News 2019

October 2019

Carnegie Mellon Team Wins Oman Programming Competition

Kara Nesimiuk

For the second year in a row, a Carnegie Mellon University in Qatar (CMU-Q) student team won the top prize at the Oman Collegiate Programming Contest (OCPC). The competition was hosted by Sultan Qaboos University in Muscat. Mohammed Nurul Hoque, Muhammad Ahmad Khan and Zaryab Shahzaib (shown top right) placed first among the nearly 30 teams who competed. Hoque, who is now a computer science senior at CMU-Q, was also on last year’s winning team.

CMU Researchers Find Google's New Congestion Control Algorithm Treats Data Unfairly

Daniel Tkacik

If the internet had its own superhero, it might be the congestion control algorithm (CCA), an essential piece of code internet giants use to ensure that the web isn't crippled by a massive data traffic jam. They've been used since the 1980s to slow data transfers when they sense a network is becoming overloaded. Like any great superhero, CCAs try to work fairly. When the network is becoming overloaded, they won't prioritize one company's services over another. But new research out of Carnegie Mellon University shows that a CCA called BBR, recently developed by Google, can be unfair when competing with other services in overloaded networks. Those findings are being presented this week at the Internet Measurement Conference in Amsterdam. "In a given network, our model shows that BBR would take up 40% of the bandwidth, leaving the remaining 60% to be split between the rest of the parties on the network," said Justine Sherry, a CyLab faculty member and an assistant professor in CMU's Computer Science Department (CSD). "This goes against the concept of internet fairness." What does this mean for users? Imagine your home uses a 50 megabit per second (Mbps) connection from an internet service provider. Most CCAs try to split the bandwidth evenly when many users want to use the network. If two users are each connected to a different internet service, the CCA should try to give 25 Mbps to one user and 25 Mbps to the other. CSD Ph.D. student Ranysha Ware, who leads the research project on internet fairness, was surprised when she ran experiments modeling network links and saw BBR exhibit different behavior. "When only two users are sharing the network, BBR's share is more than fair at 40 percent," Ware said. "But as we added more users to the network, BBR did not give up any bandwidth as more users joined the network. It kept using 40 percent." Imagine six people want to share the same 50 Mbps connection. A user connected to a service using BBR would get 20 Mbps of bandwidth, leaving the remaining 30 Mbps to be split between the other five users. Each user would get only 6 Mbps to work with. For video, this variation in bandwidth could be the difference between ultra-high definition video and standard definition. In 2017, when Google first announced its algorithm, the company claimed the design was fundamentally different from most current CCAs. "People told us that it would be too hard to say anything mathematically provable about BBR because it works differently from traditional CCAs," Sherry said. But her team found that BBR could be compared with other existing CCAs in terms of how it treats data using a mathematical approach based on congestion control windowing. Is BBR going to harm internet performance for its competitors? "Only in the most congested links," Sherry said. "At my house, I have a 1 Gbps connection and it would be very hard to generate the kind of congestion that would make BBR hurt its competitors." "BBR is a new and evolving algorithm," Sherry says. "We believe that BBR will probably change because of these findings." Other authors on the study included CSD Department Head Srinivasan Seshan and Nefeli Networks software engineer and CSD alumnus Matthew Mukerjee.

Tong Wins ICDM 2019 Tao Li Award

Roberto Iriondo

Hanghang Tong, a 2009 Ph.D. alumnus of Carnegie Mellon University's Machine Learning Department, has won the prestigious Tao Li Award from the International Conference on Data Mining (ICDM). The award recognizes researchers who have received their Ph.D. in the past 10 years and have demonstrated significant impact in research, contribution, leadership and services in the areas of data mining, machine learning and artificial intelligence. Tong is now an associate professor of computer science at the University of Illinois at Urbana-Champaign. His research focuses on large-scale data mining and machine learning, especially for graph data with applications. He has received several awards, including the SDM/IBM Early Career Data Mining Research Award (2018), National Science Foundation CAREER award (2017) and ICDM 2015 Highest-Impact Paper Award. Tong has more than 100 publications and refereed articles, and is an associate editor of SIGKDD Explorations, an action editor of Data Mining and Knowledge Discovery, and an associate editor of Neurocomputing Journal. Prior to joining the University of Illinois, Tong served on the faculties of Arizona State University and the City University of New York, as a research staff member at IBM T.J. Watson Research Center, and as a post-doctoral fellow at CMU.

Computational "Match Game" Identifies Potential Antibiotics

Byron Spice

Computational biologists at Carnegie Mellon University have devised a software tool that can play a high-speed "Match Game" to identify bioactive molecules and the microbial genes that produce them so they can be evaluated as possible antibiotics and other therapeutic agents. Working with colleagues at the University of California, San Diego, and six other institutions, Hosein Mohimani, an assistant professor in CMU's Computational Biology Department, and Liu Cao, a Ph.D. student in the department, demonstrated that their MetaMiner tool could identify bioactive molecules at least 100 times faster than was possible with previous methods. The researchers discuss their findings — including their discovery of seven previously unknown molecules of biological interest from various environments like the human gut, the deep ocean and the International Space Station — in a research paper published today by the journal Cell Systems. New techniques for obtaining the DNA of microbes directly from the environment has created intense interest in microbial communities, including those that coexist with healthy humans. Some microbes produce molecules that protect their host and, thus, are candidates to become therapeutic drugs. In the last decade, microbiologists have generated a number of large databases of microbe DNA. But microbe communities consist of hundreds or thousands of different types of microbes — and millions of different molecular products — and each microbe tends to die quickly if removed individually for study. So identifying molecules that might be drug candidates and isolating the microbes that produce them requires some innovative thinking. Cao and Mohimani decided to use an approach called genome mining. This involves looking at clusters of genes and attempting to infer what molecules these genes produce. It's much like looking at an auto assembly line and trying to determine what kind of car it can build, Mohimani said. However, predicting the molecular product of a gene cluster is fraught with errors, Cao said. To work around this shortcoming, he and Mohimani borrowed a trick from electrical engineering, called Viterbi decoding, which helps engineers detect messages in a "noisy" radio channel. This enabled them to build an error-tolerant search engine that could find matches between databases of microbial DNA and databases that identify molecular products by their mass spectra. Cao and Mohimani, working with microbiologists from multiple institutions, applied their methods to the discovery of ribosomally synthesized and post-translationally modified peptides (or RiPPs), a family of natural products that have found applications in pharmaceuticals and the food industry. About 20,000 gene clusters that encode RiPPs have been discovered, but until now only a handful of RiPPs have been matched to one of those clusters. By using MetaMiner to search millions of molecular product spectra and compare them to the gene clusters in eight datasets, the researchers were able to identify 31 known RiPPs and seven previously unknown RiPPs — all in about two weeks. "Normally, you'd be happy to find one match," Mohimani said. Obtaining these results with manual methods likely would take decades, he added. Co-authors of this study included researchers from UC San Diego, St. Petersburg State University in Russia, Ningbo University in China, California Institute of Technology, the University of Southern Mississippi, the National Oceanic and Atmospheric Administration and Iowa State University. The National Institutes of Health, the National Science Foundation, the Russian Science Foundation, NASA, the CAPES Foundation and a CMU Computational Biology Department startup fund supported this research.

Putting the Power of a Film Director in an Autonomous Drone

Virginia Alvino Young

Commercial drone products can tackle some automated tasks, but one thing those systems don't address is filming artistically. A team led by Carnegie Mellon University researchers has proposed a complete system for aerial cinematography that learns humans' visual preferences. The fully autonomous system does not require scripted scenes, GPS tags to localize targets or prior maps of the environment."We're putting the power of a director inside the drone," said Rogerio Bonatti, a Ph.D. student in CMU's Robotics Institute. "The drone positions itself to record the most important aspects in a scene. It autonomously understands the context of the scene — where obstacles are, where actors are — and it actively reasons about which viewpoints are going to make a more visually interesting scene. It also reasons about remaining safe and not crashing."As a goal, "artistically interesting" is subjective and difficult to mathematically quantify, so the system was trained using a technique called deep reinforcement learning. In a user study, people viewed scenes on a photo-realistic simulator that changed between frontal, back, left and right perspectives. Shot scale and distance were also explored, as well as the actor's position on the screen. Users scored scenes based on how visually appealing they were and how artistically interesting they found them.The system learned that some movements were more interesting than others. For example, other autonomous drone products often use a continuous backshot because it allows the drone to follow a clear, safe path behind the actor. But in the user study, participants reported that a constant backshot becomes boring after a while. They also found that the drone had to switch angles often for the shot to remain interesting, but it couldn't switch too often.Bonatti said the team wanted to make the learned behavior generalizable, going from training in simulation to deployment in real life scenarios. While the system averaged users' preferences for shots as an actor walked a narrow corridor between buildings, it can apply those preferences to similar obstacles like a forest path using topographic mapping."Future work could explore many different parameters or create customized artistic preferences based on a director's style or genre," said Sebastian Scherer, an associate research professor in the Robotics Institute.The aerial system is also skilled at maintaining a clear view of the actor, avoiding what's known as occlusions. "We were the first group to come up with new ways of dealing with occlusion that aren't just binary, but can actually quantify how bad the occlusion is," Bonatti said.Other innovations include efficient motion planners to anticipate the trajectories of actors, and an incremental and efficient mapping system of the environment using LiDAR.This system could be useful beyond entertainment and sports. Governments and police departments today already use manually flown drones for many applications, including monitoring crowds and understanding traffic patterns. But manually flying drones requires a lot of attention, and an officer cannot spend their energy actually looking at the scene. "Just like learning artistic principles, the machine could be taught the shots necessary for other applications like security," Bonatti said."The goal of the research is not to replace humans. We will still have a market for highly trained professional experts," said Bonatti. "The goal is to democratize drone cinematography and allow people to really focus on what matters to them."This work will be presented at the 2019 International Conference on Intelligent Robots and Systems next month, and has been accepted for publication in the Journal of Field Robotics. The research is sponsored by Yamaha Motor Company.

HCII Researchers Go Inside the Enigma

Shannon Riffe

Researchers had a rare opportunity to peek under the hood of the Carnegie Mellon University Libraries' two Enigma machines, opening the World War II-era machines to photograph their carefully crafted interiors and locate and record the serial numbers printed on their rotors.The University Libraries acquired the two encryption devices — one four-rotor machine and one three-rotor machine — in February 2018 as part of a collection of more than 50 calculating machines, letters and books gifted to the university by author Pamela McCorduck, wife of the late Computer Science Department Head Joseph Traub. With this gift, CMU became one of a handful of American institutions to own an Enigma machine.Enigma machines, electromechanical rotor cipher machines used to encrypt communication, were most notably used by Nazi Germany to protect military communication during World War II.During a four-hour period on Oct. 7, campus historians and researchers associated with History of Science and Technology at CMU (HOST @ CMU) — a cross-campus, interdisciplinary initiative to collect and preserve CMU's historical contributions to scientific and technical development — assembled in the Hunt Library Fine and Rare Book Room to open the machines.Andrew Meade McGee, visiting assistant professor of history and the University Libraries' CLIR Postdoctoral Fellow in the History of Science and Computing, described the procedure as a hands-on history exercise."Today, historians and engineers with screwdrivers are attempting to recapture the inner workings of a past technology and trace the intellectual connections between this electromechanical piece of the past and today's information technology ecosystem," McGee said.Under the direction of Chris Harrison, Haberman Chair and assistant professor of human-computer interaction, three researchers partially disassembled the intricate machines. With only 318 Enigma machines known to exist today, the experience offered a once-in-a-lifetime opportunity for the Human-Computer Interaction Institute's Sven Mayer, a postdoctoral researcher, and doctoral students Yang Zhang and Karan Ahuja."This is a really exciting opportunity to have a piece of computing history, to be a part of its history and to find out more about its history," said Harrison, who leads the HCII's Future Interfaces Group.When the university received the Enigma machines, their model type, creation year and the unit they may have been assigned to during WWII were unknown."Part of our investigation today is to check out the condition of the machines to see what quality they've survived in and to recover the serial numbers on the rotors in particular, which are a major clue to identifying when it was made, where it was made and the configuration it was in," Harrison said.Wearing white cotton gloves, Mayer and Ahuja carefully dismantled the three-rotor machine first, pausing occasionally to consult with Harrison or refer to online resources. As they removed the rotors, they elicited cheers from the small assembled crowd, which included Mary Kay Johnsen, special collections librarian, and Scott Weingart, program director, Digital Humanities.Mayer, a German speaker, translated inscriptions found inside the case."The inscription was like a manual with a description of how to use and maintain the machine, and inside there was a description of how to put new lamps in," Mayer said. "There was no useful information to disassemble the machine."In addition to serial numbers, the researchers inspected the construction of the boxes and internal mechanisms, the arrangement of keys, and details on the plugboards, which allowed them to identify the models as an Enigma 3-Rotor A5005 and Enigma 4-Rotor M16681. Manufacturer marks revealed through the disassembly suggested the factories of origin.The presence of an "A" at the beginning of the serial numbers on rotors of the three-rotor Enigma machine indicated that it was used by the German Army or Air Force. The four-rotor machine, however, appeared in such pristine condition that Harrison doubted that it had ever been used in the field.Despite the clues unearthed during the disassembly, questions remain."While we found one of the machines on a list of all known machines in the world, the other appears to have been previously undocumented," Harrison said. "The next steps will be some basic preservation and maintenance, and then reaching out to experts who can hopefully shed more light on the provenance."

MCDS Alum Named to Forbes 30 Under 30

Byron Spice

Forbes magazine has named Siddha Ganju, a 2016 alumnus of Carnegie Mellon University's Master of Computational Data Science program, to its annual "30 Under 30" list of the brashest entrepreneurs across 20 industries. Forbes cited her in the category of Manufacturing and Industry. Ganju, an artificial intelligence researcher at Nvidia in Santa Clara, California, is working on the company's self-driving car technology. She previously contributed to deep learning object recognition algorithms as a deep learning data scientist for Deep Vision Inc. in Palo Alto. She also is co-author of an upcoming book, "Practical Deep Learning for Cloud, Mobile and Edge: Real-World AI & Computer Vision Projects Using Python, Keras & TensorFlow," a step-by-step guide for building practical applications using deep neural networks.

NIH's HuBMAP Goals Outlined in Nature Paper

Byron Spice

Work is underway by a national collaboration of scientists, including a team at Carnegie Mellon University, on the Human Biomolecular Atlas Program (HuBMAP), a National Institutes of Health initiative to create an interactive 3D map of the human body at a cellular level. The goals of the program, announced last year, are outlined in this week's edition of the journal Nature. Sponsored by the NIH Common Fund, HuBMAP is slated to continue for four years and cost $54 million. "HuBMAP, working closely with other initiatives, aspires to help to build a foundation by generating a high-resolution atlas of key organs in the normal human body and capturing inter-individual differences, as well as acting as a key resource for new contributions in the growing fields of tissue biology and cellular ecosystems," according to the paper by the project's investigators. Those authors include Ziv Bar-Joseph, a professor in the Computational Biology (CompBio) and Machine Learning Departments, who leads HuBMAP's Computational Tools center. CompBio is leading the development of computational methods and pipelines to process, analyze and model the data collected by the HuBMAP Consortium. The CMU-led team is part of the HuBMAP Integration, Visualization & Engagement group, which is tasked with integrating and enabling access to all data collected and with creating the 3D maps that would allow users to visualize the activity of cells and tissues in the human body. "The HuBMAP vision is exciting, and the resulting 3D maps can lead to several advances in our ability to understand what goes wrong in diseases and disorders," Bar-Joseph said. "However, the computational challenges we face when trying to make sense of all the data collected, and when attempting to integrate it to create such maps, are substantial. "The computational methods and pipelines being developed by our team are aimed at addressing these challenges and enabling the general public, as well as expert users, to interact with and utilize the data collected so that they can perform detailed queries and comparisons between the reference and patient specific data," he said. "We have already implemented several pipelines for processing the large amounts of data that we expect to receive in the next few month" said Matt Ruffalo, a systems scientist in CompBio. "We expect to have all processing tools in place for the initial data release planned for next year. In addition, we have already started working on new solutions for uniform processing and integration of the different types of sequencing, imaging and mass spectrometry data being collected by the HuBMAP tissue centers." In addition to the CMU team, scientists at the Pittsburgh Supercomputing Center and the University of Pittsburgh's Department of Biomedical Informatics lead another HuBMAP center, which focuses on the computational infrastructure for the data.

Google Supports CMU Outreach to Women Computer Scientists

$story.metadata.author

The School of Computer Science's OurCS research-focused workshop for undergraduate women considering graduate studies in computer science is among the recipients of this year's Google exploreCSR grants, which were inspired, in turn, by the success of OurCS. CMU is one of 24 institutions to receive exploreCSR grants this year. Google officials established exploreCSR last year to help institutions run workshops that are modeled after OurCS. SCS hosted the first OurCS workshop in 2007, introducing more than 60 undergraduates from across the nation and around the world to the type of research they would encounter as graduate students in computer science. "OurCS has been incredibly successful, so we're delighted that Google is helping to duplicate its success at other universities," said Carol Frieze, director of Women@SCS, an organization that promotes opportunities for women, including the OurCS workshop. "We also appreciate getting additional support from Google to help sponsor this year's workshop." More than 70 undergraduate women from 32 schools; 19 US states; and countries including Uganda, Mexico, and the United Kingdom are registered for this year's OurCS, which will take place Oct. 18–20. The keynote speakers this year include Sue Black, a professor of computer science and self-proclaimed technology evangelist at Durham University in England; and Lenore Blum, a CMU professor emeritus of computer science, founder of Project Olympus and a leader of SCS's efforts to increase the number of women in computer science. Through the efforts of Blum, Frieze and many other SCS colleagues, undergraduate computer science enrollment at SCS is now almost evenly split between women and men — unique among leading computer science schools. OurCS and the exploreCSR workshops sponsored by Google help undergraduate women to enhance their research skills, create a sense of community with peers and faculty, instill confidence to problem solve beyond the classroom, and inspire and motivate them toward careers in research. Other sponsors for OurCS include Oracle, a major sponsor since 2013. "Industry sponsors such as Oracle and Google are critical for sustaining workshops like OurCS and exploreCSR. They are taking a leadership position in the global effort to recognize and encourage women in computer science to reach their full potential" Frieze said. "Making computer science education accessible and available to everyone is an important initiative," said Hejazi Moghadam, senior program manager at Google. "We're excited to provide this grant to Carnegie Mellon to help encourage more women to pursue careers in computer science research."

Female-Led Team Uses AI to Help Machines Play Nice With Humans

Stacy W. Kish

Three Carnegie Mellon University researchers are leveraging their expertise in organizational science, cognitive science and artificial intelligence (AI) to explore how AI can help humans work together better. The collaboration includes lead investigator Anita Woolley in the Tepper School of Business, along with co-investigators Cleotilde Gonzalez in the Dietrich College of Humanities and Social Sciences and Henny Admoni in the School of Computer Science.Woolley believes this award is an important achievement not only for her and her co-investigators, but also for Carnegie Mellon. "As leaders in both teams research and technology, it's an important acknowledgement of the role we have played and continue to play in pushing the frontiers of these disciplines," she said.Bringing Machines Into the FoldThe researchers received a $2.8 million DARPA grant to study team collective intelligence and the theory of mind involving human and machine interactions. Team collective intelligence explores how a group of people develops a level of intelligence that only emerges from the particular skills contributed by each individual team member. Theory of the mind explores how a person can predict what will happen next."AI has become part of our lives. Every day, we talk to our phones and these machines understand and collaborate with us," Gonzalez said. "We want to expand this concept to situations in which a group of individuals and machines are performing a task together and can interact in effective ways."The researchers aim to leverage these concepts to develop a machine theory of the mind for a synthetic team coach that can interact with a group to improve task outcomes. A synthetic team coach will benefit from a previously developed cognitive model that can remember the actions of the team members. This memory will allow the coach to understand and accurately predict what the team members will do next and intervene at the team level at the right time to improve the task at hand."I'm very interested in how intelligent synthetic agents can help teams coordinate better," said Admoni, an assistant professor in the Robotics Insitute. "I am looking forward to using my knowledge of AI and human-robot interaction in this domain of group coordination, which is a new area for me."The goal is to have the coach work with team members to aid in the exchange of ideas and improve collaborations so tasks are performed faster and more efficiently.Women in the LeadThe cross-disciplinary team behind this cutting-edge research is as intriguing as the research itself. Woolley brings expertise in collective intelligence in teams, Gonzalez in cognitive and decision sciences and Admoni in human-robot interactions."We continue to push the frontiers of these disciplines," Woolley said. "Since ours is an all-female Carnegie Mellon-based team of scientists, it shows that women are leaders in these fields where they have been typically underrepresented."The project is also drawing on the talent at all levels of professorship at CMU and offers young researchers, both undergraduate and graduate, a model of a women-led team in male-dominated fields of science and engineering."I hope that I can be a role model for women in computer science," Admoni said. "I think that science would benefit greatly from more diversity in the field."

CMU Professors Awarded Grants To Advance Public Interest Technology

Shryansh Mehta

Yulia Tsvetkov and Christopher Goranson have each received grants of $90,000 from the Public Interest Technology University Network (PIT-UN) as part of the organization's Network Challenge. The program, in its inaugural year, supports the development of new public interest technology initiatives and institutions in academia, and fosters collaboration among the network's partner institutions, including Carnegie Mellon University. Tsvetkov is an assistant professor in the School of Computer Science's Language Technologies Institute (LTI). Her grant will support a project that aims to bridge the ethical gap in computer science education by helping institutions teach socially responsible language technologies. She currently co-teaches a course in computational ethics, which introduces students to real-world language technology applications while addressing ethical implications and risks posed by language technology and other artificial intelligence (AI) tools. "Billions of citizens use social media, email and text messaging platforms built upon language technologies. These tools have become increasingly prevalent in data analysis, providing services on the web, even providing means for disaster response," Tsvetkov said. "But there is also a growing awareness about the negative side: bias in AI tools learned from user-generated data, pervasiveness of hate speech, propaganda and fake news." The award will help to expand the course for both graduate and undergraduate students and to develop open-access educational materials including video lectures and slides, lecture notes, assignments, creation of a textbook, and sample course projects. Goranson is a distinguished service professor in the Heinz College of Information Systems and Public Policy. He leads the Policy Innovation Lab (PIL) course, a new initiative that connects students with actual policy challenges and introduces an agile, design-driven framework to rapidly create solutions to those challenges. With his grant, he aims to develop an open-access, open-source starter kit and fellowship program that formalizes the PIL course framework. The starter kit will help PIT-UN member universities train future public interest technologists by adopting coursework that encourages rapid experimentation, novel approaches and viable solutions that meet the needs of end users.

CDC Funds Carnegie Mellon's Flu Forecasting Center

Byron Spice

The U.S. Centers for Disease Control and Prevention has named Carnegie Mellon University as an Influenza Forecasting Center of Excellence, a five-year designation that includes $3 million in research funding. For four of the past five years, Carnegie Mellon's forecasting efforts have proven the most accurate of all the research groups participating in the CDC's FluSight Network. In addition to expanding CMU's existing forecasting research, the new funding will enable CMU to initiate studies on how to best communicate forecast information to the public and to leaders. It will also support efforts to determine how forecasting techniques might apply to pandemics — the rare occasions when a truly novel strain of flu is prevalent around the world. Roni Rosenfeld, head of CMU's Machine Learning Department and leader of its epidemic forecasting efforts, said the designation of CMU and the University of Massachusetts at Amherst as the first two CDC flu forecasting centers of excellence marks a coming of age for the epidemic forecasting community. "When the CDC began soliciting flu forecasts, they ran it as an experiment," without funding, Rosenfeld said. But as the usefulness of the forecasts became apparent, the CDC has placed greater reliance on them. "The CDC now routinely includes our forecasting in their messaging to the public and to decision makers." "In the beginning, we had about 10 groups that voluntarily submitted forecasts," he said. "Now the CDC receives more than 40 forecast submissions. It has become a community and more and more groups are getting involved, which is the real win." The CDC has historically tracked flu epidemics through a surveillance network that includes doctors' offices and clinics. But just as weather forecasting is more useful than only reporting the current weather, accurate flu forecasting enables health officials to make more timely decisions to launch public information and vaccination campaigns and helps health providers plan clinic schedules and staffing. CMU has focused on two methods for flu forecasting — one that uses machine learning and computational statistics to make predictions based on both past patterns and input from the CDC's domestic flu surveillance system and a second that bases its predictions on the judgments of human volunteers who submit weekly predictions. Work will continue on both those methods, said Ryan Tibshirani, associate professor of statistics and machine learning and co-leader of the Delphi Research Group, which is devoted to epidemic forecasting. Baruch Fischhoff, professor of engineering and public policy and an expert on risk communication, will join the expanded effort. He will explore how flu forecast information can be communicated so both decision makers and the general public can use it effectively, while understanding the limits of the forecasts. "These audiences want to know what will happen," Tibshirani said. "We can only tell them what will probably happen. We want to be sure we send them messages that they interpret properly." Researchers at the Harvard School of Public Health also will join with CMU to explore how forecasting technology might apply to pandemics. These events only happen "once in a blue moon," Rosenfeld said, but are critically serious when they do occur. The University of Pittsburgh School of Public Health, a previous collaborator with CMU, also will be part of the new center of excellence, providing important new sources of information to improve forecasts' accuracy. Previous sponsors of CMU's forecasting research have included the Defense Threat Reduction Agency and the National Institute of General Medical Sciences' Models of Infectious Disease Agency Study (MIDAS).

One-Dimensional Objects Morph Into New Dimensions

Byron Spice

A line is the shortest distance between two points, but "A-line," a 4D printing system developed at Carnegie Mellon University, takes a more circuitous route. One-dimensional, "line"-shaped plastic structures produced with the A-line system can bend, fold and twist themselves into predetermined shapes when triggered by heat.3D-printed objects that later change shape are the very definition of 4D printing. But the process takes on special qualities when the objects can fit through narrow openings. A rod inserted through a narrow bottleneck, for instance, might transform into a hook to fish an object out of the bottle. Or a long, thin fastener inserted through holes in the seat of a chair might lock a chair leg into place.The A-line method also can be useful in making compliant devices, such as coil springs and tweezers. These are difficult to produce in final form using a 3D printer, but can be printed readily as rods that assume final form when dipped in hot water.Making sticks that morph into new objects is a feat that Lining Yao, assistant professor in CMU's Human-Computer Interaction Institute, and her colleagues in the Morphing Matter Lab have accomplished using an ordinary, hobbyist-grade 3D printer and a single type of thermoplastic material."It's not printing the line that's difficult, but it's developing the software tool that enables you to design, simulate and fabricate the line," Yao explained.The group used polylactic acid, or PLA — the most common material used in 3D printing — to produce their objects. PLA shrinks in reaction to heat along the direction in which it was printed, said Guanyun Wang, a post-doctoral fellow in the Morphing Matter lab. That makes it possible to control how an object's shape will morph based on the spacing of active and passive segments, the thickness of segments and on the printing direction of each segment, he explained.The A-line platform developed by Yao's team includes a library of eight bending directions that can be combined to produce simple or complex geometries. It also includes a customized design tool to help users combine these different types of bends to achieve desired shapes.Ye Tao, a visiting scholar at the HCII from Zhejiang University, said the team triggered the bending by immersing the engineered sticks into water heated to about 170 degrees Fahrenheit. Morphing also can be triggered by a heat gun, with embedded carbon fiber for resistive heating or with steam via hollow channels in the sticks.As with other 4D-printed objects, one advantage of the A-line rods is that they can be shipped as a flat pack and triggered on site to become tent supports, chair frames or sculptures. But Yao envisions some applications that are peculiar to line-shaped objects. By using electrical field responsive hydrogels instead of PLA, for instance, it might be possible to develop a biocompatible line that a surgeon could snake through narrow body spaces and remotely transform into surgical tweezers. By controlling electrical fields, it might also be possible to control the tweezer movement."Through this work, we hope to enlarge the design space of 4D printing technology," Yao said. "We encourage designers to think about additional novel uses of A-line."A research paper describing A-Line was presented earlier this year at CHI 2019, the Association for Computing Machinery's Conference on Human Factors in Computing Systems. The Richard King Mellon Foundation supported this research through the CMU Manufacturing Futures Initiative. In addition to Yao, Wang and Tao, the research team included Ozguc Bertug Capunaman and Humphrey Yang, both master's students in the School of Architecture. 

CompBio's Mohimani Goes Big To Find New Antibiotics

Byron Spice

Hosein Mohimani has a knack for finding ever-faster ways of searching giant databases for potential therapeutic drugs. Now, the computer scientist is trying a new approach in which the size of the databases isn't something to be conquered, but a feature he uses to his advantage. "More data can make the search easier," said Mohimani, an assistant professor in the Computational Biology Department. He'll be developing this bit of computational jiu-jitsu with the help of a National Institutes of Health Director's New Innovator Award. He is one of 60 young investigators receiving the $1.5 million award this year in recognition of their unusually creative, high-impact research efforts. With the NIH support, Mohimani plans to analyze a wide variety of microbial communities — from humans and from other animals, but also from various soil and marine environments around the world. Each community might include thousands of microbes that produce potentially tens of millions of molecules, some of which act to protect their hosts. The idea is to find these protective molecules, which might prove to be novel antibiotics or anticancer agents. Rather than search through databases for each microbiome, Mohimani plans to compare entire communities with each other. These comparisons should enable him to identify differences in the presence or relative abundance of microbes and their products. These patterns should then allow him to infer the activity of the microbes and identify candidate drugs. The findings may also be a boon to precision medicine, he said. "The interactions we detect may guide us in better understanding the microbial basis of disease," Mohimani explained, noting most diseases involve some increase or decrease in the abundance of special molecules. Developing these databases for microbiomes has been an area of intense interest in the microbiology community for the past 10 years, Mohimani said. That is one factor that has made this approach possible, though he also will collaborate with Pieter Dorrestein at the University of California, San Diego, to generate additional databases for this study. Mohimani also leverages his knowledge of computer science and electrical engineering to develop his techniques. "I really love the field," he said. When he was working on his Ph.D. in electrical engineering at UC San Diego, however, he took a course with Pavel Pevzner, a leader in computational biology, regarding computer algorithms that were applicable to biology. Looking for ways to have an impact on humanity, he asked Pevzner if he could work with him. From that point on, he was hooked on computational biology. Drug discovery, particularly of antibiotics, became his special passion. "Antibiotic resistance kills millions of people every year worldwide," he said. "Based on current trends, deaths from antibiotic resistance will surpass those of cancer by 2050. That makes it imperative to develop new methods to discover new antibiotics." That's just the type of project that is the focus of the NIH's High-Risk, High-Reward Research Program, of which the New Innovator Award is a part. It is funded through the NIH Common Fund, which seeks to close gaps in the research enterprise that are of great importance and require collaboration across the agency to succeed. The program awarded 93 grants this year for about $267 million over five years. With existing technology, each new antibiotic discovery costs millions of dollars, and getting that drug to market can costs billions, Mohimani said. The new technique he is developing might be one way to make that drug discovery pipeline faster and cheaper. "The goal of my lab is to make antibiotic discovery a high-throughput, low-cost process," he added.