students

BRETT training with VR

EECS-affiliated startup uses virtual reality to show robots how to perform

The start-up Embodied Intelligence and its founders, Prof. Pieter Abbeel and grad students Peter Chen, Rocky Duan, and Tianhao Zhang, are the focus of two news articles: one from the New York Times titled "A.I. Researchers Leave Elon Musk Lab to Begin Robotics Start-Up," and one from Berkeley News titled "Berkeley startup to train robots like puppets."  The start-up is backed by $7 million in funding from Amplify Partners and other investors and will specialize in complex algorithms that allow machines to learn new tasks on their own through extreme trial and error.  The researchers are augmenting the algorithms with a wide range of techniques, like using virtual reality tools to show a robot how to perform a task--translating the movements into digital data.  “With our advances in machine learning, we can write a piece of software once — machine learning code that enables the robot to learn — and then when the robot needs to be equipped with a new skill, we simply provide new data.” Abbeel explains.

Schematic of a magnetic memory array

EECS-affiliated team develops new, ultrafast method for electrically controlling magnetism in certain metals

A UC Berkeley/UC Riverside research group that includes Prof. Jeffrey Bokor, Prof. Sayeef Salahuddin, postdoc Charles-Henri Lambert, postdoctoral fellow Jon Gorchon, and EE graduate student Akshay Pattabi have developed a new, ultrafast method for electrically controlling magnetism in certain metals, a breakthrough that could lead to greatly increased performance and more energy-efficient computer memory and processing technologies.  Their findings are published in both Science Advances (Vol. 3, No. 49, Nov. 3, 2017) under the title Ultrafast magnetization reversal by picosecond electrical pulses and Applied Physics Letters (Vol. III, No. 4, July 24, 2017) under the title Single shot ultrafast all optical magnetization switching of ferromagnetic Co/Pt multilayers.  “The development of a non-volatile memory that is as fast as charge-based random-access memories could dramatically improve performance and energy efficiency of computing devices,” says Bokor. “That motivated us to look for new ways to control magnetism in materials at much higher speeds than in today’s MRAM.”

Pramod Subramanyan and Rohit Sinha

"A Formal Foundation for Secure Remote Execution of Enclaves" wins Best Paper Award at ACM CCS 2017

A paper co-authored by postdoc Pramod Subramanyan, grad student Rohit Sinha, alumnus Ilia Lebedev (B.S. '10), alumnus and MIT Prof. Srinivas Devadas (M.S. '86/Ph.D. '88), and EECS Prof. Sanjit A. Seshia has won Best Paper Award at the 2017 ACM Conference on Computer and Communications Security (CCS).  The paper, A Formal Foundation for Secure Remote Execution of Enclaves, introduces a formal modeling and verification methodology for secure remote execution based on the notion of a trusted abstract platform.  CCS is the flagship annual conference of the Special Interest Group on Security, Audit and Control (SIGSAC) of the Association for Computing Machinery (ACM).

EECS Accel Scholars

Amit Kumar and Accel launch Accel Scholars EECS mentorship program

EECS alumnus Amit Kumar (B.S. '03) and the venture firm Accel are launching a mentorship program called Accel Scholars to support EECS undergraduates.  Accel will work with a select group of students over the course of a year, hosting networking dinners and also guaranteeing the students an internship at a portfolio company.  Kumar initiated the program because he felt there wasn’t enough career guidance for students at Berkeley and that venture firms that ignore the ecosystem are missing out.  Chair James Demmel says EECS is grateful for the opportunity to “partner with Accel and its network to provide a fast-track for an exceptionally talented and diverse cohort of undergraduates, who will benefit from mentorship by Accel but also by and from one another.”

Ash Bhat and Rohan Phadte (James Tensuan)

Rohan Phadte and Ash Bhat are doing what Twitter won't

EECS undergraduate Rohan Phadte and Interdisciplinary Studies major Ash Bhat are the subjects of a Wired article titled "The College Kids Doing What Twitter Won't," about their creation of a Google Chrome browser extension that checks whether Twitter profiles are bots.  It describes the genesis of their partnership, which they call RoBhat Labs, and their efforts to stop the proliferation of fake Twitter accounts from flooding the internet with propaganda. It also highlights the roles played by CS Assistant Prof. Joseph Gonzalez and the class Data Science 100.

(photo Tiberio Uricchio)

Caffe team wins Everingham Prize at ICCV 2017

The Caffe team researchers ('13 alumnus and current GSR Yangqing Jia, grad student Evan Shelhamer,  '17 alumnus Jeff Donahue, '15 alumnus Sergey Karayev, grad student Jonathan Long, former postdocs Ross Girshick and Sergio Guadarrama, and Prof. in Residence Trevor Darrell) have been awarded the Mark Everingham Prize at the International Conference on Computer Vision (ICCV) 2017.  Caffe is a deep learning framework made with expression, speed, and modularity in mind,  developed by Berkeley AI Research (BAIR) and by community contributors. The Everingham Prize is bestowed by the IEEE technical committee on Pattern Analysis and Machine Intelligence (PAMI) and is given to individuals or groups "who have made a selfless contribution of significant benefit to other members of the computer vision community."  The Caffe team won "for providing an open-source deep learning framework that enabled the community to use, train and share deep convolutional neural networks. Caffe has had a huge impact, both academic and commercial. "
Rebecca Chery presenting her project (photo: Daniel McGlynn)

Rebecca Chery meets PREP design challenge

The experiences of EECS freshman Rebecca Chery, a participant in the Pre-Engineering Program (PREP), are described in a Berkeley Engineering article titled "PREP by design."  PREP is a three-week program that gives incoming engineering majors a head start on academics, networking and professional development. Chery's team used equipment at the Jacobs Institute makerspace to create a phone case with a keyfob inside that would trigger a door to open once the phone case detected a sensor in close proximity. The prototype was chosen by the PREP students as their favorite project from the design challenge.

Dan Garcia
Dan Garcia

Dan Garcia weighs in on necessary skills for coders

Teaching Prof. Dan Garcia is featured in an EdSurge article titled "Engineers, Recruiters and Professors Weigh In: Future Programmers Need Writing Skills, Too," in which he discusses how career goals should shape a student's skill set.  Although not all successful coders need to be proficient writers, flexibility is important.  “There are careers where someone doesn't need [to write]… but we want students to be able to go to any position. Maybe they want to just be a coder [at first], but later they decide to be an academic or on the documentation side or in management,” says Garcia. “My point is you never know when you need to write.”

In a somewhat related Daily Cal article, undergrad Sanil Rajput ponders the correlation between copy editing and computer science, putting forth a theory that "Copy editors make excellent coders."

Musa and Liu (photo: Mujahid Zaman)

Jimmy Liu and Zuhayeer Musa build the future

CS majors Jimmy Liu and Zuhayeer Musa are featured in a Berkeley News article titled "In undergrad startup class, students learn to build the future."  Liu and Musa co-founded a startup called Bash while still in high school.  When they came to Cal, they partnered with CS Prof. Scott Shenker to launch a student-run DeCal class on Berkeley's startup ecosystem last spring, called "How to Build the Future."  The course gives students direct experience with world-renowned entrepreneurs and faculty founders.

Brett and Chelsea Finn

The education of Brett the robot

CS Prof. Pieter Abbeel, graduate student Chelsea Finn, and Brett the robot are featured in a Wired article titled "The Education of Brett the Robot" which delves into some of the nuts and bolts of machine learning.  Brett (short for Berkeley Robot for the Elimination of Tedious Tasks) is using a reinforcement learning algorithm to allow it to learn from its mistakes.  Abbeel will speak on "Deep Learning-to-Learn Robotic Control" at the EECS Colloquium on October 11th.