues, scholarship rejections, and thousands of miles stood between her and MIT, and the challenges didn’t end when she did get to campus, halfway around the world from her home and family. She beat all those odds, however, and is ready for whatever the future brings.

“When I first arrived here, I had three main feelings: relief, hope, and doubt,” she said. “Now, I am just feeling grateful for my time here and the friendships I have made.”

From design to ownership

While growing up, Ghasemlou loved design “from the start.” That affinity led her to pursue a bachelor’s in architectural engineering, followed by a master’s in digital engineering with a focus on sustainability.

She first made serious contact with MIT while pursuing her master’s, taking the Institute’s online courses to help her with her thesis on zero-energy buildings. She chose both the thesis and the classes out of a desire to “do something positive and impactful” and learned how to use tools to optimize a building’s energy efficiency, among other important measures.

After she earned her master’s, she spent the next five years designing and developing residential buildings for a studio in Tehran. The experience sparked her interest in the financial side of architecture and real estate, and along with it, the intersection of sustainability, economics, and design — areas encompassed by MSRED’s curriculum.

She decided to apply, and was also awarded the Goldie B. Wolfe Miller Women Leaders in Real Estate scholarship.

“The Goldie Initiative is the most supportive community,” she says. “They’re the best thing that’s happen[ed] to me in the U.S. They really care about you, and they really want, in their heart, to help you.”

With women underrepresented in the real estate fields, particularly at leadership levels, awards like this emphasize both the progress that has been made as well as the work that is yet to be done. In Tehran, Ghasemlou founded Girls in Real Estate Development (GIRD), to introduce the fields of architecture and real estate to young women and help create career pathways for these traditionally male-dominated professions.

“I really love to see women being in decision-making positions and to be able to influence different industries in meaningful ways,” she said. “Whatever I learn, I try to [pass along to the next generation]. It might have a small impact on them, but I tell them, ‘If I can do it, you can do it.’”

Once she made it to MIT for her first semester, she took finance and economics courses, which were new subjects for her. Adjusting to a new environment was also jarring, but she credited her classmates and professors for being “incredibly supportive” and helping her “not feel so isolated.”

Her second semester featured sustainability courses — a friendlier prospect, given her background in design — and helped point her in the direction of sustainable portfolio management for her thesis topic.

However, enrolling at MIT was one thing. Actually getting to campus was another.

The long and winding road

Rewind back to last summer. Once the excitement of being accepted to the MSRED program wore off, reality set in. Like other international students, Ghasemlou had to apply for a visa. She did so through the U.S. embassy in Turkey’s capital, Ankara, and began the waiting game. Days turned into weeks, however, so she decided to try her luck with a different embassy and packed her bags for Toronto.

With the start of classes only weeks away, she made the decision to wait it out in the Canadian metropolis. She ended up having to take online classes during the beginning of the semester, but right on the day she “lost all hope,” her visa was finally issued.

In Ankara, that is.

She had already flown over 6,000 miles just to get from Tehran to Toronto, and she was now staring down the barrel of a 10,000-mile-plus trip to go back to Turkey for her visa and then get to MIT’s campus, all while the semester was kicking into gear. That may have been too daunting a prospect for some, but not for her.

“I calculated the hours I was in the airport and airplane: over 30 hours,” she said. “I arrived in Boston, I remember, at 11:30 p.m., then I just thought, ‘Tomorrow, I should go to my classes.’”

Luckily, her family supported her throughout the process.

“I’m so thankful for my parents and my brother — especially my brother — because he believes in me all the time,” she said. “That really helped me go through all the hard times I had to go through to be here.”

Now that she is here, she’s got a lot of big ideas for the future of housing, sustainability, and real estate. She’ll be spending the summer with a Boston-based nonprofit called Preservation of Affordable Housing, assessing units for sustainability goals and updating sustainability criteria.

Going forward, she expressed an interest in staying in Boston long-term, noting its potential to join other cities in becoming “one of the leaders in sustainability.” She’s a believer in policy for effective change-making, and cites New York City’s Local Law 97 (LL97), which requires that large buildings meet certain limits regarding energy efficiency and greenhouse gas emissions, as an example of a law that is “not just a policy” but also makes people think about the city around them.

Ghasemlou also aims to continue to support other women in the real estate fields, and expresses admiration for female industry leaders such as Fidelity’s Suzanne Heidelberger.

“When I see successful women in this industry, I feel inspired and proud of them,” she said. “I really want to see more and more female leaders in the industry.”

de MIT News https://ift.tt/eO7c3Yt

Eric Evans receives Department of Defense Medal for Distinguished Public Service

On May 31, the U.S. Department of Defense's chief technology officer, Under Secretary of Defense for Research and Engineering Heidi Shyu, presented Eric Evans with the Department of Defense (DoD) Medal for Distinguished Public Service. This award is the highest honor given by the secretary of defense to private citizens for their significant service to the DoD. Evans was selected for his leadership as director of MIT Lincoln Laboratory and as vice chair and chair of the Defense Science Board (DSB).

"I have gotten to know Eric well in the last three years, and I greatly appreciate his leadership, proactiveness, vision, intellect, and humbleness," Shyu stated in her remarks during the May 31 ceremony held at the laboratory. "Eric has a willingness and ability to confront and solve the most difficult problems for national security. His distinguished public service will continue to have invaluable impacts on the department and the nation for decades to come." 

During his tenure in both roles over more than a decade, Evans has cultivated relationships at the highest levels within the DoD. Since stepping into his role as laboratory director in 2006, he has advised eight defense secretaries and seven deputy defense secretaries. Under his leadership, the laboratory delivered advanced capabilities for national security in a broad range of technology areas, including cybersecurity, space surveillance, biodefense, artificial intelligence, laser communications, and quantum computing.

Evans ensured that the laboratory addressed not only existing DoD priorities, but also emerging and future threats. He foresaw the need for and established three new technical divisions covering Cyber Security and Information Sciences, Homeland Protection, and Biotechnology and Human Systems. When the Covid-19 pandemic struck, he quickly pivoted the laboratory to aid the national response. To ensure U.S. competitiveness in an ever-evolving defense landscape, he advocated for the modernization of major test ranges, including the Reagan Test Site for which the laboratory serves as scientific advisor, and secured funding for new state-of-the-art facilities such as the Compound Semiconductor Laboratory – Microsystem Integration Facility. He also strengthened ties with MIT campus on research collaborations to drive innovation and expand educational opportunities for preparing the next generation of the DoD STEM workforce.

In parallel, Evans served on the DSB, the leading board for providing science and technology advice to DoD senior leadership. Evans served as DSB vice chair from 2014 to 2020 and chair since 2020. Over the years, Evans led or supported more than 30 DSB studies of direct importance to the DoD. Most notably, he initiated a new Strategic Options Permanent Subcommittee focused on identifying systems and technology to prepare the nation for future defense needs.

“The medal is a wonderful and richly deserved recognition of Eric’s contributions to MIT and to national security,” said Ian Waitz, MIT’s vice president for research.

As Evans steps down from his role as Lincoln Laboratory director on July 1, he will transition to a joint appointment as a senior fellow and professor of practice appointment on the MIT campus and as a fellow in the Director's Office at Lincoln Laboratory. In these new roles, he will continue to strengthen ties between the laboratory and MIT campus and work with DoD leaders.

de MIT News https://ift.tt/wzIHDOY

Study: Titan’s lakes may be shaped by waves

Titan, Saturn’s largest moon, is the only planetary body in the solar system besides our own that currently hosts active rivers, lakes, and seas. Titan’s otherworldly river systems are thought to be filled with liquid methane and ethane that flows into wide lakes and seas, some as large as the Great Lakes on Earth.

The existence of Titan’s large seas and smaller lakes was confirmed in 2007, with images taken by NASA’s Cassini spacecraft. Since then, scientists have pored over those and other images for clues to the moon’s mysterious liquid environment.

Now, MIT geologists have studied Titan’s shorelines and shown through simulations that the moon’s large seas have likely been shaped by waves. Until now, scientists have found indirect and conflicting signs of wave activity, based on remote images of Titan’s surface.

The MIT team took a different approach to investigate the presence of waves on Titan, by first modeling the ways in which a lake can erode on Earth. They then applied their modeling to Titan’s seas to determine what form of erosion could have produced the shorelines in Cassini’s images. Waves, they found, were the most likely explanation.

The researchers emphasize that their results are not definitive; to confirm that there are waves on Titan will require direct observations of wave activity on the moon’s surface.

“We can say, based on our results, that if the coastlines of Titan’s seas have eroded, waves are the most likely culprit,” says Taylor Perron, the Cecil and Ida Green Professor of Earth, Atmospheric and Planetary Sciences at MIT. “If we could stand at the edge of one of Titan’s seas, we might see waves of liquid methane and ethane lapping on the shore and crashing on the coasts during storms. And they would be capable of eroding the material that the coast is made of.”

Perron and his colleagues, including first author Rose Palermo PhD ’22, a former MIT-WHOI Joint Program graduate student and current research geologist at the U.S. Geological Survey, have published their study today in Science Advances. Their co-authors include MIT Research Scientist Jason Soderblom; former MIT postdoc Sam Birch, now an assistant professor at Brown University; Andrew Ashton at the Woods Hole Oceanographic Institution; and Alexander Hayes of Cornell University.

“Taking a different tack”

The presence of waves on Titan has been a somewhat controversial topic ever since Cassini spotted bodies of liquid on the moon’s surface.

“Some people who tried to see evidence for waves didn’t see any, and said, ‘These seas are mirror-smooth,’” Palermo says. “Others said they did see some roughness on the liquid surface but weren’t sure if waves caused it.”

Knowing whether Titan’s seas host wave activity could give scientists information about the moon’s climate, such as the strength of the winds that could whip up such waves. Wave information could also help scientists predict how the shape of Titan’s seas might evolve over time.

Rather than look for direct signs of wave-like features in images of Titan, Perron says the team had to “take a different tack, and see, just by looking at the shape of the shoreline, if we could tell what’s been eroding the coasts.”

Titan’s seas are thought to have formed as rising levels of liquid flooded a landscape crisscrossed by river valleys. The researchers zeroed in on three scenarios for what could have happened next: no coastal erosion; erosion driven by waves; and “uniform erosion,” driven either by “dissolution,” in which liquid passively dissolves a coast’s material, or a mechanism in which the coast gradually sloughs off under its own weight.

The researchers simulated how various shoreline shapes would evolve under each of the three scenarios. To simulate wave-driven erosion, they took into account a variable known as “fetch,” which describes the physical distance from one point on a shoreline to the opposite side of a lake or sea.

“Wave erosion is driven by the height and angle of the wave,” Palermo explains. “We used fetch to approximate wave height because the bigger the fetch, the longer the distance over which wind can blow and waves can grow.”

To test how shoreline shapes would differ between the three scenarios, the researchers started with a simulated sea with flooded river valleys around its edges. For wave-driven erosion, they calculated the fetch distance from every single point along the shoreline to every other point, and converted these distances to wave heights. Then, they ran their simulation to see how waves would erode the starting shoreline over time. They compared this to how the same shoreline would evolve under erosion driven by uniform erosion. The team repeated this comparative modeling for hundreds of different starting shoreline shapes.

They found that the end shapes were very different depending on the underlying mechanism. Most notably, uniform erosion produced inflated shorelines that widened evenly all around, even in the flooded river valleys, whereas wave erosion mainly smoothed the parts of the shorelines exposed to long fetch distances, leaving the flooded valleys narrow and rough.

“We had the same starting shorelines, and we saw that you get a really different final shape under uniform erosion versus wave erosion,” Perron says. “They all kind of look like the Flying Spaghetti Monster because of the flooded river valleys, but the two types of erosion produce very different endpoints.”

The team checked their results by comparing their simulations to actual lakes on Earth. They found the same difference in shape between Earth lakes known to have been eroded by waves and lakes affected by uniform erosion, such as dissolving limestone.

A shore’s shape

Their modeling revealed clear, characteristic shoreline shapes, depending on the mechanism by which they evolved. The team then wondered: Where would Titan’s shorelines fit, within these characteristic shapes?

In particular, they focused on four of Titan’s largest, most well-mapped seas: Kraken Mare, which is comparable in size to the Caspian Sea; Ligeia Mare, which is larger than Lake Superior; Punga Mare, which is longer than Lake Victoria; and Ontario Lacus, which is about 20 percent the size of its terrestrial namesake.

The team mapped the shorelines of each Titan sea using Cassini’s radar images, and then applied their modeling to each of the sea’s shorelines to see which erosion mechanism best explained their shape. They found that all four seas fit solidly in the wave-driven erosion model, meaning that waves produced shorelines that most closely resembled Titan’s four seas.

“We found that if the coastlines have eroded, their shapes are more consistent with erosion by waves than by uniform erosion or no erosion at all,” Perron says.

Juan Felipe Paniagua-Arroyave, associate professor in the School of Applied Sciences and Engineering at EAFIT University in Colombia, says the team’s results are “unlocking new avenues of understanding.”

“Waves are ubiquitous on Earth’s oceans. If Titan has waves, they would likely dominate the surface of lakes,” says Paniagua-Arroyave, who was not involved in the study. ”It would be fascinating to see how Titan’s winds create waves, not of water, but of exotic liquid hydrocarbons.”The researchers are working to determine how strong Titan’s winds must be in order to stir up waves that could repeatedly chip away at the coasts. They also hope to decipher, from the shape of Titan’s shorelines, from which directions the wind is predominantly blowing.

“Titan presents this case of a completely untouched system,” Palermo says. “It could help us learn more fundamental things about how coasts erode without the influence of people, and maybe that can help us better manage our coastlines on Earth in the future.”

This work was supported, in part, by NASA, the National Science Foundation, the U.S. Geological Survey, and the Heising-Simons Foundation.

de MIT News https://ift.tt/Q6BA4Im

3 Questions: Catherine D’Ignazio on data science and a quest for justice

As long as we apply data science to society, we should remember that our data may have flaws, biases, and absences. That is one motif of MIT Associate Professor Catherine D’Ignazio’s new book, “Counting Feminicide,” published this spring by the MIT Press. In it, D’Ignazio explores the world of Latin American activists who began using media accounts and other sources to tabulate how many women had been killed in their countries as the result of gender-based violence — and found that their own numbers differed greatly from official statistics.

Some of these activists have become prominent public figures, and others less so, but all of them have produced work providing lessons about collecting data, sharing it, and applying data to projects supporting human liberty and dignity. Now, their stories are reaching a new audience thanks to D’Ignazio, an associate professor of urban science and planning in MIT’s Department of Urban Studies and Planning, and director of MIT’s Data and Feminism Lab. She is also hosting an ongoing, transnational book club about the work. MIT News spoke with D’Ignazio about the new book and how activists are expanding the traditional practice of data science.

Q: What is your book about?

A: Three things. It’s a book that documents the rise of data activism as a really interesting form of citizen data science. Increasingly, because of the availability of data and tools, gathering and doing your own data analysis is a growing form of social activism. We characterize it in the book as a citizenship practice. People are using data to make knowledge claims and put political demands out there for their institutions to respond to.

Another takeaway is that from observing data activists, there are ways they approach data science that are very different from how it’s usually taught. Among other things, when undertaking work about inequality and violence, there’s a connection with the rows of data. It’s about memorializing people who have been lost. Mainstream data scientists can learn a lot from this.

The third thing is about feminicide itself and missing information. The main reason people start collecting data about feminicide is because their institutions aren’t doing it. This includes our institutions here in the United States. We’re talking about violence against women that the state is neglecting to count, classify, or take action on. So, activists step into these gaps and do this to the best of their ability, and they have been quite effective. The media will go to the activists, who end up becoming authorities on feminicide.

Q: Can you elaborate on the differences between the practices of these data activists and more standard data science?

A: One difference is what I’ll call the intimacy and proximity to the rows of the data set. In conventional data science, when you’re analyzing data, typically you’re not also the data collector. However these activists and groups are involved across the entire pipeline. As a result, there’s a connection and humanization to each line of the data set. For example, there is a school nurse in Texas who runs the site Women Count USA, and she will spend many hours trying to find photographs of victims of feminicide, which represents unusual care paid to each row of a dataset.

Another point is the sophistication that the data activists have around what their data represent and what the biases are in the data. In mainstream AI and data science, we’re still having conversations where people seem surprised that there is bias in datasets. But I was impressed with the critical sophistication with which the activists approached their data. They gather information from the media and are familiar with the biases media have, and are aware their data is not comprehensive but is still useful. We can hold those two things together. It’s often more comprehensive data than what the institutions themselves have or will release to the public.

Q: You did not just chronicle the work of activists, but engaged with them as well, and report about that in the book. What did you work on with them?

A: One big component in the book is the participatory technology development that we engaged in with the activists, and one chapter is a case study of our work with activists to co-design machine learning and AI technology that supports their work. Our team was brainstorming about a system for the activists that would automatically find cases, verify them, and put them right in the database. Interestingly, the activists pushed back on that. They did not want full automation. They felt being, in effect, witnesses is an important part of the work. The emotional burden is an important part of the work and very central to it, too. That’s not something I might always expect to hear from data scientists.

Keeping the human in the loop also means the human makes the final decision over whether a specific item constitutes feminicide or not. Handling it like that aligns with the fact that there are multiple definitions of feminicide, which is a complicated thing from a computational perspective. The proliferation of definitions about what counts as feminicide is a reflection of the fact that this is an ongoing global, transnational conversation. Feminicide has been codified in many laws, especially in Latin American countries, but none of those single laws is definitive. And no single activist definition is definitive. People are creating this together, through dialogue and struggle, so any computational system has to be designed with that understanding of the democratic process in mind.

de MIT News https://ift.tt/9odw0cS

Arvind, longtime MIT professor and prolific computer scientist, dies at 77

Arvind Mithal, the Charles W. and Jennifer C. Johnson Professor in Computer Science and Engineering at MIT, head of the faculty of computer science in the Department of Electrical Engineering and Computer Science (EECS), and a pillar of the MIT community, died on June 17. Arvind, who went by the mononym, was 77 years old.

A prolific researcher who led the Computation Structures Group in the Computer Science and Artificial Intelligence Laboratory (CSAIL), Arvind served on the MIT faculty for nearly five decades.

As a scientist, Arvind was well known for important contributions to dataflow computing, which seeks to optimize the flow of data to take advantage of parallelism, achieving faster and more efficient computation.

In the last 25 years, his research interests broadened to include developing techniques and tools for formal modeling, high-level synthesis, and formal verification of complex digital devices like microprocessors and hardware accelerators, as well as memory models and cache coherence protocols for parallel computing architectures and programming languages.

Those who knew Arvind describe him as a rare individual whose interests and expertise ranged from high-level, theoretical formal systems all the way down through languages and compilers to the gates and structures of silicon hardware.

The applications of Arvind’s work are far-reaching, from reducing the amount of energy and space required by data centers to streamlining the design of more efficient multicore computer chips.

“Arvind was both a tremendous scholar in the fields of computer architecture and programming languages and a dedicated teacher, who brought systems-level thinking to our students. He was also an exceptional academic leader, often leading changes in curriculum and contributing to the Engineering Council in meaningful and impactful ways. I will greatly miss his sage advice and wisdom,” says Anantha Chandrakasan, chief innovation and strategy officer, dean of engineering, and the Vannevar Bush Professor of Electrical Engineering and Computer Science.

“Arvind’s positive energy, together with his hearty laugh, brightened so many people’s lives. He was an enduring source of wise counsel for colleagues and for generations of students. With his deep commitment to academic excellence, he not only transformed research in computer architecture and parallel computing but also brought that commitment to his role as head of the computer science faculty in the EECS department. He left a lasting impact on all of us who had the privilege of working with him,” says Dan Huttenlocher, dean of the MIT Schwarzman College of Computing and the Henry Ellis Warren Professor of Electrical Engineering and Computer Science.

Arvind developed an interest in parallel computing while he was a student at the Indian Institute of Technology in Kanpur, from which he received his bachelor’s degree in 1969. He earned a master’s degree and PhD in computer science in 1972 and 1973, respectively, from the University of Minnesota, where he studied operating systems and mathematical models of program behavior. He taught at the University of California at Irvine from 1974 to 1978 before joining the faculty at MIT.

At MIT, Arvind’s group studied parallel computing and declarative programming languages, and he led the development of two parallel computing languages, Id and pH. He continued his work on these programming languages through the 1990s, publishing the book “Implicit Parallel Programming in pH” with co-author R.S. Nikhil in 2001, the culmination of more than 20 years of research.

In addition to his research, Arvind was an important academic leader in EECS. He served as head of computer science faculty in the department and played a critical role in helping with the reorganization of EECS after the establishment of the MIT Schwarzman College of Computing.

“Arvind was a force of nature, larger than life in every sense. His relentless positivity, unwavering optimism, boundless generosity, and exceptional strength as a researcher was truly inspiring and left a profound mark on all who had the privilege of knowing him. I feel enormous gratitude for the light he brought into our lives and his fundamental impact on our community,” says Daniela Rus, the Andrew and Erna Viterbi Professor of Electrical Engineering and Computer Science and the director of CSAIL.

His work on dataflow and parallel computing led to the Monsoon project in the late 1980s and early 1990s. Arvind’s group, in collaboration with Motorola, built 16 dataflow computing machines and developed their associated software. One Monsoon dataflow machine is now in the Computer History Museum in Mountain View, California.

Arvind’s focus shifted in the 1990s when, as he explained in a 2012 interview for the Institute of Electrical and Electronics Engineers (IEEE), funding for research into parallel computing began to dry up.

“Microprocessors were getting so much faster that people thought they didn’t need it,” he recalled.

Instead, he began applying techniques his team had learned and developed for parallel programming to the principled design of digital hardware.

In addition to mentoring students and junior colleagues at MIT, Arvind also advised universities and governments in many countries on research in parallel programming and semiconductor design.

Based on his work on digital hardware design, Arvind founded Sandburst in 2000, a fabless manufacturing company for semiconductor chips. He served as the company’s president for two years before returning to the MIT faculty, while continuing as an advisor. Sandburst was later acquired by Broadcom.

Arvind and his students also developed Bluespec, a programming language designed to automate the design of chips. Building off this work, he co-founded the startup Bluespec, Inc., in 2003, to develop practical tools that help engineers streamline device design.

Over the past decade, he was dedicated to advancing undergraduate education at MIT by bringing modern design tools to courses 6.004 (Computation Structures) and 6.191 (Introduction to Deep Learning), and incorporating Minispec, a programming language that is closely related to Bluespec.

Arvind was honored for these and other contributions to data flow and multithread computing, and the development of tools for the high-level synthesis of hardware, with membership in the National Academy of Engineering in 2008 and the American Academy of Arts and Sciences in 2012. He was also named a distinguished alumnus of IIT Kanpur, his undergraduate alma mater.

“Arvind was more than a pillar of the EECS community and a titan of computer science; he was a beloved colleague and a treasured friend. Those of us with the remarkable good fortune to work and collaborate with Arvind are devastated by his sudden loss. His kindness and joviality were unwavering; his mentorship was thoughtful and well-considered; his guidance was priceless. We will miss Arvind deeply,” says Asu Ozdaglar, deputy dean of the MIT Schwarzman College of Computing and head of EECS.

Among numerous other awards, including membership in the Indian National Academy of Sciences and fellowship in the Association for Computing Machinery and IEEE, he received the Harry H. Goode Memorial Award from IEEE in 2012, which honors significant contributions to theory or practice in the information processing field.

A humble scientist, Arvind was the first to point out that these achievements were only possible because of his outstanding and brilliant collaborators. Chief among those collaborators were the undergraduate and graduate students he felt fortunate to work with at MIT. He maintained excellent relationships with them both professionally and personally, and valued these relationships more than the work they did together, according to family members.

In summing up the key to his scientific success, Arvind put it this way in the 2012 IEEE interview: “Really, one has to do what one believes in. I think the level at which most of us work, it is not sustainable if you don’t enjoy it on a day-to-day basis. You can’t work on it just because of the results. You have to work on it because you say, ‘I have to know the answer to this,’” he said.

He is survived by his wife, Gita Singh Mithal, their two sons Divakar ’01 and Prabhakar ’04, their wives Leena and Nisha, and two grandchildren, Maya and Vikram. 

de MIT News https://ift.tt/it4FlXn