Allen School News

In yet another example of how computation is transforming biology and medicine, Allen School researchers have developed a machine learning-based system that improves upon a widely used technique for analyzing interactions between DNA and the proteins that regulate gene expression. In a paper published in Nucleic Acids Research, Ph.D. students Nao Hiranuma and Scott Lundberg and professor Su-In Lee demonstrate how their system, AIControl, is more cost-effective — and yields more accurate results — than current practices for increasing our understanding of genetic factors regulating the onset of disease and other biological processes.

AIControl is designed to be used in conjunction with chromatin immunoprecipitation and DNA sequencing (ChIP-seq), a vital tool in molecular biology for determining the location and function of transcription factors that govern gene expression. ChIP-seq enables researchers to map the binding sites of a specific regulatory protein to DNA across the human genome. While ChIP-seq is one of the most advanced and popular techniques available, it is not without its shortcomings; in addition to being a costly experiment to run, the data generated by immunoprecipitation (IP) contains background signals that can lead to false positives.

To compensate, researchers are advised to generate an additional control dataset in addition to their target dataset. While the IP target dataset captures actual protein binding signals, the control captures potential biases in the data. The results of both are then subjected to a process known as “peak calling,” in which algorithms compare the two datasets and separate out the peaks, which indicate the presence of true protein binding signals, and minimize false positives stemming from background noise. It is these peaks that researchers are interested in exploring, as they indicate the site of DNA-protein interactions that influence biological processes.

Due to the time and expense associated with generating that second dataset, many users opt to rely on an existing control pulled from a public database or forego the recommended control altogether. As an alternative, the Allen School team developed a machine-learning framework, AIControl, that replaces the need for the additional control dataset by estimating it in silico using multiple, publicly available controls.

AIControl works by systematically determining the most appropriate combination of control datasets to be applied to the target experiment, then estimates the distribution of unwanted background signals based on those datasets to identify the true binding peaks. With AIControl, researchers are able to rely on efficient and cost-effective computation, rather than expensive biological experiments, to generate more accurate ChIP-seq results.

“By making use of existing datasets on a large scale, AIControl can save researchers time and expense while offering a more comprehensive and accurate peak analysis of their target dataset,” explained Hiranuma. “Because our system is capable of leveraging information from a large number of control experiments in a public database, AIControl captures potential biases in the data that might be missed using a single control — or using no control at all.”

While existing peak calling techniques require the user to decide which control datasets to apply, AIControl alleviates that burden by automatically integrating and weighing multiple datasets that are most relevant to the target dataset. The system draws upon data from 440 publicly available controls, encompassing more than 100 cell types, to infer the distribution of background signals for the peak calling comparison. Hiranuma and his colleagues evaluated their system by applying AIControl to 410 IP datasets from the ENCODE ChIP-seq database spanning five major cell types.

AIControl outperformed existing peak calling methods in identifying putative binding sites, including in cases where control datasets of the same cell type were removed. This suggests that AIControl will be capable of reliably estimating background signals in conjunction with ChIP-seq analyses performed on new cell types. According to Lee, the team’s findings have already generated interest among biotechnology companies eager to replace a costly process for generating new data with an AI-driven solution.

“Locating the binding sites of regulatory proteins on DNA is a central problem in molecular biology that will enable us to more fully understand the interplay between genetic factors and disease,” Lee said. “With AIControl, we have shown that machine learning can be used in place of expensive biological experiments to generate results with greater speed and accuracy than standard approaches. Our hope is that this will advance our understanding of genetic factors that influence disease and, ultimately, lead to better outcomes for people.”

Read the journal paper here, and visit the project webpage here.

The University of Washington community mourns the loss of our friend and colleague, Vikram Jandhyala, a committed educator, innovator, and entrepreneurial leader who made a lasting impact on our campus culture and the community. In addition to his teaching and research, Jandhyala left his mark through his tireless efforts to mentor faculty and students and help translate research into real-world impact via UW CoMotion. He will also be remembered for his enthusiastic leadership of new and innovative partnerships such as the Global Innovation Exchange (GIX) in collaboration with Microsoft and Tsinghua University in China.

Jandhyala began his UW career in 2000 as a professor in what was then known as the Department of Electrical Engineering (EE). He was Founder & Chief Technologist of UW spin-out company Nimbic (formerly Physware), which developed high-speed, 3D electromagnetic simulation solutions used in hardware design that was subsequently acquired by Mentor Graphics. Jandhyala served as ​Department ​​C​hair of ​EE from 2011 to 2014, during which time he also served as the founding director of the joint UW and Pacific Northwest National Laboratory Northwest Institute for Advanced Computing. He stepped down from his role as ​Chair to become the UW’s ​Vice ​P​resident for ​I​nnovation ​S​trategy and ​E​xecutive ​D​irector of CoMotion.

“Vikram played many different roles and was a friend to the Allen School in each of them,” said Hank Levy, Director of the Allen School. “As chair of Electrical Engineering, he worked with us to develop a joint faculty hiring program, the Experimental Computer Engineering Lab (ExCEL), which has been incredibly successful in attracting world-class engineering faculty to UW; as head of CoMotion, he helped us to multiply the impact of our technology; and as director of GIX, he built a unique and creative international educational institution. We will miss his positive and collaborative attitude, his vision, and his enthusiastic support for the work of our faculty and students.”

In the nearly five years Jandhyala led CoMotion, he molded it into more than a technology transfer and commercialization office; under his stewardship, it became the entrepreneurial hub for the entire university community. He brought the same collaborative spirit and entrepreneurial zeal to GIX, serving as Co-executive Director since 2015 and celebrating its first graduating class of 37 students from 11 countries last year. According to UW President Ana Mari Cauce, both CoMotion and GIX “will stand as testaments to Vikram’s legacy.”

Jandhyala was joined on the GIX leadership team by his faculty colleague, Shwetak Patel, who led the interdisciplinary faculty group that helped craft the GIX curriculum and serves as its Director and Chief Technology Officer. As director of the Ubicomp Lab, Patel has taken inspiration from Jandhyala to made entrepreneurial leadership a cornerstone of his teaching and mentorship. He has started multiple companies with students and colleagues since his arrival at the UW, including residential water and energy monitoring company Zensi, which was acquired by Belkin in 2010, and SNUPI Technologies, develop of the WallyHome wireless home sensing platform that was acquired by Sears in 2015. He worked directly with Jandhyala and his team on the acquisition of his most recent startup, mobile health sensing company Senosis, by Google in 2017. Patel credits Jandhyala for helping to convince him to come to Seattle and for supporting his career trajectory melding innovative research with entrepreneurial impact.

“Vikram built an infrastructure and culture around supporting an entrepreneurial spirit that cut across the entire University,” said Patel, who holds a joint appointment in the Allen School and the Department of Electrical & Computer Engineering (ECE). “He laid the groundwork for companies like Senosis to succeed by enabling us to leverage commercialization gap funding, incubation, and other programs that flourished under his leadership.

“On a personal note, Vikram was not only a colleague, but also my mentor and friend,” Patel continued. “He demonstrated through his own work and deeds that you could think big about the future of innovation and play leadership roles that help move an organization into new directions, while remaining grounded in your work with students and faculty.”

Joshua Smith, who is also a professor in both the Allen School and ECE and directs the UW’s Sensor Systems Laboratory, recalled Jandhyala’s vision and generosity in helping Smith to launch a startup company, Proprio, to commercialize a new machine-learning enabled visualization tool for surgical teams.

“Vikram selflessly catalyzed the formation of Proprio by introducing me to Dr. Sam Browd, an entrepreneurial doctor who had identified a medical need,” Smith said. “I will miss his positive, enthusiastic, and generous spirit.

“I first worked with Vikram when I was a researcher at Intel, and we collaborated on an NSF-funded research project,” Smith continued. “He went on to become faculty coordinator of EE’s Professional Master’s Program — a role I now have the privilege of filling — followed by department chair, head of CoMotion, and head of GIX. Vikram was a visionary leader and role model in all of these positions, and I marvel at how much he accomplished and how many lives he changed for the better in the short time he was here.”

Smith worked with Jandhyala on the successful spin-out of another two companies — wireless robot charging company Wibotic and Jeeva Wireless. The latter, which Smith co-founded with a team that included Allen School professor Shyam Gollakota, aims to transform the way we power the Internet of Things by enabling battery-free communication using backscatter technology. After spinning out from the UW, the company raised $5 million in venture capital, grants, and project funding. Gollakota also worked with Jandhyala on a deal with ResMed to commercialize the technology behind ApneaApp, which detects signs of sleep apnea, and the spin-out of health-oriented mobile sensing company Sound Life Sciences, which is commercializing the Second Chance opioid detection app. Both of those apps were developed in conjunction with UW Medicine — two more examples of the interdisciplinary innovation in service to big ideas that was a hallmark of Jandhyala’s leadership.

“Vikram made CoMotion a kind and supportive place that encourages researchers to take the big risks of entrepreneurship. I knew CoMotion was in safe hands because he was at the helm,” said Gollakota. “His smile and enthusiasm were extremely infectious. His death is a huge loss for the university and the Seattle tech community, and a very sad development for all of us.”

“Vikram had a huge positive impact on education, on research, and on innovation at the University of Washington and far beyond,” said professor Ed Lazowska, who holds the Bill & Melinda Gates Chair in the Allen School. “Under his leadership, CoMotion completed UW’s transformation from ‘licensing’ to ‘commercialization’ to ‘innovation.’ And the Global Innovation Exchange combined technology, design, and entrepreneurship in a project-based, global context.

“Vikram was a visionary, a friend, and an inspiring leader,” Lazowska continued. “I — and all of us in the Paul G. Allen School — are among the many who understood and appreciated the work that he did to make us all better. We will miss him terribly.”

Read President Cauce’s tribute to Jandhyala here, ECE Chair Radha Poovendran’s tribute here, and articles on Jandhyala’s life and impact in GeekWire, Puget Sound Business Journal, The Seattle Times, and Xconomy.

Helpful Resources

If you’re thinking about suicide, are worried about a friend or loved one, or would like emotional support, the Lifeline network is available 24 hours a day/7 days a week across the United States. Call the National Suicide Prevention Lifeline at 1-800-273-8255, or contact the Crisis Text Line at 741741.

Need help now? If the person you are concerned about is in immediate danger of harming themselves and/or refuses to stay safe with you, call or text 911.

The following resources are also available to the UW community:

SafeCampus: 206-685-SAFE (7233)

UW CareLink (faculty and staff): 866-598-3978

UW Counseling Center: 206-543-1240

Forefront Suicide Prevention: 206-543-1016

The team behind the Allen School’s TVM framework, an end-to-end compiler stack which enables the rapid deployment of deep learning on a variety of platforms and devices, is marking a new milestone in the project’s development with its transition to the non-profit Apache Software Foundation.

The Apache open-source developer community focuses on incubating open-source software projects for the public good. By adding TVM to its incubator portfolio, Apache will ensure that the already popular framework will continue to evolve and expand to meet the needs of developers and hardware manufacturers.

“We are thrilled that TVM is entering an exciting new phase as an Apache incubator project,” said Allen School professor Luis Ceze. “Apache’s robust community of developers and proven track record of nurturing open-source innovation are the perfect environment in which to take TVM to the next level and expand its impact.”

TVM started out as a research project in the Allen School’s SAMPL group, an interdisciplinary team of researchers with expertise spanning machine learning, computer architecture, systems, programming languages, and software engineering. In addition to Ceze, members of the TVM research team include Allen School Ph.D. students Tianqi Chen, Meghan Cowan, Ziheng Jiang, Steven Lyubomirsky, Thierry Moreau, Jared Roesch, Haichen Shen, Luis Vega, and Eddie Yan; and professors Carlos Guestrin, Arvind Krishnamurthy and Zachary Tatlock. Together, they built a full-stack compiler that bridges the gap between deep-learning systems optimized for productivity, and the various programming, performance and efficiency constraints of the latest hardware. They followed up on the initial release of TVM with the Versatile Tensor Accelerator (VTA), an extension to TVM that enabled machine learning researchers and practitioners to explore hardware-software co-design techniques, and Relay IR, a high-level intermediate representation to effectively enable front-end workloads to be compiled and optimized. The SAMPL group organized the first developer conference last December which attracted 180 attendees from all around the world.

The TVM ecosystem quickly caught on within the developer community, which enjoyed the ability to quickly and easily deploy deep learning applications on a wide range of systems — from specialized chips and mobile phones, to wearables and embedded devices — without sacrificing battery power or speed. As an open-source project from the start, TVM has benefited from the input of nearly 200 individuals and organizations, including the likes of Amazon Web Services (AWS), Qualcomm, Facebook, Google, Huawei, AMD, Microsoft, Xilinx, Cornell University, University of California, Berkeley, and more. Several major technology companies have incorporated TVM’s functionality into the products and services they provide to end users.

“The values of the original TVM community and the Apache community are very much aligned,” noted Chen. “Both have demonstrated a commitment to advancing technological innovation within the context of an open, welcoming, and neutral policy towards technology development. It has been gratifying to work with the TVM community, and we look forward to continuing on the same path under the Apache umbrella.”

“The Apache incubation is a huge milestone for TVM,” said Moreau. “Developers will be able to benefit from, and contribute to the complete TVM stack: from the flexible high-level IR down to the design of VTA’s custom hardware architecture. This is truly a first in full-stack deep learning system optimization, and it’s going to open the path to exciting research.”

“Apache incubation will allow us to better collaborate with both researchers and industry,” said Roesch. “The TVM stack has continued to open new paths for collaboration, and enabled mutual technology transfer between companies and researchers. Industry quality software can help accelerate research, and research can open new product avenues for companies in a virtuous cycle. We can all benefit from sharing technology in an open stack equally accessible to all contributors.”

“TVM is right for the Apache Software Foundation, and the Apache Software Foundation is right for TVM: One thing the ASF excels at is enabling collaboration across organizations, and encouraging collaboration even among competitors. With contributions from such a wide range of organizations, TVM clearly fits that profile. I am honored to help the project thrive in the ASF,” said Markus Weimer, the ASF member who championed the incubation of TVM at the ASF.

Read the TVM announcement here, and visit the new Apache TVM incubator page here. Learn more about the Apache ecosystem here.

Allen School Ph.D. student Venkatesh Potluri has earned a 2019 Google Lime Scholarship, which recognizes and supports students with disabilities who demonstrate the qualities of leadership, commitment to academic excellence, and passion for computer science and technology. Potluri, who is blind, is in his first year at the Allen School working on human-computer interaction (HCI) and accessibility research as a member of both the Makeability Lab and Make4All Group.

While most fledgling Ph.D. students ease into research in their first year, Potluri has initiated two research projects in his short time at the University of Washington. For the first, Potluri is collaborating with his advisors, Allen School professors Jon Froehlich and Jennifer Mankoff, to develop new techniques for enabling blind programmers to develop and test user interfaces (UIs). Through a combination of semi-structured interviews and online surveys of individuals with visual impairments, the team aims to gain a better understanding of how individuals with visual impairments engage with the spatial aspects of UIs — such as the size, shape, and location of interactive elements — and how they customize the look and feel of their social media pages. They plan to use this input to develop novel programming tools that can be used by both novice and professional programmers alike to create, evaluate, and refine interactive UIs. The project is a natural extension of Potluri’s previous work on CodeTalk, an open-source plug-in that improves the accessibility of integrated development environments (IDEs) such as Microsoft VIsual Studio. That work, which Potluri completed as a research fellow at Microsoft Research Bangalore, improved the glanceability, navigability, and alertability of code through a combination of UI enhancements and both speech and non-speech cues for developers with visual impairments.

In addition to his work on accessible UI design, Potluri is simultaneously leading a project to rethink the multi-touch laptop trackpad as an accessibility device. This effort will build upon previous work by Mankoff and researchers at Carnegie Mellon University on Spatial Region Interaction Techniques (SPRITEs), which repurposed the keyboard to enable people with vision impairments to access interactive website content. Working with three other graduate students as part of a graduate-level HCI course, Potluri aims to extend that concept by identifying challenges for visually-impaired users that could be overcome through use of the trackpad and developing new trackpad interaction techniques.

“Venkatesh’s commitment is to raising the ceiling for the sophistication and range of tasks that Blind computer users and programmers can accomplish. Both of his projects have the potential to make important contributions toward this goal” said Mankoff.

In addition to his technical projects, Potluri is collaborating with two fellow students with disabilities to generate an auto-ethnographic account of their experiences as Ph.D. students in Computer Science & Engineering with disabilities. The hope is that by sharing their unique insights and challenges, they can inspire others with disabilities to pursue advanced degrees in STEM fields. Potluri has a history of taking up the cause of technologists with disabilities with positive results. As a co-founder of Inclusive STEM (I-STEM) in his native India, he and his fellow technologists who are blind or visually impaired are helping to make STEM education more accessible. I-STEM’s initiatives include a virtual training program that teaches students with disabilities how to code; inclusive hackathons that engage developers, industry professionals, and students with visual impairments to collaborate on projects; and a series of webinars introducing students to accessibility research in India and beyond. Potluri remains active with the organization, which has succeeding in creating a community and support system for aspiring technologists with disabilities — exactly the kind of leadership and commitment that the Google Lime Scholarship is designed to promote.

“For a first-year Ph.D. student, Venkatesh has set himself ambitious goals. But with his combination of technical acumen and can-do attitude, I am confident that he will achieve them,” said Froehlich. “He is already an accomplished developer, researcher, and social entrepreneur, and I foresee that he will have an even greater impact on the world through his ideas, his technical innovations, and his mentorship of other students and technologists.”

The Google Lime Scholarship is administered through a partnership with Lime Connect, a global non-profit organization that connects students, veterans, and professionals with disabilities with educational and career opportunities. Learn more about the Google Lime Scholarship Program here.

Congratulations, Venkatesh!

Allen School professor Richard Ladner, a leading advocate for making computer science education accessible to all, and a team of collaborators earned a Best Paper Award for their work on “Computer Science Principles for Teachers of Blind and Visually Impaired Students” at the recent SIGCSE 2019, the flagship conference of the Association for Computing Machinery’s Special Interest Group in Computer Science Education.

Computer Science Principles (CSP) is a new Advanced Placement (AP) course developed by the College Board, in collaboration with computer science educators, to enable high school students to learn about computational concepts and practices while earning optional college credit or placement. In contrast to prior AP computer science courses, CSP was designed to appeal to the broadest possible spectrum of students with the goal of increasing participation in computing, particularly among groups that are underrepresented in the field. While demographic data on recent CSP exam-takers indicates that forward progress has been made in engaging more students from multiple underrepresented groups, one population was missing from the dataset: students with disabilities.

In their award-winning paper, Ladner and his co-authors — professor Andreas Stefik and software engineering student William Allee of the University of Nevada, Las Vegas, and Sean Mealin, a Ph.D. student at North Carolina State University — partnered with Code.org to perform a comprehensive review of its CSP curriculum and tools to determine how much of the content meets accessibility standards and, in the case of online content, could be used with screen readers and other technologies by students who are visually impaired. The team discovered that most of the online content — in particular, creative, interactive tools such as App Lab programming tool that rely on blocks-based rather than text-based code — are not screen-reader accessible and do not adhere to accepted web accessibility standards. They found the problem extended offline, where the so-called unplugged activities that supplement the online content heavily relied on visual metaphors and artifacts to convey concepts. In one example cited by the researchers, the “Flippy Do” template that teaches students binary numbers requires students to cut and fold a piece of paper along printed lines, then fill in blank spaces — an activity that is not accessible to students with visual impairments or certain other physical disabilities.

At Code.org’s invitation, Ladner and his collaborators spent more than a year modifying the CSP curriculum, with a particular focus on developing accessible alternatives to the visually-based unplugged activities. They also partnered with external experts to develop accessible interactive tools to replace those that were inaccessible in the original online curriculum, using the Quorum web-based programming environment that has accessibility as a primary design goal, and made the source code available to teachers and students to customize the tools to better meet their needs.

To understand how the original curriculum and proposed modifications would work in the wild, the team organized a professional development workshop with teachers having varied experience in developing or adapting computer science curriculum for students who are blind or visually impaired. The teachers were drawn from different settings, including residential schools and public school districts, to gain diverse perspectives on how the CSP curriculum is used and understand the challenges educators face in adapting content for their students. During the workshop, which took place over the course of a week, the research team gathered feedback from participants on Code.org’s online CSP curriculum and tools, the College Board’s procedures with regard to students who are blind, and its proposed modifications for making other implementations of CSP more accessible. Recommendations included encouraging teachers to make modifications themselves using an array of accessibility tools, and to share those modifications with others, and urging AP CSP curriculum providers to arrange for a complete evaluation of their content to be performed by a qualified accessibility professional and make necessary modifications to better serve the needs of teachers and students. Participants also emphasized the need for the College Board, which oversees the AP CSP exam, to clarify its procedures governing practice examples, which tend to be purely visual in nature, and the accommodations permitted during the exam, and to alter the use of Braille math on the exam to account for the existence of two, competing standards.

“If we are to embrace equity and computer science for all, we have to really mean it. Accessibility should be the default, not an afterthought,” said Ladner. “I give Code.org a lot of credit for opening up their curriculum to scrutiny and for working with us to identify strategies for making it truly accessible to all. Our hope is that this effort will serve as a blueprint for other curriculum providers to improve the accessibility of their content and increase educational opportunities for a population that is often overlooked in online curriculum development.”

The SIGCSE conference attracts more than 1,500 educators, researchers and practitioners interested in improving computer science education at the K-12 and postsecondary levels. Read the team’s research paper here, and learn about Code.org’s new accessible CSP curriculum here. Last December, Code.org recognized Ladner, Stefik and their collaborators at AccessCSforAll as part of its Champions for Computer Science Awards for their leadership in making CS education accessible to all students.

Way to go, Richard, Andreas, William and Sean!

Allen School Ph.D. student Minjoon Seo has earned a 2019 Facebook Fellowship for his research in artificial intelligence and natural language processing. Seo develops systems for end-to-end question answering, reasoning-based question answering, and scalable deep-learning architectures in collaboration with Hanna Hajishirzi and Ali Farhadi, professors at the Allen School and researchers at the Allen Institute for Artificial Intelligence (AI2).

“Minjoon is a rising star in NLP and has made substantial contributions to the field of question answering, one of the most important areas in AI and natural language processing,” said Hajishirzi. “He tends to work on high-impact projects and has already earned more than 600 citations for his research.”

One of those high-impact projects is BiDAF — short for Bi-Directional Attention Flow for Machine Comprehension — which Seo worked on as a research intern at AI2 and is now widely used in academia and industry. With BiDAF, Seo and his collaborators have built a large-scale, end-to-end, deep neural question answering system that is capable of answering questions about both textual paragraphs and diagrams. The system is modular, enabling researchers and practitioners to extend it to multiple modalities and knowledge bases. Seo’s work has proven so popular, BiDAF ranked first on the leaderboard for three important new reading-comprehension datasets shortly after its release and is actively being used in many question answering systems in academia and industry.

Seo was lead author on the groundbreaking GeoS question answering system, the first end-to-end system for solving high school-level plane geometry problems developed by AI2 and Allen School researchers. GeoS demonstrated the ability to jointly understand textual and visual cues to answer SAT geometry questions as well as the average 11th-grade test-taker.

When GeoS was publicly announced, Farhadi noted in a press release that the biggest challenge for the team was to convert the SAT question into language a computer could understand. “One needs to go beyond standard pattern-matching approaches for problems like solving geometry questions that require in-depth understanding of text, diagram and reasoning,” he explained. Among Seo’s contributions to that effort was a new system for using visual cues in diagrams to improve understanding of the accompanying text. His work incorporated novel computer vision techniques for parsing diagrams and natural language processing algorithms for accurately parsing text, as well as a new algebraic solver that can solve the question based on a combination of textual and visual cues.

More recently, Seo has turned his attention to the design of algorithms for open-domain, scalable question-answering systems. For example, developed a tool called Skim-RNN that improves the scalability of recurrent neural networks using a technique inspired by the human approach to speed reading. Seo also worked on a new approach to question-answering during an internship with Google AI that requires a standalone representation of the question and the context. His method — known as Phrase-Indexed Question Answering, or PIQA — has the potential to significantly improve the scalability of question answering in addition to enabling more efficient retrieval.

Seo is among just 21 researchers recognized in Facebook’s 2019 Fellowship competition. The prestigious Facebook Fellowship Program recognizes and supports promising doctoral students at universities around the world who are engaged in innovative research spanning 15 areas of Computer Science and Computer Engineering. Past Allen School recipients include James Bornholt and Eunsol Choi (2018), Aditya Vashistha (2016), and Lydia Chilton and Nicola Dell (2013).

Learn more about the 2019 Facebook Fellows here.

Congratulations, Minjoon!

Each year, the Computing Research Association (CRA) recognizes rising young research talent at colleges and universities across North America through its Outstanding Undergraduate Researcher Awards. As part of its 2019 competition, the CRA honored four Allen School undergraduates — Nelson Liu, Kimberly Ruth, Pathirat Kosakanchit, and Rowan Phipps — for their contributions in natural language processing, security and privacy, and technology for development and for demonstrating the potential to have an even greater impact in their future careers.

Nelson Liu is one of two Allen School students who were named finalists in this year’s CRA awards. Liu is a senior majoring in Computer Science and Linguistics whose research focuses on machine learning and natural language processing (NLP). Working under the guidance of Allen School professor Noah Smith, Liu has explored the linguistic capabilities and limitations of recurrent neural networks (RNNs), which form the backbone of state-of-the-art NLP systems. His work breaks into the “black box” of neural network NLP models to advance our understanding of their inner workings and abilities, a step toward his broader goals of improving the generalizability and interpretability of such models.

For one project, Liu and his collaborators at the Allen Institute for Artificial Intelligence (AI2) and University of Colorado Boulder found that long short-term memory recurrent neural networks (LSTMs) trained on natural language data outperform those trained on non-language sequential data in solving memorization tasks, demonstrating that recurrent neural networks exploit the linguistic attributes of data even while performing inherently non-linguistic tasks. The results earned Liu and his team a Best Paper award at the Association for Computational Linguistics’ 2018 Workshop on Representation Learning for NLP (RepL4NLP 2018). Liu’s recent work has also examined the linguistic knowledge implicitly encoded within contextualized word vectors by assessing their ability to predict a broad range of linguistic features of input text, and investigated the ability of NLP models to adapt to adversarial datasets in order to gauge their effectiveness in evaluating model generalization. Papers about both of these projects were recently accepted for presentation at the upcoming annual conference of the North American chapter of the ACL (NAACL 2019).

Liu previously earned a prestigious Goldwater Scholarship, which recognizes undergraduates who intend to pursue a research career in the natural sciences, mathematics, or engineering, and Washington Research Foundation Fellowships in 2017-18 and 2018-19 as part of a program supporting undergraduates who engage in sophisticated research projects that involve creativity and advanced subject knowledge.

Kimberly Ruth was named a finalist in the 2019 CRA competition based on her work aimed at improving the security and privacy of augmented reality applications. Ruth, a senior in the Interdisciplinary Honors Program pursuing a double-major in Computer Engineering and Mathematics, has been working with Allen School professors Tadayoshi Kohno and Franziska Roesner in the Security and Privacy Research Laboratory since her freshman year.

Ruth’s first foray into AR security focused on addressing the safety and security risks of unregulated application output, such as content that interferes with their ability to view important real-world information or distracts users from critical tasks such as driving. The research team developed, prototyped and evaluated Arya, an AR platform that enables developers to specify security policies that will prevent malicious or buggy output. Working with Arya collaborator Kiron Lebeck, a Ph.D. student in the Allen School, Ruth then turned her attention to multi-user scenarios such as AR-enabled collaborative tools and multi-player games. The team designed and conducted a user study to better understand the the privacy and security challenges of these increasingly popular applications using the Microsoft HoloLens platform. Based on her experience with both projects, Ruth became interested in how technical design solutions can be used to help developers build multi-user AR apps that protect against user-level threats. To that end, she developed a set of design principles and prototype for a multi-user AR module that would enable users to control access to sensitive information, filter inappropriate content, and engage in both co-located and remote content sharing.

Ruth was previously recognized with a 2018 Goldwater Scholarship, a Mary Gates Research Scholarship, and Washington Research Foundation Fellowships in 2017-18 and 2018-19. Last year, she was named a member of the Husky 100, which celebrates the contributions of University of Washington students who are making the most of their Husky experience. Ruth was also selected as a finalist in the CRA’s 2018 Outstanding Undergraduate Researcher competition.

Pathirat “Pat” Kosakanchit, a junior majoring in Computer Engineering, earned an honorable mention from CRA for her research as a member of the Allen School’s Information and Communication Technology for Development (ICTD) Lab. Spurred by the internet’s ability to advance social, educational, and economic opportunity, Kosakanchit has spent the past year working on the Community LTE (CoLTE) project under the guidance of professor Kurtis Heimerl and postdoc Spencer Sevilla.

CoLTE is an open-source LTE software package that enables rural communities lacking retail internet access to establish and maintain their own network. Kosakanchit’s technical contributions to the project include testing and debugging the network stack, and building a Node.js application that enables users to check their data usage, top up their account, and transfer credit to fellow users. In recognition of the unique challenges of remote communities in a network outage, she also built an offline, locally-hosted version of OpenStreetMap (OSM) that provides residents with access to basic mapping services when they are unable to access the internet. To implement the offline version, Kosakanchit had to overcome several configuration issues relating to virtual machine and database setup. She also designed and implemented a set of map markers signaling fire, water, food, and medicine that users could pin to the offline version of OSM during emergencies. The team has launched a pilot CoLTE network in two locations: the small Indonesian village of Bokondini and the town of Santa Inés in Oaxaca, Mexico. Additional networks are planned to be deployed in Washington state, Argentina, and potentially in Kosakanchit’s native Thailand.

Kosakanchit has applied her mapping skills outside of the ICTD Lab as a member of WOOF3D, the UW’s 3D-printing club, where she contributed to the development of a simplified process for 3D printing topological maps. Kosakanchit also serves as a mentor for the International Student Mentorship Program, which helps international students adjust to life in the United States, and is an active member of the Thai Student Association at UW.

Rowan Phipps, a senior majoring in Computer Science who also works with Heimerl in the ICTD Lab, received an honorable mention for his work on mobile security as part of the Digital Financial Services Research Group, a collaboration between members of the ICTD Lab and Security and Privacy Research Lab. Phipps’ research focuses on the security implications of devices and services used for mobile banking in communities where people lack access to traditional financial services.

Phipps and his colleagues were particularly interested in the growing use of ThinSIMs, which are small, programmable add-ons to a mobile SIM card that make it easier to deploy additional applications. Many mobile banking services are Slim Toolkit (STK) applications, which run directly on a SIM card. STK apps adhere to the GSM standard and are compatible with every mobile phone, but they require installation by the SIM card manufacturer. New market entrants can use ThinSIMS to get around this hurdle and provide alternatives to the services offered by incumbent providers. The disadvantage to ThinSIMs, as Phipps and his colleagues discovered, is that they can be used to intercept and initiate communications over the SIM interface without the user’s knowledge or consent, redirecting their credentials, their calls — and potentially their money — to an attacker. The team discovered a similar vulnerability with another popular protocol for delivering mobile money services, known as unstructured supplementary service data or USSD, when paired with a ThinSIM. In a paper that appeared at the Association for Computing Machinery’s 1st annual Conference on Computing & Sustainable Societies (COMPASS 2018), the team revealed the vulnerabilities associated with ThinSIMs using attacks Phipps designed to exploit mPesa, a popular STK-based app, as well as a USSD-based service he created for the experiment. Phipps and his collaborators also recommended modifications to existing apps and the GSM standard to protect users against such attacks.

Phipps was a member of the UW student team that captured first place in last year’s Pacific Rim Collegiate Cyber Defense Competition. He previously completed a technical internship at Qumulo and, before that, spent a summer as a teaching assistant for Girls Who Code.

Learn more about the CRA Outstanding Undergraduate Researcher Awards here.

Congratulations to Nelson, Kimberly, Pat, and Rowan!

Allen School professors Alvin Cheung and Shayan Oveis Gharan were named 2019 Sloan Research Fellows in Computer Science by the Alfred P. Sloan Foundation. The prestigious Sloan Research Fellowship Program recognizes early-career scientists and engineers who have already distinguished themselves through their research and exhibit the potential to make substantial contributions in their respective fields.

Alvin Cheung engages in cross-disciplinary research as a member of the Allen School’s Database and Programming Languages & Software Engineering groups. In his young career, Cheung has produced multiple, paradigm-shifting solutions spanning data management, data analysis, and end-user programming.

“From booking plane tickets to browsing social networking websites, we interact with large amounts of data everyday,” noted Cheung. “My group works on new techniques to help users process and manage data easily, with the goal to simplify software developers’ efforts to build databases and applications without compromising on performance, and enable the rapid development of database applications that provide efficient and reliable data access to all.”

One of Cheung’s key early-career contributions has been his pioneering work on verified lifting, a technique for automatically translating applications written in Java to domain-specific languages such as SQL, Spark, and Hadoop to optimize performance and reduce errors. For example, by consolidating application logic into compact SQL queries in which the SQL engine could identify optimization opportunities, Cheung’s approach increased the speed of applications up to 1,000 fold. The technique also has the effect of “future-proofing” applications driven by big-data systems that are subject to frequent updates.

For another project, Cheung and his collaborators developed methods in theorem proving and model checking to produce Cosette, an automated prover for checking complex SQL queries that can identify bugs contained in hundreds of manually written rules within seconds. Cheung has also contributed to dramatic leaps in end-user programming; for example, he helped develop Scythe to synthesize SQL queries based on input-output examples posted on Stack Overflow by users seeking expert help with writing SQL queries. Their algorithm can answer around 70% of the most SQL-related questions on the platform faster than the human experts can, making it the best SQL synthesizer ever developed.

“Alvin stands out for his interdisciplinary approach and keen intuition regarding how systems are likely to perform, which has enabled him to crack problems that appear impossible to solve,” said Allen School Director Hank Levy. “His breakthrough work on verified lifting and other projects will have a tremendous impact on the functions that future systems will be able to deliver.”

Shayan Oveis Gharan is a member of the Allen School’s Theory of Computation group who focuses on the design and analysis of efficient algorithms for solving fundamental NP-hard counting and optimization problems at the heart of the theory and practice of computing. These problems have implications for a wide range of fields, from logistics and marketing, to planning and policy-making, that cry out for new and better computational tools for managing and exploiting the vast quantities of data available.

“I encode a discrete phenomenon in a complex multivariate polynomial, and I understand it via the interplay of the coefficients, zeros, and function values of this polynomial,” explained Oveis Gharan. “Although these polynomials are so large that they cannot be stored in all computers in the world combined, I use their analytical properties to design efficient optimization algorithms for the underlying discrete phenomenon.”

Among Oveis Gharan’s most notable contributions to date are his works on the Traveling Salesman Problem (TSP) and its asymmetric variant — one of the most studied problems in optimization — and his very recent work on counting problems related to matroids. Oveis Gharan and his collaborators studied TSP using analytical techniques, proposing a new class of algorithms for variants of TSP and introducing novel analysis of classical algorithms for this problem dating back 50 years. The team’s efforts produced the first improvement on existing approximation algorithms which broke barriers that had withstood for three decades despite substantial previous attempts within the theoretical computer science community.

Oveis Gharan’s most recent work on counting bases of matroids has profound applications in many areas, such as network reliability. To illustrate, he cited the road network of Seattle, which became heavily blocked due to a recent snowstorm. If each street x in the city will be blocked with probability px, what is the probability that the whole city will be disconnected — that there is no available route from home to work for some residents — and how should the city position its snow plows to minimize the probability of such an event? Oveis Gharan and his group devised efficient algorithms to answer such questions.

“Shayan’s contributions in combinatorial optimization, such as his work on the Traveling Salesman Problem, have had a profound impact on the theoretical computer science community,” observed Levy. “His creative and driven approach has enabled him to break through long-standing barriers and expand our understanding of fundamental theoretical problems that underpin our field.”

Joining Cheung and Oveis Gharan in the class of 2019 Fellows is professor Kelley Harris of the UW Department of Genome Sciences, who was recognized in the computational and Evolutionary Molecular Biology category for her research into the evolutionary history of humans and other species based on large datasets of genetic variation. Each year, the Sloan Foundation selects a total of 126 Fellows from higher education institutions across North America who are making contributions in Chemistry, Computational and Evolutionary Molecular Biology, Computer Science, Economics, Mathematics, Neuroscience, Ocean Sciences, Physics, or a related field.

“Sloan Research Fellows are the best young scientists working today,” said Adam F. Falk, president of the Alfred P. Sloan Foundation, in a press release. “Sloan Fellows stand out for their creativity, for their hard work, for the importance of the issues they tackle, and the energy and innovation with which they tackle them. To be a Sloan Fellow is to be in the vanguard of 21st century science.”

Recent Sloan Fellowship recipients at the Allen School include Maya Cakmak, who was recognized last year for her work in robotics; Ali Farhadi and Jon Froehlich, who were included among the class of 2017 Fellows for their work in artificial intelligence and human-computer interaction, respectively; and Emina Torlak for her contributions to computer-aided verification and synthesis. A total of 35 current or former Allen School faculty members have been recognized through the fellowship program.

Read the Sloan Foundation announcement here, the complete list of 2019 Fellows here, and a related UW News story here.

Congratulations to Alvin, Shayan, and Kelley!

Yesterday, nearly 300 friends of the Paul G. Allen School gathered to celebrate the dedication of the new Bill & Melinda Gates Center for Computer Science & Engineering on the UW Seattle campus. UW and Allen School leadership were joined onstage by special guests Bill Gates, Washington Governor Jay Inslee, and Microsoft President & Chief Legal Officer Brad Smith in thanking the more than 500 individual donors, local technology companies, and state taxpayers for their support of the project.

The day marked not only the opening of a new building, but also the prospect of a new era of computer science education and impact. “The Gates Center isn’t just a building, it’s a statement about our vision of the future,” said Allen School Director Hank Levy. “We have created a world-class computer science program here, in part because of our focus on technology that helps to solve the world’s biggest challenges. This building enables us to grow those efforts and tackle even bigger challenges.”

One of the ways in which the new building will enable the school to grow its impact is by providing the physical capacity to serve more students. “It’s first and foremost a vehicle for increasing opportunity for Washington’s students,” said UW President Ana Mari Cauce. “Because that’s what we do here, in this outstanding building and all across our campuses. It’s what drives the talented faculty who teach our students and advance discovery. It’s why so many students are eager to come here to learn — both to the University and to this program.”

Cauce highlighted the care and thought that went into the design of the building, which is focused on increasing student capacity and enriching the student experience. The building’s entire first floor is dedicated to the needs of Allen School majors, including a student services center, an undergraduate commons, meeting and collaboration spaces, computer labs and support, and capstone project rooms. The Gates Center also provides much-needed instructional space, in the form of several large classrooms and seminar rooms along with a 240-seat auditorium, and numerous collaboration and community spaces — including a cafe open to the campus and the public, and a 3,000-square-foot events center for hosting workshops, research demos, career fairs, and other community-oriented events.

Emphasizing that even more important than the new building is what goes on inside, Brad Smith — who led the fundraising campaign for the building — saluted the assembled guests who helped make the vision of the Bill & Melinda Gates Center a reality. “It is extraordinary to just look across this room and see the faces of all of the people who contributed so much,” Smith said. “You didn’t hang up when the phone rang. We rolled up our sleeves together. More than 500 people donated their personal funds to this building.” Among them, he noted, were the Friends of Bill & Melinda Gates, a group of more than a dozen couples, led by Charles and Lisa Simonyi, who joined Microsoft in providing a gift to name the building in the Gateses’ honor.

Smith also acknowledged local companies such as Amazon and Zillow that joined Microsoft in supporting the project. “It was a wonderful journey that gave many of us an opportunity to partner together — even competitors,” he noted. “Other companies in Seattle really stepped up.”

In addition to enabling the Allen School to serve more students, those efforts will also help the school and the UW to stay at the forefront of the computing field. Laboratories in the Gates Center include a wet lab to support the school’s work at the intersection of information technology and molecular engineering; the UW Reality Lab, which focuses on advancing the state of the art in augmented and virtual reality education and research; the Center for Neurotechnology, which aims to use technology to revolutionize the treatment of debilitating neurological conditions; and a 3,000-square-foot robotics laboratory.

Governor Jay Inslee — a proud Husky — noted that the impact of the new building and its inhabitants would extend far beyond their core focus. “This is so much beyond the world of computing,” Inslee said, “because the world of computing feeds every single thing that we’re growing our economy and our society on now.”

Of the 500 donors Smith mentioned, roughly 300 are Allen School alumni. That means nearly 200 people without an alumni connection to the school recognized the potential impact and threw their support behind the project. “From the bottom of my heart, thanks to all of you for what you’ve done to make this amazing building a reality,” said Allen School professor Ed Lazowska.

Before the program began, Lazowska and Levy had taken Bill Gates, Cauce and Smith on a tour of the new building, visiting several labs and stopping to talk with students along the way. “It was fantastic to take the tour and not only see that it’s an incredible building, but to see some of the great work going on here,” Gates said to the assembled crowd later. Saying that he and Melinda are honored to have a building named after them that will increase the capacity of the school associated with Paul G. Allen, Gates also paid tribute to his late friend and collaborator who passed away last October.

“It would’ve been great if Paul could have been here,” he said. “He deserves so much credit for what happened at Microsoft and always believing in innovation and believing in the University of Washington. So hopefully he somewhere can appreciate the great development that is taking place here.”

Read the UW News release here, GeekWire articles on the building opening here and the dedication here, and a related Seattle Times editorial here.

View more photos of the day’s events below.

The Allen School family mourns the passing of professor emeritus Ted Kehl, one of the original group of seven faculty members who founded the Department of Computer Science at the University of Washington in the 1960s, and someone who was instrumental in introducing Seattle to a hardware revolution in the form of very large-scale integrated (VLSI) circuit design — a development that transformed computing and modern society.

Kehl earned his Ph.D. at University of Wisconsin-Madison in 1961 and joined the UW faculty six years later. As a professor in both Computer Science & Engineering and the School of Medicine, Kehl was known for his interdisciplinary approach to research long before it became fashionable. In addition to teaching physiology, biophysics and computer science, Kehl ran a hardware lab focused on the development of efficient computer hardware for biomedical research.

He took a keen interest in VLSI research and worked with Carver Mead of CalTech — who co-wrote the book on VLSI design along with Lynn Conway — to bring a new VLSI course to Seattle, jointly sponsored by UW and Boeing. That effort led to the establishment of the University of Washington/Northwest VLSI Consortium, a major initiative funded by the Department of Defense’s Advanced Research Projects Agency (ARPA) and led by Kehl in partnership with five regional companies. The course helped spawn the MicroVAX-I processor, which was developed by Digital Equipment Corporation under the leadership of Dave Cutler, then-head of the DECwest Engineering team. Kehl subsequently launched an undergraduate seminar in VLSI design in which teams of students had the opportunity to work with professional mentors drawn from the semiconductor industry on their designs.

In 1983, Kehl co-founded a startup company called IC Designs with his former Ph.D. student John Torode. IC Designs started out providing software and fabrication services for application-specific integrated circuits (ASICs) before its focus evolved into the design of ASICs for use in personal computers. The company grew to over $30 million in annual sales before its acquisition by industry leader Cypress Semiconductor Corp. in 1993.

Kehl retired from UW in 1997.

Our thoughts are with Ted’s family, friends, and the many students and collaborators whose lives he touched during his long and influential career.