News 2020

January 2020

Researchers Build a Better Lung Model

Byron Spice (Carnegie Mellon) and Gina DiGravio (Boston University)

Computational biologists at Carnegie Mellon University, working with colleagues at Boston University, have used machine learning techniques to develop an improved protocol for generating lung cells that can be used for investigating lung diseases. They reported the method in today's issue of the journal Cell Stem Cell. The lung cells are produced from induced pluripotent stem (iPS) cells, which are adult skin or blood cells that have been reprogrammed so they can potentially produce any cell or tissue type. By treating these iPS cells with growth factors over a period of a month, the researchers were able to transform them into cells very much like adult lung cells. Moreover, unlike previous attempts to create lung cells from stem cells, these lung-like cells can maintain their new characteristics for more than a year. Because the cells can be derived from any individual, the researchers believe this new method will improve their ability to study and treat lung disease, including idiopathic pulmonary fibrosis, chronic obstructive pulmonary disease (COPD), alpha-1 antitrypsin deficiency and neonatal respiratory distress or early-onset interstitial lung disease. "The machine learning methods we developed for this study can also be applied to studies of other tissues and organs," said Jun Ding, a post-doctoral researcher in CMU's Computational Biology Department (CBD), who co-authored the research paper with Dr. Killian Hurley, a researcher at the Royal College of Surgeons in Ireland. "We hope that our newly developed techniques for generating a pure, unlimited supply of cells using patient-derived stem cells can make possible new treatments or cures for diseases. These developments have the potential to prolong lives and improve the quality of those lives." Dr. Darrell Kotton, director of the Center for Regenerative Medicine at Boston University and co-corresponding author with Ziv Bar-Joseph, professor in CBD and CMU's Machine Learning Department, said the new method will allow researchers to generate an inexhaustible supply of new lung cells for any patient of any age.

Hebert Installed as SCS Dean

Susie Cribbs

It's fitting that the installation of School of Computer Science Dean Martial Hebert would begin with a musical intro from Carnegie Mellon University's robot bagpiper, McBlare. After all, Hebert spent his career at the Robotics Institute until taking the helm of SCS this past August. "When you hear bagpipes, you know something special is happening. When you hear bagpipes being played by a robot, you know you're at Carnegie Mellon University," Provost James H. Garrett Jr. said. McBlare kicked off the Jan. 29 event that honored Hebert's appointment as the sixth dean of SCS and celebrated the school's rich tradition of excellence and leadership in all facets of computer science. "Today, we officially recognize Martial as the dean of the School of Computer Science and commit our support to him as he advances the school's world-renowned educational, research and entrepreneurial mission and legacy," Garrett said. A native of France, Hebert earned a doctorate in computer science at the University of Paris. He joined the Robotics Institute in 1984, just five years after it was founded, and was named a full professor in 1999. For more than three decades, he led major research programs in autonomous systems, including ground and air vehicles, with contributions in the areas of perception for environment understanding and human interaction. "Martial has shown an unwavering commitment to excellence and a dedication to discovery, inclusion, collaboration and respect as the foundation of his leadership. His passion for SCS and for Carnegie Mellon is truly inspiring," CMU President Farnam Jahanian said. "With Martial as dean, I'm confident that SCS will continue to develop as the world-leader in computer science research and education, and deepen its connections with partners across campus and around the world.

Four SCS Students Named Facebook Fellows

Byron Spice

Four Ph.D. candidates in the School of Computer Science are among 36 outstanding students in computer science and engineering from 16 universities who have been named 2020 recipients of the Facebook Fellowship Program. Each Facebook fellow receives tuition and fees for up to two academic years and a stipend of $42,000, which includes conference travel support. Facebook received applications from 1,876 students at more than 100 universities for this year's program. The fellowship program chose Juncheng Yang, a Ph.D. student in the Computer Science Department, to be a fellow in computer storage and efficiency. Yang is broadly interested in the reliability, performance and availability in the storage and caching subsystems of internet-scale web services. Devendra Singh Chaplot, a Ph.D. student in the Machine Learning Department (MLD), was named a fellow in computer vision. His research aims to design algorithms capable of "physical intelligence," i.e., building intelligent embodied autonomous agents capable of learning to perform complex tasks in the physical world that involve perception, natural language understanding, reasoning, planning and sequential decision making. Sai Krishna Rallabandi, a Ph.D. student in the Language Technologies Institute, and Hubert Tsai, a Ph.D. student in MLD, were both named fellows in spoken language processing and audio classification. Rallabandi, who also earned a master's degree in language technologies at CMU, is developing a framework referred to as "De-Entanglement," which aims to isolate relevant causal factors of variation in the data distribution. He hopes to apply De-Entanglement to various tasks such as unsupervised acoustic unit discovery from speech, flexible and expressive Text to Speech synthesis, acoustic search, and visual question answering. Tsai's research goal is to understand computational and statistical principles in spoken language modeling. He plans to use these principles to enhance representation interpretability and improve data efficiency.

New Natural Language Processing Technique Helps Detect Microaggressions

Virginia Alvino Young

Existing processes for identifying toxic language primarily focus on overt hate speech and can automatically filter things like racist voices on social media. But no such tools exist for detecting microaggressions, which are much more subtle and can be just as harmful. New work by Carnegie Mellon University researchers helps identify these microaggressions in order to build intervention techniques in the future. "You're pretty for a black girl" is an example of a microaggression that can easily be intuited as a negative sentence, according to Yulia Tsvetkov, an assistant professor in CMU's Language Technologies Institute. But she said natural language processing (NLP) tools that determine sentiment may classify it as positive because of the words it uses. "Right now there's no way for NLP to detect these manifestations of social bias, which are often unconscious," Tsvetkov said. "While you can build a vocabulary of hate language, you can't do that for veiled negativity or discrimination." Researchers convened a user group to help develop a methodology to surface these microaggressions, which can be difficult for people to objectively recognize because the phenomenon itself is subjective. Based on individuals' biases and who they are, understanding of microaggressions change. Instead, researchers asked the group looking at sample sentences, "Is this offensive?" The researchers did not try to model the degree of offensiveness, but instead noted the level of disagreement between people. If a sentence was overtly positive or negative, users were prone to agree. But they found the higher the level of discrepancy, the more likely the sentence was to contain a microaggression. "We leveraged human biases to annotate for biases," Tsvetkov said. The researchers built a machine learning model to automatically predict disagreement between people. It also leveraged prior work that collected weak signals of bias, like specific ways verbs are used or mentions of certain social groups or topics that are more likely to contain bias. The resulting model does not detect microaggressions, but helps to bring possible microaggressions to the surface in a sea of social media content. A random sampling of social media posts included 3% microaggressions. The sample the model pulled included 10%. The most frequently encountered types of microaggressions identified on social media assigned stereotypical properties to social groups, such as "men are doctors and women are nurses." Tsvetkov said that while this first phase of detection helps increase the likelihood of finding microaggressions in a big pool of data, techniques still must be developed to classify sentences as microaggressions or not. "Eventually I hope this tool can be used to monitor civility on online forums," Tsvetkov said. "A future web plug-in could alert people in a friendly way, 'Do you want to rephrase that?'" Other researchers who contributed to this work include CMU's Luke Breitfeller and Emily Ahn, and the University of Michigan's David Jurgens.

Aaditya Ramdas stands to the right of a black chalkboard covered in scientific formulas.

Ramdas Honored for Efforts To Improve Research Reproducibility

Stacy Kish

Carnegie Mellon University's Aaditya Ramdas, assistant professor in the Department of Statistics & Data Science and Machine Learning Department, has received the National Science Foundation's (NSF) Faculty Early Career Development Award for his project, titled "Online Multiple Hypothesis Testing: A Comprehensive Treatment." "Arguably, one of the major hurdles to reproducibility of scientific studies is the cherry picking of results among the vast array of tests run or quantities estimated," Ramdas said. "We need 'online' methods to correct for cherry picking, first acknowledging that the problem exists and then designing algorithms that can account and correct for it." According to Ramdas, statistical methods that improve reproducibility in large-scale scientific studies will combat the increasing public distrust in science. The results of this five-year grant could transform how technological and pharmaceutical industries as well as the sciences perform large-scale hypothesis testing. In addition, it allows Ramdas to fund graduate and postgraduate students to prepare the next generation of researchers. This award supports junior faculty who exemplify the role of teacher-scholar through outstanding research, excellence in education, and the integration of education and research within the context of the mission of their organization. These activities build a firm foundation for a lifetime of leadership in integrating education and research. "Aaditya's work promises a unification of the central ideas underlying a fundamental statistical problem and a further advance in our understanding of the problem itself," said Christopher Genovese, head of the Department of Statistics & Data Science. "This project intersects with deep theoretical questions, raises a host of methodological challenges and will have a significant impact for practitioners across many fields." Hundreds or thousands of hypothesis tests, which, for example, can be used to evaluate the beneficial effects of different drugs, are often performed over the periods of months or years. It is natural that over time, this form of statistical testing may (simply due to chance) report some false results. Researchers often do not correct for cherry picking the 'best' results, and current statistical techniques are sometimes inadequate to correct for these results. According to Ramdas, these false discoveries not only lead to false hopes for potential treatments but also could waste millions of research dollars in needless follow-up studies. "What I'm really interested in is a comprehensive approach that can unify how we handle different error metrics," Ramdas said. "Long-term grants like this one open the door to the kind of research that is more exciting, in that we address open-ended questions where there seems to be light at the end of the tunnel, but the path to get there is still unclear." In this study, Ramdas will address this 'hidden' multiplicity to correct for selection bias that will improve long-term reproducibility. He hopes to develop statistical methods that will protect against the false discoveries using minimal assumptions. Ramdas aims to deliver an open-source software package to enable easier assimilation and application of these methods by other researchers. Ramdas received a joint Ph.D. in Statistics and Machine Learning from CMU in 2015. He continued his postdoctoral studies at the University of California Berkeley before returning to CMU in 2018 as an assistant professor, teaching courses in statistical methods for machine learning, reproducibility and sequential analysis.

Peg Calder Puts Her Heart (and Harp) in the Work

Susie Cribbs (DC 2000, 2006)

Peg Calder (MM 1966) has worn many hats in her lifetime. Student, mathematician, programmer, manager, advocate, fundraiser. But one of the most novel is "computress." Anyone who's seen "Hidden Figures" knows that before machines did the heavy lifting, humans who computed things were called "computers." But when Calder took her first job — at Bettis Atomic Power Lab — the company decided that since she and other colleagues were females who computed, they should have the title "computress." It's no surprise, then, that the Vandergrift, Pennsylvania, native went on to enroll at Carnegie Tech, studying mathematics in the mid-1960s, when computer software classes were new and nothing resembling computer science existed. In fact, most people didn't even know what a "programmer" was. "When I said I was going to be a programmer, people thought I was going to work for a radio station," Calder said. Calder studied on a campus populated with computing pioneers like Alan Perlis, Herbert Simon and Allen Newell, but it was also a campus that few current computer science students would recognize. (Or one her mother, who graduated in 1935, would know.) Sure, a few of her classes were in the "Potato Chip" building (Scaife Hall), which still stands, but her first programming class was in ALGOL, an algorithmic language created in the 1950s — not C or C++ or Python, which most students study now. "It was the first class I took, and the very first day of class I felt like I had walked into the middle of the semester," she said. "I had no idea what anything they were talking about meant. And I ended up loving ALGOL!" And forget completing a programming assignment by whipping a laptop out of your backpack and immediately compiling and checking your code. Instead, Calder and her classmates wrote their code on pieces of paper, then punched cards and gave them to an operator in Scaife. Two or three days later, they could pick up a printout of the program to see if it worked, or if a misplaced comma meant they had to do the same thing all over again. As a bachelor of arts student, Calder belonged to Margaret Morrison Carnegie College. And as a female student, her campus life experiences varied greatly from the coed dorms on campus today. Women lived in Morewood Gardens and men were prohibited from going beyond the front desk. But while living conditions were strictly segregated, the same wasn't true in classrooms. It's tempting to assume that because it was decades ago, women experienced discrimination on campus, but Calder said she always felt on equal footing with her male peers. And there were plenty of women in her field. "When I came back to campus for my 50th reunion, I looked in the yearbook and I couldn't believe how many female math majors there were!" she said. After graduation, Calder easily landed a job with the Johns Hopkins Applied Physics Lab, then moved on to The MITRE Corporation. She remained with MITRE for 10 years, still programming with pencil and punch cards, and eventually moved to the company's headquarters outside Boston. She'd go on to management roles at GTE Sylvania Telephone, Apollo Computer (a hardware workstation manufacturer), Alliant Computer Systems and GE Aerospace, which Martin Marietta acquired before merging with Lockheed. While she hadn't experienced sexism or discrimination at CMU, Calder definitely encountered it in the workplace. At her first two jobs, men were hired as members of technical staff, while women were associate members of technical staff unless they had a master's degree. After a few years at MITRE, Calder was promoted. "But that's five or six years of a lower salary because I was a woman," she said. She also met managers who refused to give her assignments, which she worked around in true CMU problem-solving form. And sometimes she simply wasn't taken seriously because of her gender. "One time, I had to go to the Pentagon to define some work. It was some time later that my boss said that whoever I talked to at the Pentagon really liked me," Calder said. "He had thought they liked me as a cute girl, but then he learned that they really liked my work. And he acted surprised!" Calder's career contained twists and turns, but she's especially proud of a formal Defense Department contract protest she led when she worked at Alliant, which designed and manufactured parallel computing systems. The company wrote proposals for government agencies to buy their computers, and the Navy rejected one they had written for Naval Air Weapons Station China Lake. Calder protested the rejection before the Navy JAG and the company won. "It wasn't the billions of dollars of a Lockheed or Boeing contract. It was maybe a million dollars. But it made a huge difference," she said. Calder also never stopped learning. She took courses in networking, C and C++. She even enrolled in a Dale Carnegie course once, and years later taught the material to her GE Aerospace colleagues while they waited for their security clearances to be approved during the Martin Marietta takeover. Calder technically retired in 2003, but she hasn't slowed down. "I always promised myself that when I retired, I was going to have a view," she said. So she worked with an architect to design and build the home of her dreams on Lake Champlain in Vermont. When the house was finished, she encountered a situation that would change her life well into the next decade. A young acquaintance had a great job, kids and home, but her marriage dissolved and she became a serious alcoholic. She was in and out of jail, and lost her home and family. Calder worked with friends to secure a spot for the woman in a state rehab facility, furnish an apartment and get her back on her feet. Calder even went to live with the woman and take her to counseling sessions and other appointments. "I entered this whole new world of counseling, probation, family services and food stamps. What I came away with is: They didn't know how to help her," Calder said. "I looked for an organization to join that was trying to find better medical treatment for alcoholism — like the American Cancer Society, but for alcoholism—and there wasn't one. So I started one." Thus began the Foundation for Alcoholism Research, a nonprofit dedicated to finding better medical methods for prevention, prediction and treatment of the disease. Their grants have supported research on alcohol tolerance and its relationship to alcoholism; the medication naltrexone; and a pilot program testing the medication baclofen's ability to reduce cravings. These days, Calder has a reduced role in the organization, and has traded in her Vermont lake house for a condo in downtown Pittsburgh. But she hasn't stopped learning and trying new things. Her next big adventure is taking harp lessons. She was even in a recital in CFA's Alumni Concert Hall, where she'd never set foot as student. Next hat? Harpist.

Filter News

News by Year