Allen School News

Allen School professor Sidd Srinivasa talks about his quest for a unified mathematical model for human-robot interaction

Each year, the Allen School welcomes industry partners, alumni and friends to our Affiliates Research Day to learn about the latest work by our faculty and students and to explore the future of computing-related research. Yesterday, more than 200 people participated in our 2017 Research Day, which featured technical talks on core and emerging areas of the field, a luncheon keynote by professor Sidd Srinivasa on human-robot interaction, and our evening open house and poster session.

During the open house, participants were invited to embark on self-guided tours of the Paul G. Allen Center, where more than 50 research teams presented posters and demos of their latest work. The evening afforded student researchers an opportunity to explain their projects to local industry leaders and technologists and, for some, to earn bragging rights (and some terrific prizes) as part of the Madrona Prize and People’s Choice Award competitions.

Madrona Venture Group sponsors the Madrona Prize — now in its 12th year — to recognize student researchers whose work represents exciting new directions and potential for commercialization.

“The newly named Paul G. Allen School has a long history of research that improves lives and delivers business and research growth to the region,” said Madrona Managing Director Tim Porter.

The winners were announced at the conclusion of the open house, along with the winners of the People’s Choice Award for visitors’ favorite posters of the evening. Read on for a complete run-down of the people and projects who were recognized as part of this year’s event.

2017 Madrona Prize

Winner

Madrona’s Tim Porter (left) congratulates Madrona Prize winners Manuel Nordhoff and Eunice Jun as Allen School Director Hank Levy looks on

The Madrona team selected two related projects in human computer interaction research developed under the guidance of Allen School professor Katharina Reinecke as joint grand-prize winners in this year’s competition:

LabintheWild: Large-scale Online Experimentation with Diverse Uncompensated Samples (postdoc Nigini Abilio Oliveira; Ph.D. student Eunice Jun; professor Katharina Reinecke) and Augury: Predicting visual appeal of website design (visiting scientist Manuel Nordhoff and professor Katharina Reinecke).

Runners up

A Unified Approach to Interpreting Model Predictions (Ph.D. student Scott Lundberg and professor Su-In Lee)

Real-time VR Video Processing with the Hardware-friendly Bilateral Solver (Ph.D. student Amrita Mazumdar; postdoc Armin Alaghi; professors Luis Ceze and Mark Oskin)

Madrona Prize winners and runners up with the Madrona Venture Group team

TVM: End-to-end IR Stack for Deep Learning Systems (Ph.D. students Tianqi Chen, Thierry Moreau, and Haichen Shen; Fudan University undergraduate and AWS intern Ziheng Jiang; professors Luis Ceze, Carlos Guestrin, and Arvind Krishnamurthy)

2017 People’s Choice Award

Winner

Learning Stylized Character Expressions from Humans (Ph.D. students Deepali Aneja, Bindita Chaudhuri; Zillow research scientist Alex Coburn; artist Gary Faigin; professors Linda Shapiro and Barbara Mones)

Runners up

Winners of the People’s Choice Award onstage with Allen School professor Ed Lazowska (left) and External Relations Director Kay Beck-Benton (right)

IDCam: Precise Item Identification for AR-Enhanced Object Interactions (Ph.D. students Hanchuan Li, Eric Whitmire, and Alex Mariakakis; Qualcomm engineer Victor Chan; former postdoc and Disney Research scientist Alanson Sample; professor Shwetak Patel)

RoyalFlush: Non-invasive toilet water overflow detector (Electrical Engineering Ph.D. students Farshid Salemi Parizi and Josh Fromm; professor Shwetak Patel)

View the complete list of featured posters and demos here, and Madrona’s blog post here. Also check out GeekWire’s coverage of Srinivasa’s talk here, and the open house here.

Many thanks to our friends at Madrona to all of the members of our extended Allen School community who joined us in supporting student innovation!

  Read more →

If, as Oscar Wilde once said, “Life imitates art far more than art imitates life,” then the art of film has a lot to answer for when it comes to the perpetuation of gender stereotypes. Thanks to researchers in the Allen School’s Natural Language Processing research group, we now have a way to measure the sometimes subtle biases in how men and women are portrayed on the big screen — and increase our understanding of how language shapes our perception of gender roles.

A team that includes Ph.D. students Ari Holtzman, Hannah Rashkin and Maarten Sap, bachelor’s alumna Marcella Cindy Prasetio, and professor Yejin Choi analyzed nearly 800 movie scripts across multiple genres and found that male characters tend to be vested with higher levels of authority and control over their own destinies than their female counterparts. The researchers applied connotation frames — a method for understanding the connotations associated with different verbs — to assess characters’ level of power and agency through their actions and speech. Even after controlling for the disparity in the number of roles, quantity of dialogue, and screen time assigned to male versus female characters, the team found that males consistently scored higher across all genres.

“What we found was that men systematically have more power and agency in the film script universe,” Holtzman said in a UW News release.

In addition to analyzing scripts, the team applied its framework to the plot summaries of several popular Disney princess movies. What they found was that, when it came to power and agency, several of Disney’s most beloved characters are not exactly living a fairy tale. This includes Anna, one of the main protagonists in the popular 2013 film, “Frozen.” While her sister, Elsa, was portrayed as being in command of her own destiny, Anna possessed much less control and often had to rely on the assistance of a man — showing that when it comes to gender stereotypes, not much has changed in the Disney universe in over half a century.

“Anna is actually portrayed with the same low levels of power and agency as Cinderella, which is a movie that came out more than 60 years ago,” said Sap, lead author of the paper that describes the team’s results. “That’s a pretty sad finding.”

The research team (clockwise from top left): Maarten Sap, Hannah Rashkin, Ari Holtzman, Yejin Choi, and Marcella Cindy Prasetio

The project built upon previous work by Choi, Rashkin, and former Allen School postdoc Sameer Singh that defined connotation frames for analyzing how a writer’s choice of verb implies certain attitudes and relationships between subjects — revealing how seemingly objective statements by an author can influence readers’ judgment about people and events. Originally conceived as a potential tool for analyzing subtle biases in online media, the framework was extended to reveal how the verbs used by and in relation to movie characters suggest power and influence along gender lines.

The researchers’ approach goes beyond a well-known method for measuring gender bias in known as the Bechdel Test. That test has gained traction over the last decade as a proxy for identifying films that offer a more robust portrayal of women that is about more than their relationships with men. Whereas the Bechdel Test stipulates that a movie must have at least two characters who are women and who speak to each other about topics other than a man, the Allen School’s framework provides a more nuanced analysis of the differences in how men and women are portrayed on the big screen.

Interestingly, the team discovered the discrepancy is not necessarily down to male writers perpetuating a deeply ingrained gender imbalance — it carries over into films scripted by female writers and overseen by female casting directors, too.

“Even when women play a significant role in shaping a film, implicit gender biases are still there in the script,” Rashkin noted.

While the team focused on characters playing on the big screen, the same method could help people recognize the subtle biases conveyed in books, plays, and more.

“We believe it will help to have this diagnostic tool that can tell writers how much power they are implicitly giving to women versus men,” Choi said, noting that these subtle biases are “deeply integrated in our language.” Eventually, the team hopes to broaden its tool to offer potential solutions, such as suggestions for ways to rephrase passages of text.

Choi and her colleagues created an online database that enables researchers and members of the public to explore their findings for hundreds of popular films. Try out the interactive tool here, and read the UW News release here. Listen to Rashkin and Sap explain their work on an episode of KUOW’s The Record here.

The team presented its findings in September at the Conference on Empirical Methods in Natural Language Processing (EMNLP 2017) held in Copenhagen, Denmark.

  Read more →

The New York Times writes:

“What recent studies have made increasingly apparent is that the greatest number of high-paying STEM jobs are in the ‘T’ (specifically, computing). …

“‘There is a huge divide between the computing technology roles and the traditional sciences,’ said Andrew Chamberlain, Glassdoor’s chief economist.

“At LinkedIn, researchers identified the skills most in demand. The top 10 last year were all computer skills, including expertise in cloud computing, data mining and statistical analysis, and writing smartphone applications.”

The Allen School’s Ed Lazowska and Berkeley’s David Culler are among those quoted.

Read more here. Read more →

Researchers in the Allen School’s Networks & Mobile Systems Lab have introduced a new kind of smart fabric imbued with computing and interaction capabilities — without the need for onboard electronics. Their work could redefine what we mean by “wearable” and usher in a fashionable new direction for computing.

Smart garments currently on the market typically pair conductive thread with electronic components, batteries, and sensors — elements that cannot be submerged under water or subjected to extreme temperatures. Allen School Ph.D. student Justin Chan and professor Shyam Gollakota discovered that, by harnessing the magnetic properties of the same, off-the-shelf thread, they could dispense with electronics altogether and overcome one of the principal barriers to widespread adoption of wearable technology.

“This is a completely electronic-free design, which means you can iron the smart fabric or put it in the washer and dryer,” said Gollakota in a UW News release.

To produce their smart textiles, the researchers used a conventional sewing machine to embroider the conductive thread onto fabric. They then manipulated the fabric, using a magnet to align the poles in a positive or negative direction to correspond with ones and zeros. The data encoded in the fabric can be read by a magnetometer — an inexpensive device that is built into most smartphones.

Chan and Gollakota envision several potential form factors and uses for the technology, including accessories such as neckties, wristbands, and belts that can do double-duty as data storage and authentication tools.

“You can think of the fabric as a hard disk — you’re doing this data storage on the clothes you’re wearing,” Gollakota explained.

Shyam Gollakota (left) and Justin Chan

To illustrate how their smart fabric could offer an alternative to expensive RFID-based authentication systems, they sewed a magnetic patch containing an identifying image onto the sleeve of a shirt, which is then passed in front of a prototype magnetic fabric reader containing an array of magnetometers and a microprocessor. The reader determines whether the signals emitted from the sleeve match a predetermined pattern; if they do, the door is unlocked.

The team’s approach can also be used to enable gesture recognition and interaction. To demonstrate, the researchers sewed magnetized thread into the fingertips of a glove and built a gesture classifier into a smartphone. They then tested six commonly used gestures, each of which emits its own unique combination of magnetic signals, and found that the phone could interpret the signals corresponding to each gesture in real time with 90% accuracy.

“With this system, we can easily interact with smart devices without having to constantly take it out of our pockets,” said Chan, lead author of the research paper describing the team’s work.

While the magnetic signal will degrade over time — think magnetized hotel keycards that are frequently encoded and wiped as guests come and go — like those same keycards, the fabric can be re-magnetized and re-programmed over and over. Another thing the fabric and key cards have in common is their susceptibility to demagnetization in the presence of a strong external magnetic field. This is because fabric constructed with commercially available thread has a weak magnetic field, making it best suited for temporary data storage. The researchers believe that custom fabrics incorporating a stronger magnetic field could offer greater resilience for longer-term applications.

Chan and Gollakota presented their work at the Association for Computing Machinery’s User Interface Software and Technology Symposium (UIST 2017) in Quebec City, Canada last week.

Read the UW News release here, visit the project page here, and check out coverage by MIT Technology Review, GeekWire, International Business Times, Engadget, New Atlas, Quartz, The Next Web, KOMO News and KING 5 News.

Read more →

Photo: Devin Coldewey / TechCrunch

Today the Paul G. Allen School was honored to host a roundtable on “Women in 21st Century Jobs” with House Democratic Leader Nancy Pelosi and Congresswoman Suzan DelBene from Washington’s 1st Congressional District.

Undergraduate students, graduate students, faculty, and staff explored a wide variety of issues with Pelosi and DelBene including:

• the barriers that women face in the tech industry
• policies to encourage more women in STEM
• public-private partnerships to help close the skills gap
• sexual harassment in society at large
• programs that can increase the success of economically disadvantaged and underrepresented minority students in STEM
• communicating the value of fundamental research
• approaches to mitigating the spread of misinformation/disinformation online
• the importance of immigration reform

Our deep thanks to Leader Pelosi and Representative DelBene for spending the morning with us in a truly engaging conversation!

(See coverage in TechCrunch, GeekWire, KING TV,  KOMO TV, KIRO TV). Read more →

Anat Caspi, director of the Allen School’s Taskar Center for Accessible Technology, is being honored with the 2017 Innovation Award from Northwest Access Fund, a non-profit organization devoted to improving access and opportunity for people with disabilities throughout the Pacific Northwest. The Innovation Award recognizes a business, organization, or individual who has developed one or more products that improve quality of life for people with disabilities.

“Northwest Access Fund was honored to have Anat Caspi present on the work of the Taskar Center at our January 2017 Board of Directors meeting,” said Emerson Sekins, executive director of Northwest Access Fund. “One highlight of her presentation was on AccessMap Seattle, which will be a powerful tool for our clients with mobility barriers — especially those who use manual wheelchairs.”

AccessMap, which grew out of a student-driven project that won the City of Seattle’s 2015 Hack the Commute competition, is a route-finding app that makes navigating Seattle streets safer and easier for everyone with the assistance of customizable route maps that take into account the local terrain, the presence or absence of curb ramps, and other potential obstacles.

“The Taskar Center is making it possible for people with disabilities to save time and energy, and in doing so allowing for greater participation in all our communities,” Sekins said.

Allen School professor emeritus Richard Ladner, a renowned accessibility researcher who has watched the Taskar Center grow under Caspi’s leadership, nominated her for the award based on her contributions to AccessMap and other innovative projects geared to improving the lives of people with disabilities.

“The mission of the center is to create new technologies and translate that technology to actual practice,” Ladner explained. In addition to AccessMap, Ladner lauded Caspi’s work on the Universal Play Kiosk. The kiosk, which was featured at the 2015 Seattle Design Festival focused on the theme “Design for Equity,” is a modular, configurable activity space that engages children of all abilities, including those living with mobility impairments, in sensory-rich games and activities.

“Anat works closely with children and their parents with disabilities,” Ladner noted. “She is an incredible advocate for including these children in the mainstream by building on the children’s strengths using technology. She is very deserving of the Northwest Access Fund Innovation Award.”

Caspi and her fellow honorees will be formally recognized at Northwest Access Fund’s annual awards dinner and benefit on November 2 in Seattle. The event celebrates organizations and individuals who endeavor to include, serve, and help people with disabilities in Washington and Oregon while building support for Northwest Access Fund’s programs that enable people to achieve greater independence with the help of assistive technologies.

Congratulations, Anat!

  Read more →

The U.S. Bureau of Labor Statistics has just released its employment projections for the decade 2016-2026. It’s a highly detailed forecast: more than 1,000 specific job categories are included.

Computing occupations once again dominate STEM, accounting for 66% of all job growth, and 60% of all job openings (whether due to growth or to replacement).

BLS projects a growth of 546,000 computing jobs over the decade, and 3,475,000 job openings.

Whoaboy! Read more →

“Luke Zettlemoyer, a professor at the University of Washington … turned down a lucrative offer from Google, instead taking a post at the nonprofit Allen Institute for Artificial Intelligence so he could continue teaching.”

Luke and Ali Farhadi are heavily engaged in AI2, which is led by long-time Allen School professor Oren Etzioni. It offers the best of both worlds.

NY Times article here. Read more →

Professor Jennifer Mankoff, a member of the Allen School’s human computer interaction research group, has been honored with a GVU Impact Award from the GVU Center at her alma mater, Georgia Tech. To mark its 25th anniversary, the center recognized Mankoff and 13 other current or former members who have had a significant impact on the world and contributed substantially to GVU’s reputation, influence, and community in pursuit of its mission to improve the human condition through technology.

Mankoff earned an Impact Award for her contributions to accessibility, health, and sustainability through research that combines empirical methods and deep technical innovation. She is widely known for her people-centric approach to technology, such as her novel use of 3D printing to create personalized assistive technologies for people with disabilities — work which, like all accessibility research, will ultimately benefit everyone. Mankoff also has explored the use of natural materials in computing, including embedding textiles in 3D printing and creating knitted objects programmatically, and developed tools and techniques to assist people in managing chronic illness.

“Each of the individuals featured…embodies the interdisciplinary mindset and commitment to real-world impact that is a hallmark of GVU’s identity,” Keith Edwards, director of the GVU Center, said in an announcement. “Through their leadership, service, and research excellence, they have changed the way we use computing technology, advanced the frontiers of knowledge, and strengthened the GVU community at Georgia Tech.”

Mankoff earned her Ph.D. in 2001 working with Gregory Abowd and fellow honoree Scott Hudson. She joined the Allen School faculty as the Richard E. Ladner Professor this fall after spending 13 years on the faculty of Carnegie Mellon University, where she was a professor in the HCI Institute. Before her arrival at CMU, Mankoff was a faculty member at the University of California, Berkeley.

The GVU Center formally recognized Mankoff and her fellow Impact Award winners at its 25th anniversary celebration earlier today.

Congratulations, Jen!

  Read more →

This map, representing an individual’s morning commute, shows the locations where the research team was able to track the person’s movements through location-based ads.

Online ads may not only be trying to sell you something; they may be selling you out. That’s according to a team of researchers in the Allen School’s Security and Privacy Research Lab, who recently discovered how easy it is for someone with less than honorable intentions to turn online ads into a surveillance tool. They found that, for as little as $1,000, a person or organization could conceivably purchase ads that will enable them to track someone’s location and app use via their mobile phone — gaining access to potentially sensitive personal information about that individual’s dating preferences, health, religious and political affiliation, and more. The team hopes that by sharing its findings publicly, it will raise awareness among online advertisers, mobile service providers, and customers about a potential new cybersecurity threat.

This threat stems from how the existing online advertising ecosystem enables ad purchasers to precisely target consumers based on their geographic location, interests, and browsing history for marketing purposes. The problem, as researchers explained in a UW News release, is that the same infrastructure can be exploited by people and organizations other than advertisers to precisely target individuals in ways that could compromise their privacy and security. According to former Allen School Ph.D. student Paul Vines, lead author on the project, it would be easy for anyone from a foreign agent to a jealous spouse to sign up with an online advertising service and track another individual.

“If you want to make the point that advertising networks should be more concerned with privacy, the boogeyman you usually pull out is that big corporations know so much about you. But people don’t really care about that,” Vines explained in a Wired article about the project. “[T]he potential person using this information isn’t some large corporation motivated by profits and constrained by potential lawsuits. It can be a person with relatively small amounts of money and very different motives.”

As the team discovered, online advertising can deliver fairly detailed information about a person’s behavior. For example, the researchers were able to determine an individual user’s location within a distance of 8 meters based on where their ads were being served. By establishing a grid of hyperlocal ads, the team was able to discern an individual’s daily routine based on where ads were served to the user’s device at various points along the way.

The team refers to this method of information gathering as ADINT, or “advertising intelligence,” reminiscent of well-known intelligence collection tactics such as SIGINT (signals intelligence) and HUMINT (human intelligence). To test the capabilities of ADINT, Vines and his coauthors — Allen School professors Franziska Roesner and Tadayoshi Kohno — purchased a series of ads through a demand-side provider, or DSP, which is an entity that facilitates the purchase and delivery of targeted advertising. They set up their ads to target a mix of 10 actual users and 10 facsimile users with the help of each device’s unique mobile advertising identifier (MAID), which functions as a sort of “whole device” tracking cookie. The team then repurposed the tools designed to deliver relevant ads for commercial purposes to instead collect information on each user’s whereabouts and behavior.

The ADINT research team, from left: Tadayoshi Kohno, Franziska Roesner, and Paul Vines Dennis Wise/University of Washington

Movement was not the only thing they could track; it turns out that ad purchasers have the ability to learn a lot about a person by viewing what apps they use, including popular dating and fitness-tracking apps. The team’s experiments also revealed that the individual being tracked does not need to actually click on an ad in order for ADINT to work, because purchasers can see where the ad is being served regardless of whether the target interacts with it.

“To be very honest, I was shocked at how effective this was,” said Kohno, who co-directs the Allen School’s Security and Privacy Research Lab with Roesner. “There’s a fundamental tension that as advertisers become more capable of targeting and tracking people to deliver better ads, there’s also the opportunity for adversaries to begin exploiting that additional precision.”

The team surmises that ADINT attacks could be driven by a variety of motives, from criminal intent, to political ideology, to financial profit. According to Roesner, the ease with which the team was able to deploy targeted ads against individuals calls for heightened awareness and vigilance — not just within the computer security community, but on the part of the policy and regulatory communities, as well.

“We are sharing our discoveries so that advertising networks can try to detect and mitigate these types of attacks,” she explained, “and so that there can be a broad public discussion about how we as a society might try to prevent them.”

The team will present its findings at the Association for Computing Machinery’s Workshop on Privacy in the Electronic Society taking place in Dallas, Texas later this month.

To learn more about ADINT, visit the project website here. Read the UW News release here and the Wired feature here, and check out additional coverage by The Verge, Mashable, and Mic.

  Read more →