Allen School News

The Puget Sound Business Journal posted an article this week in which Microsoft President Brad Smith, a vocal proponent of expanding access to computer science education, reiterated his call for a significant expansion of UW CSE as the leading generator of computing talent in the state.

From the article:

“As Microsoft expands outside the country to mitigate the lack of engineering talent in the Puget Sound region, local technology industry leaders renew a call for investments in the next generation of local computer science graduates.

“‘It’s important for Puget Sound to continue to strengthen our own ability to develop more talent,’…Smith recently told the Puget Sound Business Journal. ‘We need to get more computer science into schools here in Washington state, including the University of Washington, which is a world leader. It’s so important to build a second UW computer science building.'”

The article points out that local technology companies are “talent-starved” due to a lack of qualified graduates and quotes UW CSE professor Ed Lazowska on the need to expand capacity to meet the explosive demand from employers and students.

“This will consolidate us as one of the preeminent programs in the country,” Lazowska said. “The tragedy is, most applicants don’t get in. These are terrific kids from Washington state who could graduate here, stay here and power our economy.”

Read the full article here. Learn more about the Campaign for UW CSE here. Read more →

UW CSE Ph.D. student Kira Goldner has been named a 2016 Google Anita Borg Memorial Scholar. Goldner, who works with professor Anna Karlin in UW CSE’s Theory group on algorithmic game theory and approximation algorithms, is one of only 20 student researchers from across the United States selected by Google to receive this award.

Google established the Anita Borg Memorial Scholarship in honor of Borg’s tireless dedication to advancing women and other underrepresented groups in the field of computing. The program celebrates recipients’ academic excellence and their potential to become active role models and leaders in the field. The scholarship includes an invitation to participate in the Google Scholars’ Retreat at the company’s Mountain View, California headquarters.

The scholarship includes an invitation to participate in the Google Scholars’ Retreat at the company’s Mountain View, California headquarters. Goldner is the latest in a distinguished line of UW CSE students who have received the award, including Nell O’Rourke and Irene Zhang (2015), Jenny Abrahamson and Nicki Dell (2012), Janara Christensen and Katie Kuksenok (2011), Lydia Chilton and Kristi Morton (2010), Saleema Amershi (2009), and Julie Letchner and Kate Everitt (2008).

Way to go, Kira! Read more →

Just how accurate are facial recognition algorithms—which may have been trained and tested on fewer than 15,000 photos—when put to the test on a larger scale? Researchers in UW CSE’s Graphics and Imaging Lab (GRAIL) aimed to find out by launching the MegaFace Challenge, a new competition in which teams from all over the world were invited to put their algorithms through their paces using the MegaFace dataset of one million images.

The results showed that, when it comes to the size of datasets used for training and testing these algorithms, bigger tends to be better.

From the UW News release:

“‘We need to test facial recognition on a planetary scale to enable practical applications—testing on a larger scale lets you discover the flaws and successes of recognition algorithms,’ said Ira Kemelmacher-Shlizerman, a UW assistant professor of computer science and the project’s principal investigator. ‘We can’t just test it on a very small scale and say it works perfectly.’

“The UW team first developed a dataset with one million Flickr images from around the world that are publicly available under a Creative Commons license, representing 690,572 unique individuals. Then they challenged facial recognition teams to download the database and see how their algorithms performed when they had to distinguish between a million possible matches.

“Google’s FaceNet showed the strongest performance on one test, dropping from near-perfect accuracy when confronted with a smaller number of images to 75 percent on the million person test. A team from Russia’s N-Tech.Lab came out on top on another test set, dropping to 73 percent.”

Testing against the larger dataset revealed differences in performance across facial recognition algorithms that were masked when tested against a much smaller dataset, with accuracy rates of some other algorithms that had previously performed well on a small scale—some surpassing 95%—falling to as low as 33% when using the larger dataset. Algorithms that were trained on larger datasets to begin with tended to outperform those that were trained on smaller datasets when tested on the MegaFace collection.

The team, which also includes UW CSE professor Steve Seitz, Master’s student Aaron Nech, undergraduate student Evan Brossard, and former student Daniel Miller, will present its results at the IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2016) in Las Vegas next week.

In addition to the challenge—which is ongoing—researchers are building a large-scale training dataset that will incorporate multiple photos of half a million identities. This new dataset will benefit researchers who previously had to rely on smaller sample sizes to train their algorithms. The project could help enable new applications for facial recognition technology, such as face-based security features for mobile devices and the ability for enforcement to quickly and accurately identify individuals captured on video surveillance footage for public safety purposes.

Read the full news release here and the research paper here. Visit the MegaFace website here, and check out articles in TechCrunch, The Atlantic, IEEE Spectrum, and Silicon Republic. Read more →

Professor Anup Rao of UW CSE’s Theory group has been recognized by the Society for Industrial and Applied Mathematics (SIAM) with its Outstanding Paper Prize for the 2013 paper, How to Compress Interactive Communication. Each year, SIAM selects just three papers from among thousands published in its journals over the previous three years for the award, which honors authors who have made original contributions to the field of applied mathematics.

In 1948, Claude E. Shannon published his foundational paper on information theory. He defined the entropy of a message, and showed that this quantity captures the number of bits required for one party to communicate the message to another, thereby laying the mathematical groundwork for the field of data compression.

Shannon’s result addresses the setting of one-way communication tasks. But what if two parties engage in a conversation, possibly exchanging many short messages in order to accomplish a task interactively. Can such a protocol be compressed? If so, what is the appropriate notion of entropy?

The groundbreaking paper by Rao and co-authors Boaz Barak, Mark Braverman, and Xi Chen, addresses precisely these questions. They present sophisticated methods to compress interactive communication protocols. Intuitively, their idea is for the communicating parties to predict the future communication transcript based on their current partial knowledge. Moreover, they put forth a precise and appealing definition of the “information cost” of a protocol which measures the amount of information exchanged by the two parties.

The team’s work helped to launch the field of information complexity, and led to progress in many areas of computer science, including streaming computation, differential privacy, communication complexity, and quantum communication.

Learn more about the award here, and read the winning paper here.

Congratulations, Anup! Read more →

Arvind Krishnamurthy, Will Scott, and Tom Anderson

Newly minted UW CSE Ph.D. Will Scott has captured the Best Student Paper Award at the 2016 USENIX Annual Technical Conference underway in Denver for the paper Satellite: Joint Analysis of CDNs and Network-Level Interference.

The winning paper presents Satellite, an efficient tool for understanding global trends in the distribution and accessibility of website content from a single vantage point. The system, which collects and analyzes data on DNS resolution and resource availability by monitoring the IPv4 address space, enables researchers to measure the growth and behavior of content distribution networks (CDNs) and prevalence of online censorship among the top 10,000 domains as ranked by Alexa.

Scott co-authored the paper with his Ph.D. advisers, Tom Anderson and Arvind Krishnamurthy, and CSE professor Yoshi Kohno. Read the full paper here, and check out the Satellite website here.

This is the second year in a row that UW CSE research has been recognized at the conference. Last summer, the GRAPPA team collected the Best Paper Award for its work on software distributed shared memory for data-intensive applications.

Go team! Read more →

Josh Smith

An international team led by researchers at the UW has been chosen as one of three finalists in the GlaxoSmithKlein Bioelectronics Innovation Challenge—and made itself eligible for $1 million in new research funding as a result.

UW CSE and electrical engineering professor Josh Smith is working with colleagues at UW’s Center for Sensorimotor Neural Engineering (CSNE), the University of Cambridge and University College London to develop an implantable device that would restore bladder function for people living with spinal cord injury or incontinence. Technology developed in Smith’s Sensor Systems Lab is being used to wirelessly power the implanted device.

From the UW News release:

“An international team led by researchers at the Center for Sensorimotor Neural Engineering (CSNE) based at the University of Washington is one of three finalists in a race to produce an implantable wireless device that can assess, stimulate and block the activity of nerves that control organs….

“‘For people with spinal cord injuries, restoring sexual function and bladder function are some of their top priorities — higher than regaining the ability to walk,’ said Chet Moritz, deputy director of the CSNE and UW associate professor of rehabilitation medicine and of physiology and biophysics.

“‘The vision is for these neural devices to be as ubiquitous as pacemakers or deep brain stimulators, where a surgeon implants the device and it’s seamless for the patient,’ he said. ‘We’re really excited to make a difference in people’s lives and to help push these technologies forward.’

Prototype implanted device designed at the UW.

“The final implantable wireless device will be able to stimulate and block electrical signals that travel along the nerves and control specific organs. Stimulating the pelvic nerve causes the bladder to empty, for example, while blocking those signals could help someone who is unable to control his or her bladder.”

The team designed its device to interact with the pelvic nerve both electronically and optically. This approach, in which light is used to control neurons, means the researchers can stimulate the pelvic nerve without having to physically touch it, thus reducing the risk of swelling and scarring that can occur with direct nerve interfaces.

“It is gratifying to see the center’s hardware research efforts paying off so quickly,” said Smith. “Selection by GlaxoSmithKline in this rigorous international competition shows that technologies emerging from the CSNE are at the leading edge of what is possible.”

In addition to the $1 million prize for reaching the finals, GSK will award $1 million to the first team to deliver a functional device in small animal models. The other UW members of the GSK competition team include physiology and biolophysics professor Greg Horwitz, physiology and biophysics postdoc Tom Richner, biology professor and eScience Institute Data Science Fellow Bingni Brunton, and Ryan Solinsky, a resident in rehabilitation medicine. The CSNE is a NSF-funded partnership of multiple research institutions that is housed at the UW and directed by UW CSE professor Rajesh Rao.

Read the full UW News release here. Learn more about the GSK Bioelectronics Innovation Challenge here.

Amazing work—congratulations to Josh and the entire team! Read more →

UW CSE alum Martha Kim, now a member of the computer science faculty at Columbia University, has been honored with a 2016 Borg Early Career Award from the Committee on the Status of Women in Computing Research (CRA-W). This award—named in honor of the late Anita Borg, an early member of CRA-W and a leading advocate for increasing women’s participation in computing—recognizes one or two women each year in the early stages of their careers who have made significant contributions to the computing community through research and through efforts to advance women and diversity in the field.

Kim completed her Ph.D. in 2008 working with UW CSE professor Mark Oskin and currently directs Columbia’s Architecture and Design (ARCADE) Lab, where her research focuses on developing ways to increase power and efficiency and reduce energy consumption while meeting the growing demands on computing systems of big data sets. She is particularly interested in developing hardware and software techniques to improve usability of hardware accelerators and data-centric accelerator design. In each of the past three years, papers she has co-authored have been selected among IEEE’s “Top Picks in Computer Architecture” for their novelty and long-term impact. Kim has won numerous awards, including a NSF CAREER Award in 2013, in recognition of her research and teaching.

In announcing the award, CRA-W cited Kim’s outreach to high school students, including participation in conferences such as “Code Like A Girl” and service as a faculty advisor for the Artemis Project, which is a summer school for girls in grades 9 and 10 from underserved schools. Kim also served on a working group charged with revamping her own high school’s approach to STEM education. Kim’s extensive outreach efforts are bearing fruit: her first two Ph.D. graduates are women, as are two-thirds of her current Ph.D. students.

Read the CRA-W award citation here, and a nice post on the Columbia CS department blog here.

Way to go, Martha! Read more →

Ras Bodik accepts the Influential Paper Award at ISCA.

Professor Ras Bodik, a member of UW CSE’s Programming Languages & Software Engineering (PLSE) group, collected the 2016 ACM SIGARCH and IEEE-CS TCCA ISCA Influential Paper Award (a “test of time” award) at the International Symposium on Computing Architecture being held this week in Seoul, South Korea. Bodik and his co-authors, former University of Wisconsin-Madison Ph.D. students Brian Fields and Shai Rubin, received the award for their 2001 ISCA paper, Focusing Processor Policies via Critical-Path Prediction, in which they introduced a simple yet effective hardware predictor of instruction criticality to optimize performance.

The predictor employs a dependence-graph model of the microarchitectural critical path that incorporates both data and machine-specific dependences to identify execution bottlenecks, and uses a token-passing algorithm to compute the critical path without having to actually build the dependence graph. By focusing processor policies on critical instructions, the predictor enables the prioritization of critical instructions for scarce resources, and suppressed speculation on non-critical instructions to reduce the risk of misspeculations. 

Ras Bodik, Luis Ceze and Karin Strauss at ISCA 2016.

Bodik and his colleagues demonstrated their predictor was capable of supporting fine-grain optimizations and producing significant improvements in performance—as high as 21%, and 10% on average, in the case of cluster instruction scheduling and steering.

Read the winning paper here.

Congratulations, Ras!

Ras is at ISCA with CSE professor Luis Ceze and affiliate faculty member Karin Strauss of Microsoft Research. In addition to Ras’ big win, they are cheering on their colleague from the Molecular Information Systems Lab, CSE affiliate faculty member Doug Carmean of Microsoft Research, who delivered a keynote talk titled Quantum and Cryo and DNA oh my! Sights Along the New Yellow Brick Road. Read more →

Alex Mariakakis demonstrates Bilicam to Sen. Cantwell. (No actual babies were harmed during this demonstration!)

U.S. Senator Maria Cantwell spent nearly two hours at UW CSE on Saturday afternoon, along with staffers Dayna Lurie and Nate Caminos. The agenda:

• Ed Lazowska: student demand, workforce demand, diversity, and data science.
• Yoshi Kohno: cybersecurity.
• Alex Mariakakis: smartphone medical diagnostic apps.
• Steve Seitz: virtual reality.

Steve Seitz demonstrates content created using Cardboard Camera, software developed by his team at Google.

Many thanks to Senator Cantwell for taking time from her busy schedule to learn more about what we do. Washington State is blessed with an extraordinary Congressional delegation. Read more →

Original study by Change the Equation, the Amgen Foundation, and C+R Research

What courses do high school students like “a lot”?

AP Computer Science blows away all other STEM fields! Among 52 fields, its popularity is exceeded only by Dance, Drawing, Graphic Design, Painting, and Choir.

The study, by Change the Equation, the Amgen Foundation, and C+R Research, was unearthed and analyzed by Code.org’s Hadi Partovi.

Hadi Partovi’s aggregated version

Hadi corrected the study for selection bias, and aggregated related fields – the results were just as dramatic!

Read Hadi’s blog post here. Read more →