News

Stuart Russell shares his favorite algorithms

CS Prof. Stuart Russell is one of four top experts asked by The Next Web (TNW) "which algorithms they think made the biggest contribution to artificial intelligence and science in general?"  Russell chose Lookahead and Backchaining, which he described as "fundamental modes of decision making.”  Other experts chose Gradient Descent, Convolutional Networks, and Forward-backward.

Joe Hellerstein on the must-haves of a modern data prep platform

CS Prof. Joseph Hellerstein is the subject of a feature in InsideBigData titled "The Must-Haves of a Modern Data Prep Platform."  Hellerstein is the co-founder and Chief Strategy Officer at Trifacta, a company that develops data wrangling software for data exploration and self-service data preparation for analysis.  he discusses how the challenge of data preparation sits squarely between the growth of BI and visualization tools and the specific data needed to fuel them.  Efficient data preparation is key to alleviating new demand from business users. The article offers three key requirements that a data preparation platform should have.  Hellerstein's career in research and industry has focused on data-centric systems and the way they drive computing. Fortune Magazine included him in their list of 50 smartest people in technology , and MIT’s Technology Review magazine included his work on th eir TR10 list of the 10 technologies “most likely to change our world.”

Constantinos Daskalakis wins Rolf Nevanlinna Prize

CS alumnus Constantinos Daskalakis (Ph.D. '08, advisor: Christos Papadimitriou) has won the Rolf Nevanlinna Prize at the International Congress of Mathematicians, one of the highest awards in theoretical computer science.  Daskalakis, who is currently a professor at MIT,  was cited for his work on game theory and machine learning.  He is profiled in a Quanta Magazine article titled "A Poet of Computation Who Uncovers Distant Truths," that describes his fruitful time at Berkeley with Papadimitriou.

How Robot Hands Are Evolving to Do What Ours Can

The New York Times has published a front page article featuring research being done in the EECS department.  "How Robot Hands Are
Evolving to Do What Ours Can" details how robotic hands could once only do what vast teams of engineers programmed them to do but--thanks to research being done at places like Berkeley--can now learn more complex tasks on their own.  The article breaks tasks down into 5 categories, 4 of which are illustrated by work being done in Prof. Ken Goldberg's AUTOLAB:  gripping, picking, bed-making, and pushing.    Although these tasks are limited, the machine learning methods that drive these systems point to continued progress in the years to come.

Five Questions for David Patterson

CS Prof. Emeritus David Patterson, winner of the 2017 ACM A.M. Turing Award, answers 5 questions posed by the Cal Alumni Association's California Magazine.   Topics include the unsurpassed number of Berkeley Turing laureates, the dangers of AI, the RISC revolution, Patterson's classic textbook on computer architecture, and how much weight he can bench press.  You can attend lectures by many of U.C. Berkeley's prominent Turing laureates, including Patterson,  this fall at the Berkeley ACM A.M. Turing Laureate Colloquium.

Pieter Abbeel stresses cooperation key to advancing AI application

CS. Prof. Pieter Abbeel is the subject of a China Daily article titled "Professor stresses cooperation key to advancing AI application."  Abbeel attended the China-US Entrepreneur and Investment Summit  in Santa Clara where he discussed recent advances and trends in artificial intelligence (AI) and applications in robotic automation, calling for more collaboration worldwide in order to make robots ultimately serve the people.  "The articles of our research findings are in archives," he said.  "Anyone can read it. There are also three important global deep-learning conferences so you can present your work and meet people in the AI field worldwide."

Ming Wu and Steven Conolly named Bakar Fellows

EECS Profs. Ming Wu and Steven Conolly been selected for the Bakar Fellows Program, which supports faculty working to apply scientific discoveries to real-world issues in the fields of engineering, computer science, chemistry and biological and physical sciences.  Wu's fellowship support will be used accelerate commercialization of his invention: a high-performing silicon photonic switch for data center networks.  Conolly's laboratory is developing a high-resolution three-dimensional imaging method, Magnetic Particle Imaging, which does not use any radiation and has unprecedented sensitivity.

Jitendra Malik takes position at Facebook

Facebook has announced that it has hired EECS Prof. Jitendra Malik in an effort to expand its artificial intelligence research.  Malik, one of the most influential researchers in computer vision, will be based at the Menlo Park lab, where Facebook Artificial Intelligence Research (FAIR) is headquartered.  He will retain part-time affiliation with U.C. Berkeley to advise students; the Berkeley AI Research (BAIR) Lab is one of several receiving funding from FAIR.  “He has been influential in shaping Berkeley’s AI group into the exceptional lab that it is today, and we look forward to his help in continuing the growth of FAIR,” Facebook chief AI scientist Yann LeCun wrote in a news release.  LeCun added that Facebook plans to support a number of doctoral students who will conduct research in collaboration with researchers at FAIR and their university faculty, or on topics of interest to FAIR under the direction of their faculty.

Lensless Cameras May Offer Detailed Imaging of Neural Circuitry

EECS graduate students Nick Antipa and Grace Kuo, along their advisor Associate Prof. Laura Waller, have penned an article for Photonics Media titled "Lensless Cameras May Offer Detailed Imaging of Neural Circuitry" about a new architecture which could enable simultaneous monitoring of millions of neurons in 3D space at frame rates limited only by image sensor read times.  Instead of using a large, lens-based light-field microscope to image individual brain neurons, the DiffuserCam lensless imaging architecture consists of a diffuser placed in front of a 2D image sensor. When an object is placed in front of the diffuser, its volumetric information is encoded into a single 2D measurement.   Borrowing tools from the field of compressed sensing, a 3D image is reconstructed by solving a sparsity-constrained optimization problem.

RAFAR wins Best Student Paper Award at MARSS 2018

"Bidirectional thin-film repulsive-/attractive-force electrostatic actuators for a crawling milli-robot," written by recent EE alumnus Ethan Schaler (Ph.D. '18), his advisor Prof. Ron Fearing, and two undergraduates from other departments (Loren Jiang in BioE and Caitlyn  Lee in E3S), received the Best Student Paper Award  from the International Conference on Manipulation, Automation, and  Robotics at Small Scales (MARSS) 2018 in Nagoya, Japan in July. The authors demonstrated a new thin-film electrostatic actuator (RAFA)  capable of generating bidirectional repulsive- and attractive-forces:  156 Pa in repulsion and 352 Pa in attraction, when operating at up to  1.2 kV. They used this actuator to power RAFAR, a 132 mg milli-robot  that crawls at 0.32 mm/s with anisotropic friction feet.   Schaler will be joining NASA Jet Propulsion Laboratory (JPL) this summer.