Allen School News

In the late 1990’s, members of the Allen School faculty experimented with a new — some would say unorthodox — way to mark the conclusion of Visit Days, the annual pilgrimage made by prospective graduate students to computer science programs around the country. To commemorate the visitors’  time in Seattle, professors in what was then the Department of Computer Science & Engineering would cheerfully send them on their way with a surprise parting gift: a palm-sized chunk of concrete.

The concrete in question had, without any human intervention, become dislodged from the crumbling facade of Sieg Hall — the building that, should the recipients choose the University of Washington, would become a home away from home for the duration of their Ph.D. 

“The souvenir definitely made us memorable, and it helped our cause when it came to recruitment,” Allen School professor Ed Lazowska, who chaired the department at the time, recalled wryly. “One student emailed that they just couldn’t say ‘no’ to us after we literally gave them a piece of our building. But giving out chunks of the building, like the building itself, was a joke. We were woefully behind other top computing programs when it came to facilities.”

While outside the building was crumbling, inside it was cramped — so much so that, as a prank, someone set up a “graduate student office” on a ledge in the stairwell, complete with a handy rope ladder for access. More than two decades after it first housed UW’s burgeoning Computer Science & Engineering program, Sieg was no longer fit for the purpose. In 1999, the department stepped up its campaign for a new, permanent home.

Lazowska and local technology industry leaders led the charge, forging a public-private partnership that was unprecedented in UW’s history. All told, they raised $42 million in private funds — substantially more than half the project’s cost — from more than 250 donors. Lazowska’s faculty colleague Hank Levy oversaw the design and construction of the building in tandem with LMN Architects and general contractor M.A. Mortenson. He saw to it that the funds were put to good use.

“Our goal was to create a warm and welcoming environment that would facilitate teaching, research and collaboration,” said Levy. “Every aspect of the building — the materials, the artwork, the abundant natural light, the open spaces that encourage people to gather and exchange ideas — were intentional choices made with this goal in mind.”

Those choices were supported in large part by leadership gifts from the building’s namesake, the late Paul G. Allen, along with the Bill & Melinda Gates Foundation and Microsoft. Completion of the 85,000 square-foot facility, which was dedicated on October 9, 2003, tripled the program’s available space and set off a chain of events that made the Allen School into the powerhouse it is today. 

Allen himself understood at the time that he was investing in something more meaningful than bricks and mortar.

“I’m proud to have supported this beautiful and unique facility, but what really sets UW’s computer science program apart are the people,” Allen observed during the grand opening celebration. “The faculty here is unparalleled, and the undergrad and graduate students are dedicated and inspiring.”

Allen’s faith would inspire a period of expansion that no one — including Lazowska, who has been the program’s most vocal cheerleader over the years — could have foreseen in 2003. 

“I cannot stress enough the importance of the Allen Center to the trajectory of our program,” he said. “It provided us with competitive space for the first time in our history. It was the spark that set us on a path to triple our degree production, ramp up our ability to deliver computer science education to students across campus, and attract the brightest researchers in the field to Seattle.

“And in the midst of all that,” Lazowska added, “we became a full-fledged school!”

On move-in day in the summer of 2003, fewer than 40 faculty members unpacked boxes in their shiny new offices; two decades later, that number is approaching 100. And faculty recruiting has barely kept pace with the explosive growth in student interest, with the Allen School the most requested major among freshman applicants to the University for several years running. It now serves roughly 2,800 students across its degree programs — and thousands more who take one or more courses as non-majors each year.

As the program grew in size, it also grew in stature, thanks in no small part to its new and improved laboratory space.

“Computer Science & Engineering at the University of Washington is an engine of opportunity,” Allen had said at the time, “and I want to ensure it’s an even more cutting-edge resource for the coming generation.”

That engine has been going full throttle ever since. One high-profile example of how the move to the Allen Center greased the wheels of innovation is UW’s emergence as a center for mobile health. By tapping into the built-in sensing capabilities of smartphones coupled with advances in machine learning, Allen School researchers, in conjunction with UW Medicine clinicians, have developed a range of mobile tools for screening and monitoring of a variety of health conditions spanning fever, pre-diabetes, sleep apnea, infant jaundice, reduced lung function, ear infection, newborn hearing loss and more. All got their start in the Allen Center’s labs, and several led to the creation of Allen School spinout companies.

The collaborations don’t stop there, as the Allen Center provided a launch pad for multiple cross-campus initiatives, some supported by significant federal and/or private investment. These include efforts to advance accessible technologies and more accessible communities, data science for discovery and innovation, neurotechnologies for people with spinal cord injury, stroke or other neurological disorders, next-generation cloud computing infrastructure, computing for environmental sustainability and more. In the past five years alone, the Allen School has secured more than $200 million in grants and contracts to support its research. Along the way, the school has strengthened its leadership in core areas such as systems, architecture and theoretical computer science even as it has expanded its expertise to encompass new areas, including cryptography, molecular programming, quantum computing and natural language processing.

And that list is by no means exhaustive. 

“We took Paul’s words to heart, and the impact of the community’s investment continues to be felt today far beyond the Allen Center’s walls,” said Magdalena Balazinska, director of the Allen School and Bill & Melinda Gates Chair in Computer Science & Engineering. “It is felt through the graduates we’ve mentored, the technologies we’ve developed, the companies we’ve started, the opportunities we’ve created, and the leadership we’ve provided.”

The growth sparked by the Allen Center eventually led UW to break new ground in computing literally as well as figuratively; nearly 16 years later, with its first building now bursting at the seams, the Allen School dedicated its second building, the Bill & Melinda Gates Center, which doubled its physical space.

That additional space came just in time, too. Thanks to advocacy by the University and additional investments from the state legislature, the school is currently on track to award 820 degrees annually and has cemented its place in the top echelon of computer science programs in the nation.

“I said back then that the true measure of this building will be what we do inside to take our programs to the next level of excellence,” said Levy. “I’d like to think that we lived up to that promise, and then some.”

For more on the Allen Center’s history, see the Allen Center dedication brochure, a special pre-dedication insert in the Most Significant Bits newsletter, and the dedication issue of MSB from fall 2003.

Read more →

In 2020, a group of researchers unveiled a tool called Face Depixelizer that would take a low-resolution image as an input and, with the help of a generative machine learning model called StyleGAN, produce a high-resolution image in its place. But the model, which was not designed to “fix” the original low-quality image but instead generate an imaginary replacement, had a tendency to predominantly imagine white people — even when the original image depicted someone of another race.

The following year, a group of web developers and accessibility experts signed an open letter urging website owners to avoid using accessibility overlays on their sites. The signatories had become alarmed by the growing reliance on these automated tools, which are marketed under the guise of helping website owners improve the user experience while avoiding potentially costly litigation, when it became apparent that they can actually make the experience worse for people with screen readers — to the point of making a site unusable. To date, nearly 800 individuals have added their names to the letter.

These are just two examples of how technology can have unforeseen, and ostensibly unintended, negative consequences in the real world. Spurred on by these and other cautionary tales, a team of researchers at the Allen School want to assist their colleagues in anticipating and mitigating the consequences of their own work. With support from a five-year institutional transformation grant through the National Science Foundation’s Ethical and Responsible Research (ER2) program, the team hopes their project will usher in a new paradigm in computing-related research not just at the University of Washington, but across the field.

One member of the team, Allen School Ph.D. student Rock Yuren Pang, already had begun thinking about how society increasingly bears the brunt of unintended consequences from new technologies. After enrolling in a graduate-level computer ethics seminar taught by professor Katharina Reinecke, he began to fully appreciate the difficulties researchers face in attempting to understand, let alone mitigate, what those might be.

“Emerging technologies are being used for a growing range of applications that directly impact people’s lives — from how communities are policed, to which job applicants are called for an interview, to what content someone sees online,” Pang said. “As a young Ph.D. student, I thought the question of how we as researchers might think about the downstream impacts of our work to be a really important problem. But I also felt overwhelmed and didn’t know how to even begin tackling it.”

In a new white paper, Pang, Reinecke and Allen School professors Dan Grossman and Tadayoshi Kohno offer a potential starting point. Dubbed PEACE — short for “Proactively Exploring and Addressing Consequences and Ethics” — their proposal offers a vision for empowering researchers to anticipate those consequences “early, often, and across computer science.” 

The latter is important, Reinecke notes; while artificial intelligence may dominate the headlines at the moment, these issues extend throughout the field.

“We can’t just point fingers at AI; every technology, no matter how seemingly benign, has the potential to have undesirable impacts,” said Reinecke, the PI on the NSF grant whose research in the Allen School’s Wildlab includes investigating how people relate to technology differently across languages, cultures and abilities. “When we interviewed researchers across multiple subfields, they generally acknowledged the importance of trying to anticipate the consequences of innovation. But to translate that into practice, they need some scaffolding in place.”

To that end, Reinecke and her co-authors propose a holistic approach that would weave such considerations into the school’s teaching and research while making it easier for researchers to tap into existing resources for assistance in anticipating and mitigating undesirable impacts. Two of the resources the team intends to explore as part of the NSF grant, the Tarot Cards of Tech and guided stakeholder analysis, have Seattle roots. The latter is a pillar of Value Sensitive Design, co-conceived by UW iSchool professor and Allen School adjunct faculty member Batya Friedman, that engages individuals or groups who could be directly or indirectly affected by technology. As part of the process, researchers could save the results of their analysis in the form of a PEACE report that could be shared with collaborators on a particular project and updated anytime.

Researchers will also have the option to share their PEACE reports with an ethics board comprising faculty colleagues from across campus with expertise in areas such as law, bioethics, science and technology studies, and gender, women and sexuality studies. Members of this group will act as a sounding board for researchers who wish to follow up on the results of their exploration — and help them think through how they could address any potential unintended consequences they’ve identified in the process.

As with other elements of the proposed PEACE process, consultation with the ethics board would be entirely voluntary.

“We want to give researchers a low-friction, low-stakes mechanism for seeking diverse perspectives on how a technology might be used or misused. This could help surface potential implications we may not think of on our own, as computer scientists, that can inform how we approach our work,” Reinecke said. “We aren’t saying ‘don’t do this risky piece of research.’ What we’re saying is, ‘here’s a way to anticipate how those risks might manifest’ in order to mitigate potential harm.”

In his role as co-director of the Allen School’s Security and Privacy Research Lab and the Tech Policy Lab at UW, Kohno has had ample opportunity to analyze the harm that can result when researchers haven’t thought ahead.

“Many times during my career have I wondered if the original researchers or developers could have prevented a problem before deployment,“ said Kohno. “For years, I and my colleagues have encouraged the people who build new technologies to apply a security and privacy mindset from the start rather than having to fix vulnerabilities later, after damage has been done. That’s essentially what we’re suggesting here — we’re asking our colleagues to apply a societal mindset, and to front-load it in the research process instead of relying on hindsight, when it may be too late.”

Grossman is vice director of the Allen School and often teaches the undergraduate computer ethics seminar, which the school began offering to students on a quarterly basis in 2020. He sees an opportunity for the PEACE project to eventually transform computing education and research on a massive scale. 

“We are in a position to guide the future leaders of our field toward thinking not only about the technical aspects of computing, important as they are, but also the ethical ones — to train future researchers and technologists how to rigorously consider the potential ramifications socially, politically, environmentally, economically or any combination thereof,” said Grossman. “We need the people who understand their proposed technology to grapple with these issues as well as to learn how to interact with non-technologists, such as public-policy experts, who have complementary expertise.”

The team will deploy and evaluate the PEACE project within the Allen School to start, with plans to extend access to other academic units on campus in later years. Eventually, Pang and his colleagues plan to distill the findings from their evaluation of the UW deployment into detailed design guidelines that can be adapted by other institutions and companies.

“I want to create the go-to place for UW researchers to learn, anticipate and bounce ideas off other researchers about the potential consequences of our work,” Pang said. “But I hope this initiative encourages a broader culture in which computer scientists are unafraid to think critically and openly about these issues. And I believe we can do it in a way that supports, not stifles, innovation.”

Read the team’s white paper here. This work is supported by National Science Foundation award #2315937. Read more →

In the days before single-cell RNA sequencing, researchers investigating the mechanisms and treatment of disease had to make do with running experiments on bulk cell profiles created by taking tissue samples and grinding them up, “sort of like putting them in a blender,” in the words of Allen School Ph.D. student Ethan Weinberger.

That milkshake may have brought all the biomedical scientists to the lab, but the bulk sequencing technique limited them to studying aggregations of populations of cells, with no way to distinguish among individual cell types. Nowadays, researchers can take measurements at the level of individual cells, enabling the exploration of such finer-grained distinctions and advancing our understanding of various biological functions. But without the right computational tools, even single-cell datasets can yield distinctions without a difference.

Weinberger is a member of the Allen School’s AIMS Lab, where he works with fellow Ph.D. student Chris Lin and professor Su-In Lee to leverage advances in artificial intelligence to help scientists get the most out of these increasingly robust datasets. In a paper published this week in Nature Methods, the team introduced ContrastiveVI, a deep learning framework for applying a powerful technique called contrastive analysis, or CA, to single-cell datasets to disentangle variations in the target, or treatment, cells from those shared between target and control cells when running experiments. 

“Scientists want to investigate questions like ‘How does perturbing this particular gene affect its response to a pathogen?’ or ‘What happens when I hit a diseased cell with such-and-such a drug?’,“ explained Weinberger. “To do that, they need to be able to isolate the variations in the cell data caused by that perturbation or that drug from those that are shared with a control dataset. But existing models can’t separate those out, which might lead someone to draw erroneous conclusions from the data. ContrastiveVI solves that problem.”

CA has proven effective at this type of isolation in other contexts, but its utility in relation to single-cell datasets has so far been limited. That’s because existing computational models for analyzing single-cell data mostly rely on a single set of latent variables to model all variations in the data, effectively lumping them all together and precluding the ability to perform CA.

ContrastiveVI is the first deep learning model designed for performing CA on single-cell data. Unlike other approaches, the ContrastiveVI model explicitly separates latent variables into two categories, each with their own encoding function: shared variables, or those that are found in both the target and control cells, and salient variables, which are found exclusively among the target cells. 

It is that second category that will excite scientists testing potential cancer drugs or analyzing the role of gene expression in the body’s response to disease. 

“ContrastiveVI effectively distinguishes the factors that are salient — that is, relevant — to an experiment from confounding factors. This enables us to capture variations that are unique to the treated cells,” said Lee, senior author of the paper and holder of the Paul G. Allen Career Development Professorship in the Allen School. “ContrastiveVI will reveal tiny but important variations in the data that may be obscured by other models.”

Lee and her co-authors validated ContrastiveVI using real-world datasets with previously verified results as their ground truth. In one experiment, the researchers applied ContrastiveVI to a target dataset of measurements taken from two dozen cancer cell lines treated with idasanutlin. This small-molecule compound has shown therapeutic potential owing to its activation of a tumor-suppressing protein in wild type — that is, unmutated — TP53 genes. The team used ContrastiveVI to analyze data on both wild type and mutated TP53 cell lines, which are non-responsive to idasanutlin, using a background dataset from the same cell lines treated with a different compound, dimethyl sulfoxide, as the control. 

“A good result — one that agreed with prior knowledge — would show separation by cell line accompanied by increased mixing of treatment and control cells in the shared latent space, but mixing across mutant cell lines with clear separation based on mutation status in the salient latent space,” said Lin, co-lead author of the paper with Weinberger. “And that is exactly what we observed. In addition, our model indicated a separation between wild-type cell lines in the salient space that suggested a differential response to treatment, which spurred us to run additional analyses to identify the specific genes that contribute to those variations.”

Such findings, which could build upon prior knowledge and lead scientists to new hypotheses, is precisely the sort of progress Lin and his colleagues hope their model will support. In another demonstration of ContrastiveVI’s potential, the researchers applied the model to a dataset drawn from intestinal epithelial cells of mice displaying variations in gene expression due to infection with the bacteria Salmonella or the parasite H. polygyrus (H. poly), a type of roundworm, using healthy cells as the control. Once again, the model aligned with expectations by separating along cell type and mixing across infections in the shared latent space, while largely mixing across cell types and separating by pathogen in the salient latent space.

Like the cancer cell example, the pathogen infection experiment also yielded unexpected patterns that prompted the team to analyze further. These patterns included differences in the upregulation of multiple genes between H. poly–infected tuft cells and other infected cell types that may have been masked in prior experiments — and could point to a distinctive role in the body’s immune response.

The researchers also explored how the model could be adapted to isolate variations in multimodal single-cell datasets, such as a combination of RNA and surface protein expression data in CRISPR-perturbed cells. They layered their CA modeling techniques onto TotalVI, a deep generative model developed to analyze joint RNA-protein datasets, to create TotalContrastiveVI. In a series of experiments, they showed how their extended model could be used to identify clusters of cells in the salient latent space and apply downstream analysis to identify patterns that warranted further investigation.

TotalContrastiveVI may be a proof of concept, but the underlying model is no mere demonstration project. The team designed ContrastiveVI to make it easy for researchers to integrate the tool into existing workflows.

“Our software is essentially plug and play,” noted Lin. “Computational biologists can deploy ContrastiveVI right now in conjunction with standard tools in the field such as Scanpy to begin exploring single-cell datasets in greater detail than they could before.”

Those details could yield new hypotheses that could, in turn, lead to new biomedical breakthroughs.

“There are so many contexts in which scientists would want to do this,” said Weinberger. “People were already excited by the potential of single-cell datasets. With ContrastiveVI, they can unlock even more insights and expand our knowledge of the mechanisms and treatment of disease.

“To borrow a popular metaphor in biomedical circles: before, we had a smoothie; now we can zoom in on each part of the corresponding fruit salad.”

Read the paper in Nature Methods here. Read more →

Take a generous helping of mathematical brilliance, cover it in copious amounts of curiosity about the most vexing problems underpinning computer science, add a generous dash of humility, and what do you get? 

Shayan Oveis Gharan, a professor in the Allen School’s Theory of Computation group, combines all the essential ingredients of a trailblazing researcher who, as his colleagues will attest, also happens to be a genuinely nice guy. The combination has also proved to be a genuine recipe for success, as he has racked up a series of accolades in theoretical computer science since he arrived at the University of Washington in 2015. His most recent honor, the Stephen Smale Prize from the Society for the Foundations of Computational Mathematics (FoCM), celebrated Oveis Gharan’s “breakthrough results on the applications of algebraic and spectral methods to the design of algorithms and to combinatorial optimization” that have made him “the architect of surprising and profound discoveries on foundational problems in computing.”

Oveis Gharan began acquiring the building blocks of those career triumphs as a middle school student in Iran, when he encountered the book “Mathematics of Choice: How to Count without Counting” by the Canadian-American mathematician Ivan Niven. Under Niven’s written tutelage, the young Oveis Gharan learned how to quickly count combinatorial objects — say, the number of ways one could assemble a basketball team consisting of 10 players and a captain from a class of 30 students — on paper. The lesson was a slam dunk, and counting problems still comprise a major portion of his research years later — only now, he’s designing computer algorithms to handle more sophisticated problems than the possible permutations with himself at point guard. 

Later, as he prepared to do battle in regional and world informatics competitions, a teenaged Oveis Gharan practiced with hundreds of combinatorial and graph theoretic problems drawn from past Russian Mathematics Olympiads. That practice paid off, as he took home a silver medal from the 2003 Central European Olympiad; a year later, he won gold at the International Olympiad.

“Those experiences are the backbone of most problem-solving techniques I still use today,” Oveis Gharan noted, pointing out that old approaches can come in handy when it comes to new problems. It’s one of the aspects he loves most about his chosen line of work.

“One of the amazing characteristics of research in discrete mathematics and combinatorics is that there is rarely a unified theme to approach hard problems,” he explained. “One often needs to cook up a new method from scratch. So in one sense, it feels like we are going to fight with a challenging problem empty-handed, but in reality, previous work on related problems can offer a menu of ideas with which to approach this other problem.”

One challenge that Oveis Gharan found he could really sink his teeth into is the infamous Traveling Salesperson Problem, which he first encountered over a decade ago as a Ph.D. student at Stanford University. There, he and his colleagues set out to design an improved approximation algorithm for metric TSP. 

At first, the team thought they had the solution — until they realized they didn’t.

“About halfway through we figured some of our intermediate conjectures were wrong,” Oveis Gharan recalled. “So, we made the problem simpler and instead only managed to prove that the algorithm breaks the long-standing barrier for a special family of metrics called ‘graph metrics.’ It wasn’t until two years ago that I and another group of co-authors finally achieved a result for all metrics.”

The aforementioned result was the first performance improvement in metric TSP in more than 40 years. Along the way, Oveis Gharan contributed to what co-author and Allen School professor Anna Karlin has described as a “deep mathematical machinery” mixing elements of graph theory and probability theory that researchers have since applied to other open problems. Among the tools in this expanded mathematical toolbox were the use of maximum entropy sampling, new theorems related to the geometry of polynomials, Strong Rayleigh probability distributions and negative dependence, and new insights into the combinatorial structure of the minimum and near-minimum cuts of a graph. Oveis Gharan and his co-authors, including Ph.D. student Nathan Klein, subsequently used the latter to build on their initial result by showing the integrality gap of the subtour linear programming relaxation for TSP is below 3/2 last year.

What is it about TSP that he finds so alluring? As Oveis Gharan explains it, TSP is different from most computational problems encountered by theoreticians, like the graph coloring problem, that require one to satisfy a range of local constraints. With TSP, one has to satisfy both a set of local constraints and a global constraint — connectivity — simultaneously.

“Oddly enough, each of these two sets of constraints is easy to satisfy optimally on their own, but the challenge is to satisfy both,” he said. “The quest in studying TSP is that you want to construct solutions which are locally correct while globally connected.”

For Oveis Gharan, satisfying that dual challenge is where the rubber meets the road.

”Think of a driver going from Seattle to San Francisco. They need to keep an eye on the road to make sure they are ‘locally’ driving safely — i.e., not ramming into the next car and not driving out of bounds,” he said. “But they also need to keep the bigger picture of the route in mind, choosing the right highway at every junction. Now in this example, perhaps, the bigger picture is easy to keep in mind when there’s only a single highway, I-5, running all the way south. But imagine how difficult it would be with millions of possible roads to choose from, and no GPS! That is similar to the dilemma in designing an algorithm for TSP.”

Despite his devotion to that problem, Oveis Gharan is also driven to tackle other challenges. For example, he is widely known for his work analyzing the Markov Chain Monte Carlo (MCMC) technique for sampling from high dimensional distributions and as a method for studying large, complicated sets. As part of that work, Oveis Gharan and his collaborators — including former student Kuikui Liu (Ph.D., ‘23), who will join the faculty of MIT this fall — developed the theory of spectral independence, a revolutionary approach for approximate sampling of Markov chains that has implications for computational biology, machine learning, physics and more. 

Oveis Gharan and his colleagues leveraged that approach to produce the first efficient approximation algorithm for counting the bases of a matroid and simultaneously proved a 30-year-old conjecture concerning the minimum edge expansion of the bases exchange graph of any matroid. In another paper of the same series, he and his co-authors answered another open question that had gone unanswered for nearly three decades by proving that an algorithmic tool used in statistical physics called Glauber dynamics mixes in polynomial time to generate a random independent set for any graph up to the phase-transition threshold.

The aforementioned work relates to expander graphs — which, in Oveis Gharan’s view, are “one of the most extraordinary inventions in mathematics.” He is keen to further explore the theory of high-dimensional expanders, which have numerous practical applications in computing.

“On one hand, these graphs are as sparse as, say, a cycle; on the other hand, they preserve almost all properties of a complete network,” Oveis Gharan explained. “If one wants to build a sparse routing network that will be as reliable as possible against node or connection failures, the best design is to use an expander graph. High dimensional expanders are a generalization of expander graphs to hypergraphs, which have been a subject of intense study over the last decade leading to several breakthroughs, from improved analysis of MCMC algorithms, to the construction of new locally testable codes.”

To someone whose work is typically celebrated in optimization and algorithm design circles, the Smale Prize came as a pleasant surprise. 

“Although not the community in which I normally publish my research, I am truly honored and amazed that my work has been recognized by leaders in computational mathematics,” he said. “This certainly motivates me and my research group to pursue a deeper understanding of problems at the intersection of math, computer science and efficient computation.”

The Smale Prize — named for Stephen Smale, one of the founding members of FoCM — is awarded every three years. The organization formally honored Oveis Gharan, who joins rarefied company as only the fifth recipient since the prize’s inception, during its annual conference in Paris last month. Read more →

On Friday, June 9, more than 4,000 family and friends from near and far gathered on the University of Washington campus to celebrate the Allen School’s 2023 graduates. The celebration commenced with a casual open house and meet-and-greet with faculty and staff in the Paul G. Allen Center and Bill & Melinda Gates Center. It culminated in a formal event in the Hec Edmundson Pavilion at the Alaskan Airlines Arena, where graduates made the brief journey across the stage to mark the start of a new journey as Allen School alumni.

‘Remember tonight‘

In her remarks opening the evening’s program, Magdalena Balazinska, professor and director of the Allen School, observed that most of those seated before her in their caps and gowns started their Allen School education prior to the COVID-19 pandemic. Balazinska congratulated them for overcoming the challenges and isolation of the intervening years to emerge victorious. She also noted that, while this milestone is cause for celebration, still more challenges — as well as opportunities — await them. And they should fear neither.

“When opportunities arise, take them. If you hesitate because opportunities are often scary and they often look like a lot of work, remember tonight. Remember how loud your families, your friends, the faculty, the staff have cheered for you, how much they believe in you, and take the opportunity,” Balazinska advised them. “Like generations that preceded you, you will face personal challenges, and the world around you will face challenges. Remember: you have a very strong education. Use that education, your passion, your kindness, your cleverness to have an impact on the people and the world around you.”

Recognizing the impact of those who came before

As if to illustrate the point, the Allen School welcomed back two graduates who have applied their education, along with their passion, kindness and cleverness, in very different ways: 2023 Alumni Impact Award recipients Janet Davis (Ph.D., ‘06) and Paul Mikesell (B.S., ‘96).

Allen School professor and Bill & Melinda Gates Chair Emeritus Ed Lazowska noted that the award is not only intended to honor outstanding alumni for their contributions throughout their careers, but also to remind the new graduates that they are joining a long and distinguished line of former Allen School students who have built on their education to change the world. Davis’ contributions include building Whitman College’s computer science program from the ground up to reflect that institution’s liberal arts traditions. Mikesell helped build scalable data storage company Isilon Systems into a multi-billion dollar company before expanding into agricultural technology by founding Carbon Robotics, a Seattle-based startup whose LaserWeeder provides farmers with a cost-effective, environmentally friendly alternative to pesticides.

‘Take advantage of the doors that open’

This year’s graduation speaker, Barbara Liskov, traveled from Boston to share her wisdom and encouragement with the newly minted graduates. Liskov is an Institute Professor of MIT — the highest institutional honor bestowed upon faculty — and received the A.M. Turing Award from the Association of Computing Machinery for her contributions to the theory and practice of programming language and systems design. She is also, as Lazowska noted in welcoming her to the stage, ”a wonderful human being — an example of what we should all strive to be.”

Before she touched down in Seattle, Liskov was asked what message she hoped the newly minted graduates would take away from their big day. 

“One thing that strikes me when I look back on my career is the importance of the unexpected,” Liskov explained. “Doors close and doors open. It’s important in your career not to be discouraged when doors close and to take advantage of the doors that open. You may end up doing something quite different from where you started, and this is absolutely ok.”

Liskov delivered that message and more to those gathered in the arena, drawing from a career spanning six decades. As the first woman in computer science at MIT — at a time when there were only 10 women on the entire faculty numbering nearly 1,000 — Liskov was a trailblazer in more ways than one. Having reluctantly accepted a position with industry when she couldn’t land a faculty position following her own graduation with a Ph.D. from Stanford, she relayed how she turned that disappointment into opportunity by transitioning her research from artificial intelligence to systems. 

Noting that where she ended up “was not at all where I might have predicted when I got started,” Liskov suggested that many of the graduates seated before her are likely to experience the same. And while they should not be deterred by detours, they should remain true to themselves.

“There will be setbacks, and there will be opportunities,” she said. “When there are setbacks, you want to persevere. When there are opportunities, you want to decide whether it’s a good idea for you to accept them. And these decisions that you make, you need to make by thinking about what’s going to work for you.”

Liskov also hoped the graduates would think about what that work will mean for the world at large. Pointing to technologies that enabled remote learning during the pandemic, computer-assisted surgery, and other contributions, she noted that computer science has created “marvelous opportunities.” But it also has created problems like fake news, bias stemming from the naive use of machine learning, and potential misuse of recent developments in AI.

“As you do your job, think about what’s ethical to do. If you develop tools, think about tools that will be good for humanity,” Liskov advised, noting that the entire field has an obligation to contribute its technical knowledge to mitigating such problems.

“I had a wonderful career. I had a lot of fun,” she concluded, “and I hope all of you have the same in your careers.”

Celebrating scholarship and service

Outstanding Senior Awards

Liskov’s words would have resonated with the recipients of the Allen School’s Outstanding Senior Awards, which recognize superior scholarship, potential for leadership and the ability to both apply and create new knowledge. While it is a nearly impossible task to choose from among the outstanding graduates each year — all of whom would have demonstrated some combination of those qualities to be admitted in the first place — five were singled out for their extraordinary contributions.

Maggie Jiang distinguished herself as an insightful and creative researcher in the Allen School’s Security and Privacy Research Lab who wasn’t afraid to ask questions about technology and scientific methodology. Operating at a level associated with experienced Ph.D. students, Jiang contributed to the publication of a longitudinal study of public opinion around the use of contact tracing apps to slow the spread of COVID-19 and concerns over individual privacy. She will continue on at the Allen School as a student in the combined B.S./M.S. program.

Sarah Khan was honored for her contributions as a teaching assistant for CSE Startup, a course for first-year students focused on problem solving, communication and computational thinking skills. In that role, Khan contributed to the development of curriculum for the Allen School Scholars Program and STARS with an emphasis on interdisciplinarity and the diversity of student experiences. Khan, who double-majored in computer science and education, communities and organizations, will continue her studies in the Allen School’s B.S./M.S. program.

Lansong (Ryan) Li was recognized for his remarkable contributions to interdisciplinary research projects bridging natural language processing and social computing. As a member of the Social Futures Lab, Li worked with UW and external collaborators to develop a harm-reduction framework for assessing and triaging misinformation online. He also explored how to leverage state-of-the-art neural network models to assess misinformation believability. Li will pursue a master’s degree at Stanford University following his graduation from the Allen School.

Alex Mallen is known as an ambitious and talented researcher with a passion for AI safety. As a member of the H2Lab, he spearheaded a project that sought to identify when large language models’ outputs can become untrustworthy — revealing his skills at building diagnostic datasets and running experiments in the process. Mallen, who is an active member of the grassroots research collective EleutherAI, previously earned a prestigious Goldwater Scholarship on his way to earning his undergraduate degree in just three years at the Allen School. 

Katherine Murphy earned recognition as an outstanding leader as a teaching assistant for Software Design and Implementation for nine quarters. Having taken over the course software infrastructure when many experienced TAs were about to graduate, she worked with the rest of the teaching team to keep the course running smoothly and ensure continuity across multiple instructors and offerings. Although many of her contributions were behind the scenes, Murphy was responsible for the positive experiences many of her fellow graduates had in the course.

Best Senior Thesis Award

Each year, the school recognizes an undergraduate student for original research contributions through the Best Senior Thesis Award. The recipient of this award has completed an independent research project under the supervision of one or more faculty members culminating in a thesis presenting their results. The school received eight nominations this year, from which it selected one award winner and one honorable mention recipient.

Ximing Lu received the 2023 Best Senior Thesis Award for “The Art of Algorithm and Knowledge in the Era of Extreme-Scale Neural Models.” In her thesis, Lu demonstrated how to empower small to moderate sized neural language models to work competitively against industry-scale models. A prolific researcher working under the supervision of Yejin Choi, the Brett Helsel Career Development Professor in the Allen School and Senior Research Director at AI2, Lu has already published multiple papers in the preeminent venues for natural language processing. She is currently pursuing her Ph.D. at the Allen School.

Matt Deitke earned an honorable mention for his thesis, “Scaling Embodied Artificial Intelligence: Massive 3D Simulations to Real-World Distributional Robustness.” That work, which Deitke completed under the supervision of professor Ali Farhadi, explored ways to improve robots’ abilities to generalize to novel scenarios.

Undergraduate Service Awards

Since 2011, the Allen School has recognized graduating seniors who devoted their time and energy to building community and benefiting their fellow students through various events and activities throughout their time on campus. This year, the school recognized five outstanding graduates with Undergraduate Service Awards.

Camila Christensen (B.S., ‘22) has been “an amazing Allen School ambassador” — particularly to transfer students. Christensen served as a teaching assistant for the school’s transfer seminar, which supports newly arrived students to acclimate to the program, in addition to supporting various outreach efforts and serving as a frequent volunteer for Allen School events.

Described as “an incredible leader,” Hayoung Jung has been an engaged member of the student group Computing Community (COM^2) throughout his time on campus. He applied his leadership skills during the pandemic to analyzing and reporting on how COVID-19 was impacting his fellow students. Jung, who double-majored in computer science and political science, was recognized in the Husky 100 last year.

Samuel Levy has served as a developer for Impact++, a student group focused on the intersection of computing and social good, and as an Allen School peer adviser, assisting his fellow students with advice and resources. For the past year, Levy has served as a lead peer adviser and is known as “an exceptional leader, advocate, and student employee.”

Eman Mustefa co-founded GEN1, a student group focused on building community among Allen School students who are the first in their families to pursue a four-year degree. Passionate about supporting women of color in computing, Mustefa has also “generously stepped up, time and time again” to share her experiences with K-12 students as an Allen School Ambassador.

Earning a reputation as an “outreach and recruitment rockstar,” Lynn Nguyen volunteered for more high school visits and information sessions than any other member of the Allen School Ambassadors team. A community-minded leader who is exceptional at supporting students, Nguyen is also described as the glue that held various events and activities together.

A tip of the hat to great teaching

Bob Bandes Memorial Awards

Teaching assistants play a vital role in the Allen School’s educational mission, serving not only its own majors but also thousands more students across campus who enroll in computing courses. More than 750 students served as Allen school TA’s in 2022-2023, supporting student learning through office hours, tutoring and review sessions while assisting instructors with various course administration duties. As teaching professor Justin Hsia observed, “We could not do what we do without you.” During the annual graduation celebration, the school recognizes outstanding TA’s from the preceding academic year with the Bob Bandes Memorial Award for Excellence in Teaching. Out of over 700 total nominations spread out over 200 individual TA’s, the school selected three winners and three runners up who went above and beyond to provide a supportive student experience.

Winner Anthony Chung has supported six different courses over nine quarters as an undergraduate TA, including multiple courses in the Allen School’s Introduction to Computer Programming series, Data Structures and Algorithms, and Distributed Systems. Chung was lauded for his “clear intent to do right by all of his students” through his diligence in providing them with clear feedback and consistent grading. He was also proactive in identifying areas where students were struggling, offering solutions such as the creation of alternative visualizations to assigned problems and hosting one-on-one Zoom calls with students outside of office hours.

Fellow winner Wen Qiu has been a TA for 12 quarters, first as an undergraduate and then as a graduate student in the Allen School’s B.S./M.S. program, for Web Programming, Data Programming and Intermediate Data Programming. In addition, she served as the instructor for the latter course last summer. Known for her going to “tremendous lengths” to share her expertise not only with students in her courses but also her fellow TA’s, Qiu earned accolades for “seeing something needed to be done and jumping in and doing it.” One student nominator noted, “She would be an amazing professor!” Qiu also founded and served as president of the Association for Computing Education (ACE).

The final winner, Sylvia Wang, served as a TA or head TA for nine quarters across five different courses, including Intermediate Data Programming, Data Structures and Parallelism, Systems Programming and Database System Internals. Wang’s teaching style resonated with her students, who described her as kind, encouraging, supportive, patient and helpful. “She stays with each student until they understand the concepts they’re struggling with and does not let any student leave with any confusion.” She was also an excellent advocate for both students and other TA’s, including being proactive in identifying how an assignment that multiple students struggled with could be adjusted to improve the experience of everyone in the class.

Honorable mentions went to Melissa Lin, Sudheesh Singanamalla and Ben Zhang.

Lin, a student in the B.S./M.S. program, has served as a TA for eight quarters spanning the introductory series, Foundations of Computing I and II, and Data Structures and Algorithms and earning students’ appreciation for giving 110% to her students — even at 8:30 in the morning.

Singanamalla, a Ph.D. student, served as a TA for graduate-level courses in Computer Systems and Computer Security and Privacy, where his wisdom and compassion for students “shined brightly in all aspects of his work.”

Zhang, an undergraduate, TA’ed for multiple offerings of the Foundations of Computing series, earning plaudits for being thorough and proactive, making his sections “very welcoming and open” and responding to questions that seem ambiguous with examples generated from his own understanding.

Undergraduate Teaching Awards

Each year, student leaders in COM^2 — formerly the UW chapter of the Association for Computing Machinery — recognize selected faculty members for contributing to the Allen School’s educational mission and enriching the student experience through the Undergraduate Teaching Awards. This year, the group highlighted two professors for their role in “shaping our minds and inspiring our achievements.”

Andrea Coladangelo, a professor in the Allen School’s Cryptography and Theory of Computation groups, was honored for his first quarter of teaching. “He has worked tirelessly this quarter to build up the quantum computing course, expanding the frontiers of knowledge for his students,” COM^2 Chair Vidisha Gupta said.

“What distinguishes him, however, is his ability to discern the ‘real’ question behind the question.” 

Known not only for his technical expertise but also for his kindness and generosity with his time, Coladangelo earned students’ admiration for his willingness to provide one-on-one help and to modify his teaching to better accommodate his class.

Teaching professor Ryan Maas (M.S.’18) was recognized for his impact that reaches far beyond the classroom.

Maas is known for his “captivating lectures” that make challenging concepts seem easy. But what truly sets him apart, Gupta noted, is his extraordinary care and dedication to his students.

“No question is too repetitive or silly for him, as he treats each inquiry with patience and thoughtfulness,” she said. “Ryan’s commitment to his students’ success extends beyond teaching, as he provides guidance and support to help them excel academically and grow as individuals.”

Congratulations to all of our Allen School graduates! And remember — our doors will always be open to you! Read more →

“There is not one area of the school that she does not touch in some way.”

“She” is Jennifer Worrell, the Allen School’s director of finance and administration. And that observation was made by a colleague advancing her successful nomination for a 2023 Professional Staff Award from the University of Washington College of Engineering. Each year, the College of Engineering Awards honor faculty, research and teaching assistants, and staff like Worrell whose extraordinary contributions benefit the college community.

Now approaching two decades of service at the Allen School, Worrell started out as an office manager and moved into successively more complex roles — event coordinator, grants manager, lead grants manager — before stepping into her current position in 2017. In highlighting her achievements since taking over as the school’s first new DFA in more than 30 years, the College noted that Worrell’s “combination of warmth and organizational know-how contributes to a culture that benefits her team and the Allen School as a whole.” 

That combination makes Worrell so effective at her job, some in the school are convinced that she possesses special powers. 

“Jen is like the great and powerful Oz,” said Kellus Stone, operations analyst at the Allen School and author of the aforementioned letter. “She’s the woman behind the curtain who makes sure everything runs smoothly as folks go about their business without giving a second thought as to how it all works.”

“How it all works” has only become more complex in recent years owing to the roll-out of new systems for managing everything from payroll to print jobs that coincided with a period of rapid growth. That growth has led to the school doubling its degree production, doubling its physical space, and surpassing $75 million in yearly expenditures — with roughly half going toward research.

“Jen has been a key contributor to the Allen School’s success and why it is thriving and growing,” said Megan Russell, assistant director of human resources. “Any time there is a need for someone to fill a gap, Jen raises her hand and says, ‘I’ll do it.’ When an employee says they’re overwhelmed, she responds with, ‘What can I do to help?’ 

“She will never take any credit for it, but she deserves it,” Russell continued. “We are all better for her presence here.”

Despite her can-do attitude and willingness to fill any gap, Worrell would have been forgiven for questioning her presence here after enduring a trial by fire immediately upon ascending to her position. When she took the reins of the school’s Business Office, her first task was to implement and train her team on a new online payroll system, Workday, that was being rolled out across the University. If that wasn’t sufficiently daunting, her second task was to fill two open positions responsible for entering Allen School payroll into this same system after the incumbents left shortly after the big roll-out.

There were times, in those early days, that Worrell wasn’t sure how she would make it past lunch, let alone to the end of the day. But make it she did, repeatedly rising to the occasion while overseeing not only the Business Office, but also Research Administration and Human Resources. Two more teams, Facilities & Operations and Events, would be added to her portfolio later. Each time she was called upon for advice or assistance in response to a crisis, she would answer with a genuine smile on her face — and a genuine concern for the wellbeing of her colleagues.

“Jen leads the entire business administration team, and yet when I have meetings with her, she is fully present and engaged, offering helpful solutions and encouragement,” said Amber Cochran, assistant director of events for the Allen School. “She has the ability to make each staff member feel seen and valued.”

She also leads by example as the de facto head of the Staff Executive Committee, a group that comprises staff directors and assistant directors responsible for various functions that make up the administrative and operational side of the school. This group, which encompasses not only Worrell’s functional teams but also Undergraduate Student Services, Graduate Student Services, Technical Support, External Relations and Communications, engages in high-level organizational planning and develops unified policies and procedures along with consistent messaging across the entire school.

The role is challenging enough on a good day; it reached a whole new level when COVID-induced remote working scattered those directors, assistant directors and their teams across the region — and sometimes even farther afield. And yet, Worrell worked with her colleagues to quickly adapt, taking steps to ensure staff maintained a sense of connection and had the resources they needed remain both agile and resilient in the face of uncertainty.

“Jen is the linchpin — she is the central pillar of our school,” said Magdalena Balazinska, professor and director of the Allen School. “She has very deep expertise and can answer any question on almost any topic. We couldn’t have achieved our current growth without her help and leadership.”

Worrell is extending that help and leadership to assist with the latest overhaul of campus-wide systems known as the University of Washington Finance Transformation (UWFT). Her colleague Debbie Carnes, who serves with her on the Process Transformation Team, has witnessed firsthand how Worrell has employed her professional skills and personal empathy to assist UWFT program staff in understanding how changes to the administration of research grants and other fiscal processes are likely to impact operational staff college-wide.

”Jen’s longtime service to the College, ability to come up with innovative and creative solutions, resourcefulness and positivity are an asset to us all,” said Carnes, administrator for the UW Department of Chemical Engineering. “I cannot think of anyone more deserving of this award than Jen.”

Aside from her well-earned reputation as a skilled leader and a veritable fountain of knowledge about how the University works, perhaps the greatest endorsement Worrell has collected is that of colleagues who point to the time and care that she gives them — even when she is busy. Make that especially when she is busy.

“Not long ago, I ran into Jen in the hallway as she was rushing from one meeting to the next. Despite that, she stopped and asked me how I was doing,” recalled Stone. “I started to answer and then stopped myself and apologized, as I could see she was in a hurry. ‘It’s okay,’ she replied with a smile. ‘You are important, too.’

“I believe that encapsulates who Jen is at her core,” continued Stone. “She is one of a kind.”

Members of the Allen School will gather to celebrate Worrell’s recognition, and that of UW Distinguished Staff Award recipient Chloe Dolese Mandeville, at a reception on June 14. 

Learn more about the College of Engineering Awards here.

Read more →

Champion, advocate, role model…based on her colleagues’ descriptions, Chloe Dolese Mandeville sounds like a regular Girl Scout. Which, it so happens, she is: for the past two and a half years, the Allen School’s Assistant Director for Diversity & Access has volunteered as a troop leader for the Girl Scouts of Western Washington, hosting activities on campus and inspiring girls to see computing as a potential career path.

It is but one example of the many ways in which Dolese Mandeville has helped students to engage with the field — efforts that have now earned her a 2023 Distinguished Staff Award from the University of Washington. Part of the UW Awards of Excellence, the Distinguished Staff Award is the highest honor bestowed upon staff by the University. 

As the saying goes, not all heroes wear capes.

“Chloe’s responsibilities are enormous — hers is definitely not a job for a mere mortal,” said professor Dan Grossman, Vice Director of the Allen School. “But she built a strong team to help her get it done, and she is a phenomenal leader. People love working with her.”

After graduating from UW with a bachelor’s in psychology with a minor in education, Dolese Mandeville joined the Allen School’s undergraduate advising team in 2016 to assist students in charting their own educational journeys. She took a particular interest in transfer students and the unique challenges they face in acclimating to the UW, teaching a seminar designed to help ease the transition. She simultaneously worked on a transition of her own as she pursued a master’s degree in leadership in higher education. 

That degree would come in handy when, a mere two months after completing it, she took the reins of the school’s Diversity & Access program.

At the time, the Allen School’s undergraduate program had earned a national reputation for its success in recruiting and retaining women in computer science. But gender was the only area in which the school seemed to be making headway when it came to the breadth of students it serves. Not long before Dolese Mandeville assumed her present role, Jeff Dean (Ph.D., ‘96), Google Senior Fellow and Chief Scientist, and his wife Heidi Hopper approached school leaders with a challenge to extend the same energy and fervor they had devoted to growing the school’s gender diversity to other underrepresented groups. 

Dolese Mandeville embraced that challenge — and ran with it. Among her first priorities was morphing the school’s K-12 outreach programs from “broad and shallow” to “narrow and deep” by building substantive, sustainable partnerships with a set of schools and community organizations that directly served student populations the school was trying — and to that point, largely failing — to reach. With this new approach, the school soon surpassed the Seattle campus-wide average in the proportion of students who are Black or African American, from economically disadvantaged backgrounds, or among the first in their families to pursue a four-year degree. Previously, the share of the school’s students who identified with these groups was half, or less, than that of the campus as a whole.

“We are serving an increasingly diverse undergraduate population that is more reflective of the face of Washington and of technology users around the world,” said professor Ed Lazowska, Bill & Melinda Gates Chair Emeritus in Computer Science & Engineering in the Allen School. “Chloe has been instrumental in this remarkable transformation. We wouldn’t have made this progress without her.”

The notion of transformation comes up repeatedly in conversations about Dolese Mandeville’s impact. It is among the many superlatives offered by members of the Allen School’s undergraduate student services team who work alongside her every day.

“Chloe’s compassion, skill, talent and hard work have truly had a transformational effect on the Allen School student experience,” said Crystal Eney, director of undergraduate student services. “Chloe’s tenacity and creativity are among her greatest strengths, and the Diversity & Access team has risen leaps and bounds from where it started under her leadership.”

Another word that is mentioned in connection with Dolese Mandeville is “fierce” — but her peers are quick to point out that such fierceness is accompanied by compassion and kindness. And, they note, her leadership is paying dividends not only for UW but also for the broader field of computing.

“Chloe’s impact on the Allen School and computing is vast and unparalleled. Her leadership in building equitable, justice-oriented programs and systems while centering the student experience is one of Chloe’s greatest strengths.” observed Leslie Ikeda, who manages the Allen School Scholars Program, “If anyone can transform the work we are doing to support our field’s most vulnerable populations, it’s Chloe.”

The program Ikeda manages, formerly known as Allen School Startup, was initially conceived as an immersive, four-week summer experience to assist incoming first-year students who are first-generation, low-income and/or from underserved communities in their transition to college. It has since evolved under Dolese Mandeville’s direction into a comprehensive, year-long cohort-based program with wraparound support. The summer bridge course remains, but that is now accompanied by increased staff support, one-to-one mentorship, workshops that supplement students’ first-year coursework, a new study hall course and community-building events throughout the year.

“Our mission is to educate the next generation of outstanding computer scientists and computer engineers who reflect the diverse needs, backgrounds and experiences of people in society at large,” said Juliet Quebatay, senior program manager for K-12 outreach programs. “Chloe supports us all with time, energy, constructive feedback and a clear vision of where we want to go — all while creating realistic, sustainable collaborations and programming that will help the school get there.”

“Us all” is the close-knit team of full-time professional staff that Dolese Mandeville has assembled to execute on that vision. In addition to Quebatay and Ikeda, the team includes Christina Huynh, academic adviser for the Allen Scholars; Kayla Shuster Sasaki, who focuses on high school and transfer student recruiting, and EJ Pinera, who works directly with Allen School student groups such as Ability, Women in Computing (WiC), GEN1, Minorities in Tech (MiT) and Q++ — to name only a few. Like much of the school’s current DEIA-focused initiatives, those groups got their start with Dolese Mandeville’s encouragement.

“Chloe championed the importance of student groups in building community and a sense of belonging for all students in the Allen School,” Pinera said. 

Dolese Mandeville also championed a mentorship initiative alongside undergraduate students called Changemakers in Computing. CiC is a summer program for rising juniors and seniors in Washington state high schools interested in exploring technology and its intersection with society and justice. Through a combination of culturally relevant project-based learning and networking opportunities, the program empowers students from marginalized backgrounds to engage with computing as a potential career while building a community of future computer scientists and engineers who will be changemakers in the field. Importantly, CiC is completely free to participants; meals, public transportation to campus and all activities are covered by the program, as is an education stipend, to ensure that a lack of financial resources is no barrier to student participation. 

The program has grown from serving roughly 20 high school students when it was launched in 2021 to 40 students in the most recent cohort. Encouraging students to lead the way, as she did with CiC — and backing them up with the tools and resources that will help them to succeed — is characteristic of Dolese Mandeville’s approach.

“Chloe prioritizes student voices,” said Chelsea Navarro, senior academic adviser. “She takes actions big and small to ensure that students of all backgrounds feel that they belong and can thrive here.”

Those actions include teaming up with Pinera, Assistant Director of Advising Jenifer Hiigli and Senior Academic Adviser Rakeb Million to push for the creation of physical spaces in the school’s buildings that reflect its values around DEIA. Spaces such as the Diversity & Access student lounge and a dedicated prayer/meditation room offer places where students can support each other, share experiences and honor their whole selves.

In addition to taking concrete steps that contribute to a more welcoming and inclusive culture, Dolese Mandeville is also committed to setting the school up for success over the long haul.

“Chloe is amazing and an incredible asset to the Allen School. Our entire community — students, staff and faculty — benefit from her presence,” said professor Tadayoshi Kohno, the Allen School’s associate director for diversity, equity, inclusion and access. “In summer 2020 Chloe and I started working on a 5-year strategic plan to guide our DEIA work, and her vision, leadership and wisdom have been instrumental in getting us to where we are today.”

“Where we are today” is a testament to how effective Dolese Mandeville has been in helping the Allen School rise to the challenge issued by Hopper and Dean since she stepped into her role.

“I see firsthand, every day, the amount of energy, compassion and thought Chloe puts into building out our DEIA efforts,” said Hiigli. “The Allen School would absolutely not be the same if she had not been here building these programs over the past several years. 

“Chloe’s work has benefited thousands of computer science students in countless ways.”

Two of Dolese Mandeville’s Allen School colleagues were also among the nominees for 2023 Distinguished Staff Awards: Senior Academic Adviser Chelsea Navarro, in the individual impact category, and Robotics Lab Manager Selest Nashef, in the individual collaboration category.

Dolese Mandeville and her fellow honorees will be formally recognized at a campus ceremony on June 8.

Learn more about the UW Awards of Excellence here. Read more →

University of Washington professor Thomas Rothvoss, a member of the Allen School’s Theory of Computation group with a joint appointment in the UW Department of Mathematics, has received the 2023 Gödel Prize recognizing outstanding papers in theoretical computer science for “The matching polytope has exponential extension complexity.” In the paper, Rothvoss proved that linear programming — a core technique in combinatorial optimization for modeling a large class of problems that are polynomial-time solvable  — cannot be used to solve the perfect matching problem in polynomial time. He originally presented these results at the 46th Association for Computing Machinery Symposium on Theory of Computing (STOC 2014).

Rothvoss shares this year’s accolade with researchers Samuel Fiorini and Serge Massar of the Université Libre de Bruxelles, Hans Raj Tiwary of Charles University in Prague, Sebastian Pokutta of the Zuse Institute Berlin and Technische Universität Berlin, and Ronald de Wolf of the Centrum Wiskunde & Informatica and the University of Amsterdam. Two years before Rothvoss published his seminal result, that team proved the extension complexity of the polytope for the Traveling Salesperson Problem is exponential — confirming that there is no polynomial-sized extended formulation, and therefore no small linear program, that can be used to solve the TSP. 

That result provided a partial, yet definitive, answer to a problem posed by theoretician Mihalis Yannakakis two decades prior. For Rothvoss and his colleagues in the tight-knit theory community, it was a watershed moment.

“Most of the time in complexity theory, we deal in conjectures but can’t actually prove any of them,” Rothvoss said. “On a good day, we can maybe prove that one conjecture implies another. So it was rather surprising when Sam and the rest of that group proved, completely unconditionally, the exponential extension complexity of the TSP polytope.”

The group further proved that its result extended to the maximum cut and stable-set polytopes, as well. But that proof, significant as it was, only answered the question for problems that are NP-hard. Inspired by the progress Fiorini and his collaborators had made, Rothvoss aimed to settle Yannakakis’ question once and for all when it comes to linear programs applied to polytopes that are not NP-hard — that is, well-understood polytopes, such as that of the perfect matching problem, for which polynomial-time algorithms for optimizing linear functions are known to exist. 

And settle it, he did.

“I focused on it full-time for half a year,” Rothvoss recalled. “A couple of the technical aspects of that 2012 paper were also useful for my purposes, such as a technique drawn from Razborov’s classic paper on communication complexity, while others I had to modify.

“In particular, we knew the so-called rectangle covering lower bound used by Fiorini et al. to great effect in the case of TSP would not suffice for the matching polytope,” he continued. “In fact, the rectangle cover number for matchings is polynomial in the number of vertices, so it turned out that a more general technique — hyperplane separation lower bound — works instead.”

In the process of arriving at his proof, Rothvoss confirmed that Edmonds’ characterization of the matching polytope, made nearly half a century earlier, is essentially optimal. According to Allen School professor James R. Lee, his colleague’s work was — and remains — a significant insight with ramifications in mathematics, algorithm design and operations research.

“Thomas’ work is a masterful combination of ideas from two seemingly disparate areas of TCS,” said Lee. “It’s the synthesis of really profound insights of Yannakakis and Razborov from three decades ago, weaving together polyhedral combinatorics and communication complexity to settle a problem that essentially predates the era of P vs. NP.”

Rothvoss previously received the Delbert Ray Fulkerson Prize from the Mathematical Optimization Society and the American Mathematical Society for the same work. He is also the past recipient of a Packard Fellowship, a Sloan Research Fellowship, a National Science Foundation CAREER Award and Best Paper Awards at STOC, the Symposium on Discrete Algorithms (SODA) organized by the ACM and the Society for Industrial and Applied Mathematics (SIAM), and the Conference on Integer Programming and Combinatorial Optimization (IPCO). 

The Gödel Prize, named for mathematical logician Kurt Gödel, is co-sponsored by the ACM Special Interest Group on Algorithms and Computation Theory (SIGACT) and the European Association for Theoretical Computer Science (EATCS). Rothvoss and his fellow honorees will be formally recognized at STOC 2023 in Orlando, Florida next month. Learn more about the Gödel Prize here. Read more →

Since 2020, communities around the globe have endured more than 1,100 natural disasters combined. From floods and drought to earthquakes and wildfires, these events contribute to human suffering and economic upheaval on a massive scale. So, too, do pandemics; since the emergence of SARS-CoV-2 at the end of 2019, nearly 7 million people have died from COVID-19. 

Then there is the human, economic and geopolitical toll caused by cyberattacks. While there is no way to know for certain, one oft-cited study estimated hackers launch “brute force” attacks against a computer once every 39 seconds, the equivalent of roughly 800,000 attacks per year. The fallout from malicious actors gaining unauthorized access to these and other systems — ranging from an individual’s laptop to a country’s electrical grid — is projected to cost as much as $10.5 trillion worldwide by 2025.

Whether natural or human-made, events requiring rapid, coordinated responses of varying complexity and scale could be could be addressed more efficiently and effectively with the help of artificial intelligence. That’s the thinking behind two new National Artificial Intelligence Research Institutes involving University of Washington researchers, including Allen School professors Simon Shaolei Du and Sewoong Oh, and funded by the National Science Foundation.

AI Institute for Societal Decision Making

Allen School professor Simon Shaolei Du will contribute to the new AI Institute for Societal Decision Making (AI-SDM) led by Carnegie Mellon University. The institute will receive a total of $20 million over five years to develop a framework for applying artificial intelligence to improve decision making in public health or disaster management situations, when the level of uncertainty is high and every second counts, drawing on the expertise of researchers in computer science, social sciences and humanities along with industry leaders and educators.

“AI can be a powerful tool for alleviating the human burden of complex decision making while optimizing the use of available resources,” said Du. “But we currently lack a holistic approach for applying AI to modeling and managing such rapidly evolving situations.”

To tackle the problem, AI-SDM researchers will make progress on three key priorities to augment — not replace — human decision making, underpinned by fundamental advances in causal inference and counterfactual reasoning. These include developing computational representations of human decision-making processes, devising robust strategies for aggregating collective decision making, and building multi-objective and multi-agent tools for autonomous decision-making support. Du will focus on that third thrust, building on prior, foundational work in reinforcement learning (RL) with long-time collaborators Aarti Singh, professor at CMU who will serve as director of the new institute, and Allen School affiliate professor Sham Kakade, a faculty member at Harvard University, along with CMU professors Jeff Schneider and Hoda Heidari.

Adapting RL to dynamic environments like that of public health or disaster management poses a significant challenge. At present, RL tends to be most successful when applied in data-rich settings involving single-agent decision making and using a standard reward-maximization approach. But when it comes to earthquakes, wildfires or novel pathogens, the response is anything but straightforward; the response may span multiple agencies and jurisdictions, the sources of data will not have been standardized, and each incident response will unfold in an unpredictable, situation-dependent manner. Compounding the problem, multi-agent decision making algorithms have typically performed best in scenarios where both planning and execution are centralized — an impossibility in the evolving and fragmented response to a public health threat or natural or human-made disaster, where the number of actors may be unknown and communications may be unreliable. 

Du and his colleagues will develop data-efficient multi-agent RL algorithms capable of integrating techniques from various sources while satisfying multiple objectives informed by collective social values. They will also explore methods for leveraging common information while reducing sample complexity to support effective multi-agent coordination under uncertainty.

But the algorithms will only work if humans are willing to use them. To that end, Du and his collaborators will design graduate-level curriculum in human-AI cooperation and work through programs such as the Allen School’s Changemakers in Computing program to engage students from diverse backgrounds — just a couple of examples of how AI-SDM partners plan to cultivate both an educated workforce and an informed public.

“There is the technical challenge, of course, but there is also an educational and social science component. We can’t develop these tools in a vacuum,” Du noted. “Our framework has to incorporate the needs and perspectives of diverse stakeholders — from elected officials and agency heads, to first responders, to the general public. And ultimately, our success will depend on expanding people’s understanding and acceptance of these tools.”

In addition to CMU and the UW, partners on the AI-SDM include Harvard University, Boston Children’s Hospital, Howard University, Penn State University, Texas A&M University, the University of Washington, the MITRE Corporation, Navajo Technical University and Winchester Thurston School. Read the CMU announcement here.

AI Institute for Agent-based Cyber Threat Intelligence and Operation

Allen School professor Sewoong Oh and UW lead Radha Poovendran, a professor in the Department of Electrical & Computer Engineering, will contribute to the new AI Institute for Agent-based Cyber Threat Intelligence and OperatioN (ACTION). Spearheaded by the University of California, Santa Barbara, the ACTION Institute will receive $20 million over five years to develop a comprehensive AI stack to reason about and respond to ransomware, zero-day exploits and other categories of cyberattacks. 

”Attackers and their tactics are constantly evolving, so our defenses have to evolve along with them,” Oh said. “By taking a more holistic approach that integrates AI into the entire cyberdefense life cycle, we can give human security experts an edge by rapidly responding to emerging threats and make systems more resilient over time.”

The complexity of those threats, which can compromise systems while simultaneously evading measures designed to detect intrusion, calls for a new paradigm built around the concept of stacked security. To get ahead of malicious mischief-makers, the ACTION Institute will advance foundational research in learning and reasoning with domain knowledge, human-agent interaction, multi-agent collaboration, and strategic gaming and tactical planning. This comprehensive AI stack will be the foundation for developing new intelligent security agents that would work in tandem with human experts on threat assessment, detection, attribution, and response and recovery.

Oh will work alongside Poovendran on the development of intelligent agents for threat detection that are capable of identifying complex, multi-step attacks and contextualizing and triaging alerts to human experts for follow-up. Such attacks are particularly challenging to identify because they require agents to sense and reason about correlating events that span multiple domains, time scales and abstraction levels — scenarios for which high-quality training data may be scarce. Errors or omissions in the data can lead agents to generate a lot of false positives, or conversely, miss legitimate attacks altogether. 

Recent research using deep neural networks to detect simple backdoor attacks offers clues for how to mitigate these shortcomings. When a model is trained on data that includes maliciously corrupted examples, small changes in the input can lead to erroneous predictions. Training representations of the model on corrupted data is an effective technique for identifying such examples, as the latter leave traces of their presence in the form of spectral signatures. Those traces are often small enough to escape detection, but state-of-the-art statistical tools from robust estimation can be used to boost their signal. Oh will apply this same method to time series over a network of agents to enable the detection of outliers that point to potential attacks in more complex security scenarios.

Oh and Poovendran’s collaborators include professors João Hespanha, Christopher Kruegel and Giovanni Vigna at UCSB, Elisa Bertino, Berkay Celik and Ninghui Li at Purdue University, Nick Feamster at the University of Chicago, Dawn Song at the University of California, Berkeley and Gang Wang at the University of Illinois at Urbana-Champaign. The group’s work will complement Poovendran’s research into novel game theoretic approaches for modeling adversarial behavior and training intelligent agents in decision making and dynamic planning in uncertain environments — environments where the rules of engagement, and the intentions and capabilities of the players, are constantly in flux. It’s an example of one of the core ideas behind the ACTION Institute’s approach: equipping AI agents to be “lifelong learners” capable of continuously improving their domain knowledge, and with it, their ability to adapt in the face of novel attacks. The team is keen to also develop a framework that will ensure humans continue to learn right along with them.

“One of the ways this and other AI Institutes have a lasting impact is through the education and mentorship that go hand in hand with our research,” said Oh, who is also a member of the previously announced National AI Institute for Foundations in Machine Learning (IMFL). “We’re committed not just to advancing new AI security tools, but also to training a new generation of talent who will take those tools to the next level.”

In addition to UCSB and the UW, partners on the ACTION Institute include Georgia Tech, University of California, Berkeley, Norfolk State University, Purdue University, Rutgers University, University of Chicago, University of Illinois Chicago, University of Illinois Urbana-Champaign and University of Virginia. Read the UCSB announcement here and a related UW ECE story here.

The ACTION Institute and AI-SDM are among seven new AI Institutes announced earlier this month with a combined $140 million from the NSF, its federal agency partners and industry partner IBM. Read the NSF announcement here.

Read more →

There was a time when cookies were considered something to be savored — back when chips referred to chocolate rather than silicon. Once “cookies” became synonymous with online tracking, privacy researchers weren’t so sweet on the concept. 

That includes Allen School professors Franziska Roesner and Tadayoshi Kohno, who investigated the online tracking ecosystem for their 2012 paper “Detecting and Defending Against Third-Party Tracking on the Web.” Last month, Roesner, Kohno and co-author David Wetherall, a former Allen School professor who is now a Distinguished Engineer at Google, received the Test of Time Award at the 20th USENIX Symposium on Networked Systems Design and Implementation (NSDI 2023) for their influential work, which offered the first comprehensive evaluation of third-party trackers and their intrusion into people’s activities online. 

The team’s findings informed the nascent policy debate around web privacy that has become all the more relevant with the proliferation of social media and reliance on targeted advertising as a revenue model. They also led to the creation of new tools like Privacy Badger, a browser extension that learns and automatically blocks hidden third-party trackers used by millions of people to protect themselves and their browsing histories online. The work also inspired a significant body of follow-on research, including team members’ subsequent paper that appeared at NSDI 2016 chronicling the increase in both the prevalence of online tracking and the complexity of tracker behavior over time.

“Considering how much time we spend online and the variety of activities we engage in, this type of tracking can yield a lot of information about a person,” said Roesner, a co-director of the Security and Privacy Research Lab at the University of Washington along with Kohno. “That’s even truer today than it was a decade ago, and I’m gratified that our work helped initiate such an important conversation and informed efforts to educate and empower users.”

At the time of the original paper’s release, third-party tracking had started to gain attention in security and privacy circles. But researchers were just nibbling around the edges, for the most part; they had a fragmented understanding of how such trackers worked and their impact on people’s online experience. Roesner — an Allen School Ph.D. student at the time — worked with Kohno and Wetherall to develop a client-side method for detecting and classifying trackers according to how they interact with the browser. They analyzed tracker prevalence and behavior on the top 500 website domains, as identified by the now-defunct web traffic analysis firm Alexa Internet, examining more than 2,000 unique pages.

“We identified 524 unique trackers, some of which had sufficient penetration across popular websites to enable them to capture a significant fraction of a user’s browsing activity — typically around 20%, and in one case, as much as 66%,” Roesner recalled.

Roesner and her colleagues cataloged five types of tracker behavior, varying from the relatively benign, to the opportunistic, to the infuriating. The behaviors spanned analytics that are generally confined to a specific site, Google Analytics being an example; “vanilla” trackers, which rely on third-party storage to track users across sites for the purposes of additional analytics or targeted advertising, such as Doubleclick; forced, which include the dreaded popup or redirect that compels the user to visit its domain; referred, which rely on unique identifiers leaked by other trackers; and personal trackers, which engage in cross-site tracking based on a user’s voluntary visit to its domain in other contexts. Some trackers exhibit a combination of the above.

Despite the existence of multiple tools intended to give users more control, from third-party cookie blockers to “private” browsing mode, the team found those options insufficient for preventing certain trackers from following people across the web while maintaining any semblance of functionality. This was particularly true for popular social widgets by the likes of Facebook, Twitter, LinkedIn, Digg, and others that were embedded on a growing number of sites ranging from news outlets to online storefronts.

“While users could prevent some tracking, that was not the case for social widgets,” noted Roesner. “If a user was logged into a social media site like Facebook, for instance, their activity elsewhere on the web would be tracked — non-anonymously, I would add — even if they didn’t interact with the ‘like’ button embedded on those sites.”

For those who would prefer to cover their tracks while continuing to enjoy the convenience of interacting with social widgets on their terms, Roesner and her collaborators developed ShareMeNot. The browser extension took a bite out of social widgets’ ability to construct browsing profiles of users by only allowing activation of third-party tracking cookies when a user explicitly interacted with the “like,” “share,” or other relevant buttons; if a user visited a site but did not click on the social widgets, ShareMeNot stripped the cookies from any third-party requests to those trackers.

The team worked with an undergraduate research assistant in the lab, Chris Rovillos (B.S., ‘14) to refine ShareMeNot following the paper’s initial publication and address instances of the trackers attempting to circumvent the restrictions on cookies via other means. Instead of just blocking cookies, the new and improved version of the tool blocked tracker buttons altogether. In their place, ShareMeNot inserted local, stand-in versions of the buttons that users could click to either “like” a page directly or load the real button — putting users, not the trackers, in control. Roesner partnered with the nonprofit Electronic Frontier Foundation to incorporate ShareMeNot into the previously mentioned Privacy Badger, which remains an important tool for protecting users from intrusion by third-party trackers to this day.

The team’s work is notable for inspiring not only new technologies but also a new wave of researchers to focus on web tracking. One of those researchers, Umar Iqbal, followed that inspiration all the way to the Allen School.

“This is one of the seminal works in the space of web privacy and security. It had an immense influence on the community, including my own research,” observed Iqbar, a postdoc in the Security and Privacy Research Lab. “I extended several of the techniques proposed in the paper as part of my own doctoral thesis, from the measurement of online trackers, to their characterization, to building defenses. It was, in fact, one of the reasons I decided to pursue a postdoc with Franzi at UW!”

Roesner, Kohno and Wetherall were formally recognized at NSDI 2023 last month in Boston, Massachusetts. Read the research paper here.

Read more →