News

Shruti Agarwal and Hany Farid use facial quirks to unmask ‘deepfakes’

CS graduate student Shruti Agarwal and her thesis advisor Prof. Hany Farid have created a new weapon in the war against "deepfakes," the hyper-realistic AI-generated videos of people appearing to say and do things they never actually said or did.  The new forensic technique, which uses the subtle characteristics of how a person speaks to recognize whether a new video of that individual is real, was presented this week at the Computer Vision and Pattern Recognition conference in Long Beach.  “The basic idea is we can build these soft biometric models of various world leaders, such as 2020 presidential candidates," said Farid, "and then as the videos start to break, for example, we can analyze them and try to determine if we think they are real or not.”

Berkeley distinguished by number of graduating startup founders

According to Crunchbase News,  UC Berkeley graduated 108 startup founders--not including business school graduates --who raised $1M or more after May 1, 2018.   This makes Berkeley the top-ranked public university, and the third-ranked university of any kind after Stanford and MIT, in founding graduates.  In the Crunchbase tally of all funded founders graduating from public universities (including those with business school degrees), Berkeley (with 240) had more than three times the number of funded founders than second-ranked UCLA (with 85).  Berkeley News notes that you would have to combine the second- through fifth-ranked schools (UCLA, Michigan, Illinois and Washington)  to get to Berkeley’s level. “Berkeley is the original question-the-status-quo, do-disruptive-thinking place,” says Caroline Winnet of Berkeley SkyDeck. “I like to say that we don’t just think outside the box. There is no box.”

Elizaveta Tremsina is 2019 ACM SRC Grand Finals Winner

A paper written by recent graduate Elizaveta Tremsina (B.S. '19 CS/Physics/Applied Math) has taken third place in the undergraduate category of the 2019 ACM Student Research Competition (SRC) Grand Finals.  The paper, titled "Your Story Recorded in a Magnet: Micromagnetic Simulations of Spin-Orbit Torque in Multi-layer Structures," was a continuation of the first place poster she presented at the 2018 Richard Tapia Celebration of Diversity in Computing.    "I am extremely thankful to the Berkeley EECS department for the amazing 4.5 years and for the unique chance to participate in cutting-edge research with Dr. Salahuddin's group and also attend the Tapia conference (my first one back in 2016 and last year)," she said. "I hope that more Berkeley undergrads participate in this competition in the future, be it at Tapia or other ACM conferences."  Tremsina was presented with her award at the ACM awards banquet last weekend.

Berkeley paper wins 2018 IEEE EDS George E. Smith Award

"Improved Subthreshold Swing and Short Channel Effect in FDSOI n-Channel Negative Capacitance Field Effect Transistors," has won the 2018 IEEE Electron Devices Society (EDS) George E. Smith Award.  The paper was co-authored by current postdoc Korok Chatterjee, graduate student Ava J. Tan, former postdocs Daewoong Kwon,  Angada B. Sachid, Ajay K. Yadav and Hong Zhou, EE Profs. Chenming Hu and Sayeef Salahuddin, and LBNL's Roberto dos Reis. The award recognizes the best paper appearing in a fast turn around archival publication of the IEEE Electron Devices Society, targeted to IEEE Electron Device Letters.

Justin Yim wins Best Student Paper Award at ICRA 2019

EECS PhD student Justin Yim (with advisor EECS Prof. Ron Fearing and ME undergraduate co-author Eric Wang) has won the best student paper award at the 2019 IEEE International Conference on Robotics and Automation (ICRA) (May 20-24, Montreal) for his paper "Drift-free Roll and Pitch Estimation for High-acceleration Hopping."  The robot "Salto" (Saltatorial Agile Locomotion on Terrain Obstacles) was previously restricted to only indoor operation in a room equipped with a motion tracking system. In the newest paper, Salto can now estimate its own position by combining its onboard inertial sensor with a model of its takeoffs and landings. This improved estimate allows
Salto to hop outside with human steering.

Tianshi Wang and Jaijeet Roychowdhury win UCNC 2019 Best Paper Award

A paper co-authored by freshly minted alumnus Tianshi Wang (Ph.D. '19, winner of the 2019 EECS David Sakrison Memorial Prize for "truly outstanding research") and Prof. Jaijeet Roychowdhury has won Best Paper Award at the International Conference on Unconventional Computation and Natural Computation (UCNC) 2019.  The paper, titled "OIM: Oscillator-based Ising Machines for Solving Combinatorial Optimisation Problems" will be presented at the conference in Japan next week.

Moses Surumen plugs Kenya’s skills gap with peer to peer learning

Moses Surumen, who graduated with a degree in EECS this week, has been sharing his knowledge with peers in Kenya for the past two years, helping them develop the skills to solve challenges back home.  Surumen, who has 10 siblings, grew up in Kajiado, a Masai area south of Nairobi.  In 2017, he implemented a program called M-Soma, running a six-week summer course for Kenyan high school graduates in computer science.  “We were building skills the way Berkeley does, providing the best skeletal code for setting up the platform and building onto that several features they wanted to use,” he explains.  Surumen has accepted a position at Qualcomm but plans to continue to explore how to scale his project to work in different African countries.

With a hop, a skip and a jump, Salto leaps over obstacles with ease

Salto the robot, first unveiled in 2016 by the Biomimetic Millisystems Lab, is now equipped with a slew of new skills, giving it the ability to bounce in place like a pogo stick and jump through obstacle courses like an agility dog. Salto can even take short jaunts around campus, powered by a radio controller.  Salto creators Justin Yum, Eric Wang and Ronald S. Fearing will describe the robot’s new skills today (Tuesday, May 21) at the 2019 International Conference on Robotics and Automation in Montreal.

Chelsea Finn wins 2018 ACM Doctoral Dissertation Award

Recent graduate Chelsea Finn (Ph.D. '18, advisors: Pieter Abbeel and Sergey Levine), has won the prestigious ACM Doctoral Dissertation Award. This award is presented annually to "the author(s) of the best doctoral dissertation(s) in computer science and engineering."  In her dissertation, "Learning to Learn with Gradients," Finn introduced algorithms for meta-learning that enable deep networks to solve new tasks from small datasets, and demonstrated how her algorithms can be applied in areas including computer vision, reinforcement learning and robotics.  Finn  is currently a research scientist at Google Brain, a post-doc at the Berkeley AI Research Lab (BAIR), and an acting assistant professor at Stanford.  Last year's recipient, Aviad Rubinstein, was also a Berkeley EECS alum.

Soham Phade and Venkat Anantharam win GameNets Best Paper Award

Graduate student Soham Phade and his advisor, Venkat Anantharam, have won the Best Paper Award at the 9th EAI International Conference on Game Theory for Networks (GameNets 2019).  Their paper, titled "Optimal Resource Allocation over Networks via Lottery-Based Mechanisms," was in the Games for Economy and Resource Allocation category.  Phade's current focus is on "designing market-based mechanisms and algorithms on presumably more accurate models of human behavior from psychology and decision theory, for increasing human welfare and for building more efficient commercial systems that interact with humans."