Lecture: Dissertation Talk: CS | April 29 | 1-2 p.m. | 405 Soda Hall
Modern deep learning techniques are data-hungry, which presents a problem in robotics because real-world robotic data is difficult and expensive to collect. In contrast, simulated data is cheap and scalable, but jumping the "reality gap" to use simulated data for real-world tasks is challenging. In this talk, we discuss applications of Domain Randomization: a technique for bridging the reality... More >
Dissertation Talk: Optoelectronics for refrigeration and analog circuits for combinatorial optimization
Presentation: Dissertation Talk: EE | April 29 | 1-2 p.m. | 540AB Cory Hall
Tianyao Xiao, PhD candidate, Department of EECS, UC Berkeley
In this dissertation talk, I will cover two topics. First, I will discuss the prospects of using light-emitting diodes as solid-state refrigerators. Second, I will present a non-von Neumann computer, built from coupled analog electrical oscillators, that can rapidly search for solutions to difficult combinatorial optimization problems.
Seminar: Berkeley Sensor & Actuator Center (BSAC): EE | April 30 | 12:30-1:30 p.m. | 521 Cory Hall
Millions suffer from spinal cord problems and many have bladder dysfunction. We aim to build a non-invasive device utilizing LEDs to infer bladder expansion to minimize patient inconvenience. Clinical problems will be presented and potential solutions will be discussed.
RSVP online by April 29.
Dissertation talk: Evaluation of Methods for Data-Driven Tools that Empower Mental Health Professionals
Lecture: Dissertation Talk: CS | April 30 | 4-5 p.m. | 190 Doe Library
It is estimated that nearly one in five adults in the United States live with mental illness, and for individuals who struggle with mental health, the experience can be excruciating. The rise of mobile devices presents a unique opportunity to improve mental health outcomes, in part through empowering mental health professionals. Because many individuals always have their smartphones with them,... More >
Scientific Computing and Matrix Computations Seminar: Why Deep Learning Works: Traditional and Heavy-Tailed Implicit Self-Regularization in Deep Neural Networks
Seminar: Scientific Computing: CS | May 1 | 12-1 p.m. | 380 Soda Hall
Michael W. Mahoney, ICSI and Department of Statistics, University of California at Berkeley
Random Matrix Theory (RMT) is applied to analyze the weight matrices of Deep Neural Networks (DNNs), including both production quality, pre-trained models and smaller models trained from scratch. Empirical and theoretical results clearly indicate that the DNN training process itself implicitly implements a form of self-regularization, implicitly sculpting a more regularized energy or penalty... More >
Seminar: Dissertation Talk: EE | May 1 | 3-4 p.m. | 380 Soda Hall
Stephen Tu, UC Berkeley
Reinforcement learning (RL) has demonstrated impressive performance in various domains such as video games, Go, robotic locomotion, and manipulation tasks. As we turn towards RL to power autonomous systems in the physical world, a natural question to ask is, how do we ensure that the behavior observed in the laboratory reflects the behavior that occurs when systems are deployed in the real world?... More >
Rapidly mixing random walks on matroids and related objectsidly mixing random walks on matroids and related objects
Seminar: CS | May 1 | 3-4 p.m. | 1011 Evans Hall
Nima Anari, Stanford University
A central question in randomized algorithm design is what kind of distributions can we sample from efficiently? On the continuous side, uniform distributions over convex sets and more generally log-concave distributions constitute the main tractable class. We will build a parallel theory on the discrete side, that yields tractability for a large class of discrete distributions. We will use this... More >
Center for Computational Biology Seminar: Sohini Ramachandran, Associate Professor, Brown University
Seminar: Biosystems and Computational Biology: CS | May 1 | 4:30-5:30 p.m. | 125 Li Ka Shing Center
Leveraging linkage disequilibrium to identify adaptive and disease-causing mutations
Correlation among genotypes in human population-genetic datasets complicates the localization of both adaptive mutations and disease-causing mutations. I will describe our latest efforts to develop new methods for localizing adaptive and disease-causing mutations, motivated by (1) incorporating... More >
Seminar: Dissertation Talk: EE | May 2 | 2-3 p.m. | 606 Soda Hall
Frank Li, UC Berkeley
Seminar: Dissertation Talk: CS | May 3 | 11 a.m.-12 p.m. | 465H Soda Hall
In this thesis, we argue that serverless functions represent a viable platform for analytics workloads, eliminating cluster management overhead, fulfilling the promise of elasticity. We identify and provide solutions to address two major issues in realizing this vision.
Solid State Technology and Devices Seminar: 24/7 Electricity Produced by Intermittent Power Requires Its Energy Storage
Seminar: Solid State Technology and Devices: EE: CS | May 3 | 1-2 p.m. | Cory Hall, The Hogan Room, 521
Jerry Woodall, Electrical and Computer Engineering, UC Davis
This is a simple story with a no-brainer punchline included in the title. Except for geothermal and nuclear energy, the sun is, and has been, the source of nearly all energy used on our planet. The problem is that the earth receives plenty of intermittent solar power, but not as solar energy.
Seminar: Dissertation Talk: CS | May 3 | 1:30-2:30 p.m. | 306 Soda Hall
Jingcheng Liu, UC Berkeley
In classical statistical physics, a phase transition is understood by studying the geometry (the zero-set) of an associated polynomial (the partition function). In this talk I will show that one can exploit this notion of phase transitions algorithmically, and conversely exploit the analysis of algorithms to understand phase transitions. As applications, I will give efficient deterministic... More >
Seminar | May 3 | 2-3 p.m. | 4 LeConte Hall
Watch a brief animated explainer of bionanotechnology at http://www.shawndouglas.com
Seminar: Dissertation Talk: EE | May 3 | 2:30-3:30 p.m. | 531 Cory Hall
Ankush Pankaj Desai, University Of California, Berkeley
Asynchronous event-driven systems can be found in myriad domains including cloud computing systems, device drivers, and robotics.
These systems are notoriously hard to get right as the programmer needs to reason about numerous control paths resulting from the complex interleaving of events (messages) and failures.
Unsurprisingly, it is easy to introduce subtle errors while attempting to fill... More >
Seminar: Dissertation Talk: EE | May 3 | 3-4 p.m. | 250 Sutardja Dai Hall
As robots become more capable and commonplace, it is increasingly important that the policies they execute are transparent. For instance, engineers should have an idea of which situations their robot may act incorrectly in, and end-users should be able to anticipate how a robot they are interacting with will behave in various situations. This is essential for building trust, enabling seamless... More >