The 2015 Microsoft Research Faculty Summit is over, but I am still recovering my voice from all the great hallway conversations! The summit reminded my fellow Microsoft researchers, our myriad collaborators in academia, and me of what we have already accomplished and the exciting opportunities ahead.
Harold Javid (opens in new tab) blogged about day one of the summit in detail, but I would like to call out a few of my own takeaways from that day. First, I was extremely pleased by the reactions to Jeannette Wing (opens in new tab)’s announcements of the RFPs for Microsoft HoloLens and Catapult. I overheard my academic colleagues discussing ideas for taking advantage of these grant opportunities. I enjoyed the spirited discussions that characterized Eric Horvitz’s (opens in new tab) panel on artificial intelligence—there’s a palpable excitement about the entire AI field that Eric’s panel captured. You can watch on demand (opens in new tab) Jeannette’s keynote and Eric’s session.
The Universal Design (opens in new tab) breakout session offered some fresh motivation and approaches for diversity, which Harold covered in his blog (opens in new tab). Although diversity is my passion, Professor Charles Isbell (opens in new tab) of Georgia Tech delivered an “aha moment” for me. More than 60 percent of the faculty at the top 4, 10, 20, and 25 computer science programs are graduates of one of these same programs. The ranking of one’s PhD institution is a huge factor in hiring—departments hire at their own rank or higher. This is common knowledge, but Charles connected it to diversity. If the very top programs would make a truly concerted effort to increase the participation of women and minorities in PhD programs, the effect would propagate throughout the entire computer science field. Only a few people, those who lead and serve on the PhD admissions committees, can make it happen.
Spotlight: On-demand video
AI Explainer: Foundation models and the next era of AI
Explore how the transformer architecture, larger models and more data, and in-context learning have helped advance AI from perception to creation.
Finally, I really enjoyed socializing that evening with colleagues from around the world. It was great to see so many old friends—and to make new ones. I was especially excited that so many assistant professors attended for the first time and were enjoying the research breadth of the content and making new connections. My abiding memory of the reception will be the way Harry Shum (opens in new tab) used his “rock star” status to spark animated conversations throughout the night.
On day two, I had the honor of introducing one of the most influential researchers in my field, Stanford professor Monica Lam (opens in new tab). Monica delivered a provocative keynote on Omlet (opens in new tab), an app that she and her students at Stanford created as an open alternative to Facebook and China’s WeChat, which she indicts as “big-brother” social networks that trample privacy and exploit users’ personal data to their own profitable advantage. I could try to summarize her arguments, but I wouldn’t do them justice, so I encourage you to listen to her presentation (opens in new tab) for yourself. Everyone was hungry for more technical details.
Monica Lam delivered her keynote, A Revolution Against Big-Brother Social Networks.
After Monica’s address, it was time for the Research Showcase (opens in new tab). This year, we expanded the showcase to include 47 demo and poster booths with a range of exciting projects underway at Microsoft, many of them joint efforts with academic institutions. There was truly something to interest researchers from every domain of computer science. The demos were clustered around six broad themes—Artificial Intelligence, Software, Devices, Computing in Society, Research in Action and Engaging With Microsoft—and featured crowd-pleasers like the RoomAlive Toolkit (opens in new tab) and cutting-edge topics like What Can We Solve with a Quantum Computer?
My own favorite was the demo of an AI project that can accurately guess your weight, waist size and body mass index (BMI) based on a handful of inputs. (The estimate was over in my case!) The TouchDevelop (opens in new tab) demo showed some of the software capabilities of the tiny Microbit programmable device that will soon be in the hands of every middle-school-aged student in the UK. The booth on Programmability of Scalable, Geo-Distributed, Interactive Applications—surely in the running for the wordiest name—is using a variant of the actor model, which was invented by Professor Gul Agha (opens in new tab) of the University of Illinois at Urbana-Champaign, who enjoyed the demo with me and had some great feedback for the interns! Serendipitous moments like this are what make the Faculty Summit special.
The Research Showcase featured 47 demo and poster booths with a range of exciting projects underway at Microsoft, many of them joint efforts with academic institutions.
More than 400 summit attendees viewed the showcase, and a number of them voted for their favorite demos. The results of this highly unscientific popularity contest were as follows:
- Top AI demo: Fully articulated hand tracking (opens in new tab)
- Top Software demo: Madoko (opens in new tab), Code Hunt (opens in new tab), and Vigil (three-way tie)
- Top Devices demo: RoomAlive Toolkit (opens in new tab)
- Top Computing in Society demo: Project Premonition (opens in new tab)
- Top Research in Action demo: Microsoft HoloLens (opens in new tab)
- Top Engaging with Microsoft demo: Project Catapult (opens in new tab)
After the showcase and lunch, it was time for more breakout sessions, including one I led on Programming Models for Estimates, Approximation, and Probabilistic Reasoning (opens in new tab) (maybe I shouldn’t have been so hard on that wordy demo title). A panel of experts—Noah Goodman (opens in new tab) of Stanford University, Dan Grossman (opens in new tab) of the University of Washington, Michael Carbin (opens in new tab) of Microsoft Research and MIT, and Todd Mytkowicz (opens in new tab) of Microsoft Research—provided multiple perspectives on how to deal with the problems that result from using data from sensors, machine learning, approximation and crowdsourcing, all of which have errors. From programming to type theory to probabilistic reasoning, there were enough equations to satisfy even hardcore mathletes. Their presentations attracted a very diverse set of researchers, including experts and theorists in programming languages and AI, and generated a lot of pointed questions.
Then I was off to sit in on the biggest draw of all the breakout sessions—the standing-room-only discussion of The Holoscene: Virtual and Mixed Reality (opens in new tab). This informative and entertaining presentation traced the history and promise of holography, and talked honestly about the problems, both technical and social, that remain to be solved before we all begin living in a mixed environment of the virtual and physical worlds. You will be able to view this session and many of the other breakout sessions online soon. If you’d like to review the presenters’ slides, many of them are already downloadable from the agenda page (opens in new tab).
Peter Lee, corporate VP of Microsoft Research, delivered the concluding keynote.
The concluding keynote from Peter Lee (opens in new tab), corporate VP of Microsoft Research, was aspirational. He described the philosophy and example research projects in our new research division at Microsoft, called NeXT (New Experiences and Technologies), which he is leading. NeXT aspires to create a new model for conducting research and technology transfer—with joy, risk taking and fearless dedication to discovery. He exhorted the audience to double-down on basic and disruptive research, especially today, when politicians and the public often demand to know the ROI for their research investments. He ended his talk with the metaphor of research as a rollercoaster ride: it’s terrifying as you climb up the first hill, but exhilarating once the ride gets rolling.
I hope all of you will heed Peter’s advice—create your own rollercoaster!
—Kathryn S. McKinley (opens in new tab), Principal Researcher, Microsoft Research
Learn more