Logo: University of Southern California

Events Calendar



Select a calendar:



Filter March Events by Event Type:



Events for March 30, 2017

  • CS Colloquium: Kevin Jamieson (UC Berkeley) - Efficient scalable algorithms for adaptive data collection

    Thu, Mar 30, 2017 @ 11:00 AM - 12:20 PM

    Computer Science

    Conferences, Lectures, & Seminars


    Speaker: Kevin Jamieson, UC Berkeley

    Talk Title: Efficient scalable algorithms for adaptive data collection

    Series: CS Colloquium

    Abstract: This lecture satisfies requirements for CSCI 591: Computer Science Research Colloquium.

    In many applications, data-driven discovery is limited by the rate of data collection: the skilled labor it takes to operate a pipette, the time to execute a long-running physics simulation, the patience of an infant to remain still in an MRI, or the cost of labeling large corpuses of complex images. A powerful paradigm to extract the most information with such limited resources is active learning, or adaptive data collection, which leverages already-collected data to guide future measurements in a closed loop. But being convinced that data-collection should be adaptive is not the same thing as knowing how to adapt in a way that is both sample efficient and reliable. In this talk, I will present several examples of my provably reliable -- and practical -- adaptive data collection algorithms being applied in the real-world. In particular, I will show how my adaptive algorithms are used each week to crowd-source the winner of the New Yorker Magazine Cartoon Caption Contest. I will also discuss my application of adaptive learning concepts at Google to accelerate the tuning of deep networks in a highly parallelized environment of thousands of GPUs.

    Biography: Kevin Jamieson is a postdoctoral researcher working with Professor Benjamin Recht in the Department of Electrical Engineering and Computer Sciences at the University of California, Berkeley. He is interested in the theory and practice of machine learning algorithms that sequentially collect data using an adaptive strategy. This includes active learning, multi-armed bandit problems, and stochastic optimization. Kevin received his Ph.D. from the University of Wisconsin - Madison under the advisement of Robert Nowak. Prior to his doctoral work, Kevin received his B.S. from the University of Washington, and an M.S. from Columbia University, both in electrical engineering.

    Host: CS Department

    Location: Ronald Tutor Hall of Engineering (RTH) - 217

    Audiences: Everyone Is Invited

    Posted By: Assistant to CS chair

    OutlookiCal
  • CS Colloquium: Jyotirmoy V. Deshmukh (Toyota Technical Center) -Ninja Temporal Logic: Making formal methods relevant in engineering practice

    Thu, Mar 30, 2017 @ 04:00 PM - 05:20 PM

    Computer Science

    Conferences, Lectures, & Seminars


    Speaker: Jyotirmoy V. Deshmukh, Toyota Technical Center

    Talk Title: Ninja Temporal Logic: Making formal methods relevant in engineering practice

    Series: CS Colloquium

    Abstract: This lecture satisfies requirements for CSCI 591: Computer Science Research Colloquium.

    The software that controls the operation of critical systems such as vehicles, medical devices, buildings, and transportation infrastructures is getting smarter due to the increased demands for autonomy. The push for increased automation is a worthy goal, but can we do so without compromising the safety and reliability of such systems?
    Furthermore, can formal methods truly improve a design engineer's productivity? In this talk, I will introduce some of the most important questions facing academic and industrial development of software for the cyber-physical systems of tomorrow. We will consider solutions based on the use of formal logics, that, on one hand allow rigorous reasoning about system designs, while on the other, do not place an undue burden on the engineer. In particular, I will explain how formal requirements using real-time temporal logics have had an impact in the development of cutting-edge alternate-energy vehicles and advanced control problems within Toyota. I will guide the audience through an ecosystem built around temporal logic that permits automatic testing, efficient monitoring, requirement engineering and controller synthesis for highly complex automotive systems. The talk covers topics from what I consider the trifecta for designing reliable cyber-physical systems: formal logic, machine learning, and control theory, and will lay out my vision for future research and open problems within this domain.

    Biography: Jyotirmoy V. Deshmukh is a Principal Engineer at Toyota R&D. He received his Ph.D. from the University of Texas at Austin under the supervision of E. Allen Emerson on topics including tree automata, verifying data structure libraries, static analysis for concurrent programs and program repair. He worked as a post-doctoral researcher at the University of Pennsylvania with Rajeev Alur's research group, investigating theoretical models of streaming computation and program synthesis techniques. For the last five years at Toyota, Jyo's research has focused on the design and analysis of industrial cyber-physical systems. Drawing on areas such as hybrid systems, real-time temporal logics, control theory, machine learning and dynamical systems theory, Jyo has been attempting to bridge the gap between academic research and its applicability to industrial-scale systems.

    Host: CS Department

    Location: Henry Salvatori Computer Science Center (SAL) - 101

    Audiences: Everyone Is Invited

    Posted By: Assistant to CS chair

    OutlookiCal