CS Assistant Prof. Sergey Levine explores how deep learning will unleash robotics in an NVIDIA AI Podcast which first aired on Sept 1st.  “One of the most important things is that you have to somehow communicate to the robot what it means to succeed,” Levine said in a conversation with AI Podcast host Michael Copeland. “That’s one of the most basic things …You need to tell it what it should be doing.”  He points out that it’s important that the robots don’t just repeat what they learn in training, but understand why a task requires certain actions. “If you want to get a robot to do interesting things, you kind of need it to learn on its own,” Levine said

Graduate student Yang You (advisor: James Demmel) has won a 2017 ACM IEEE Computer Society George Michael Memorial Fellowship for his work on designing accurate, fast, and scalable machine learning algorithms on distributed systems.   The award, which was named in honor of George Michael, one of the founding fathers of the Supercomputing (SC) Conference series, is given in recognition of overall potential for research excellence in subjects of interest to the High Performance Computing (HPC) community.  In You's most recent work, “Scaling Deep Learning on GPU and Knights Landing Clusters,” his goal is to scale up the speed of training neural networks so that networks which are relatively slow to train can be redesigned for high performance clusters. This approach has reduced the percentage of communication from 87% to 14% and resulted in a five-fold increase in speed.

Undergraduate students Jian Lu (EECS junior), Walt Leung (CS sophomore), Jiayi Chen (CS junior), and Malhar Patel (EECS junior) placed 3rd at the Greylock Hackfest in July.  Their platform, BeAR, allows multiple users to connect to the same #AR (augmented reality) session.  The Hackfest, sponsored by Greylock Partners, allows 45 teams of up to four university students the opportunity to show what they can build to a panel of tech industry  judges.  Hacks are judged based on five different criteria: level of difficulty, aesthetics, originality, usefulness, and your project’s “WOW factor.”

CS alumnus Andrew Ng (Ph.D. '02), one of the world's leading authorities on AI, interviews EE Prof. Pieter Abbeel for Heroes of Deep Learning, an interview series from Ng's cousera course, Deep learning AI.  “Work in Artificial Intelligence in the EECS department at Berkeley involves foundational research in core areas of knowledge representation, reasoning, learning, planning, decision-making, vision, robotics, speech and language processing," Abbeel says. "There are also significant efforts aimed at applying algorithmic advances to applied problems in a range of areas, including bioinformatics, networking and systems, search and information retrieval. There are active collaborations with several groups on campus, including the campus-wide vision sciences group, the information retrieval group at the I-School and the campus-wide computational biology program. There are also connections to a range of research activities in the cognitive sciences, including aspects of psychology, linguistics, and philosophy. Work in this area also involves techniques and tools from statistics, neuroscience, control, optimization, and operations research. Berkeley Artificial Intelligence Research Lab (BAIR)."

CS alumnus Andrew Ng (Ph.D. '02, adviser: Michael Jordan) has been singled out by NewsCenter.io as one of 7 leaders shaping the AI revolution.  Ng founded the “Google Brain” project, which developed massive-scale deep learning algorithms.  He led the AI group at Baidu, China’s largest search engine company, which directed research into advertising, maps, take-out delivery, voice and internet searching, security, consumer finance, among others. Ng also co-founded Coursera, an online education company that has raised more than $200 million venture capital funding.  He is also currently an adjuct professor at Stanford.

CS graduate student Rebecca Portnoff (adviser: David Wagner) has developed the first algorithm to identify adult ads tied to human trafficking rings by linking the ads to public information from Bitcoin — the primary payment method for online sex ads.  “The technology we’ve built finds connections between ads,” says Portnoff.  “Is the pimp behind that post for Backpage also behind this post in Craigslist? Is he the same man who keeps receiving Bitcoin for trafficked girls? Questions like these are answerable only through more sophisticated technological tools – exactly what we’ve built in this work – that link ads together using payment mechanisms and the language in the ads themselves.”  Her team will present their findings this month at the Association for Computing Machinery’s SIGKDD Conference on Knowledge Discovery and Data Mining.

CS Prof. Katherine Yelick will give the HPC keynote on Exascale computing at the upcoming ACM Europe Conference. Yelick also serves as Associate Laboratory Director for Computing Sciences at Lawrence Berkeley National Laboratory.  The event, which will take place on Sept. 7-8 in Barcelona, Spain, will focus on the themes of Cybersecurity and High Performance Computing.

CS graduate student Grant Ho, Aashish Sharma (LBNL),  CS alumna Mobin Javed (Ph.D. 2016), and CS Profs. Vern Paxson and David Wagner have won the 2017 Internet Defense Prize, worth $100,000, for their paper "Detecting Credential Spearphishing in Enterprise Settings."  CS graduate student Thurston Dang,  Petros Maniatis (Google Brain), and Prof. David Wagner, were finalists for their paper "Oscar: A Practical Page-Permissions-Based Scheme for Thwarting Dangling Pointers."  The award, which is funded by Facebook and offered in partnership with USENIX, recognizes research that meaningfully makes the internet more secure.

CS Assistant Prof. Anca Dragan has been named one of 2017's 35 Innovators Under 35  by MIT Technology Review.   Each year, exceptionally talented young innovators are singled out for the honor because their work is thought to offer the greatest potential to transform the world.   Dragan was nominated in the Visionary category for "Ensuring that robots and humans work and play well together" and is profiled in an MIT Technology Review article.  She will also be recognized at a special ceremony at EmTech MIT.

Assistant Prof. Sergey Levine, Prof. Pieter Abbeel, and graduate student Chelsea Finn are featured in a CSO article highlighting research they presented at the International Conference on Machine Learning (ICML).  The research was done in partnership with the NVIDIA AI Labs (NVAIL) programme.  Levine’s team wants to help intelligent agents learn faster and require less training by teaching deep neural networks to learn more like humans.  “Look at how people do it,” said Levine. “We never learn things entirely from scratch. We draw on our past experience to help us learn new skills quickly. So we’re trying to get our learning algorithms to do the same.”  Levine and his team have been using an NVIDIA DGX-1 system to train their algorithms how to coordinate movement and visual perception.

NVAIL helps keep AI pioneers ahead of the curve with support for students, assistance from researchers and engineers, and gives them access to the industry’s most advanced GPU computing power.