Value Function Approximation workshop at ML95

July 9, 1995
VFA workshop home -- Other ML95 workshops -- ML95 conference home
8:30-9:20 am: Introduction
  • Introduction: Value Function Approximation
    Andrew Moore (CMU)
  • On The Virtues of Linear Learning and Online Training
    Rich Sutton (U. Mass.)
    9:20-10:00: Morning A
  • Online Fitted Reinforcement Learning
    Geoff Gordon (CMU)
  • [Effects of Overestimation in Q-learning with Function Aproximation]
    Sebastian Thrun (Bonn) [to be presented briefly by Andrew Moore]
  • Discussion
    10:30-12:00: Morning B
  • Counterintuitive aspects of TD-Gammon
    Gerry Tesauro (IBM)
  • Reinforcement Learning With Compact Function Approximation: A Case Study from the Elevator Dispatching Domain
    Bob Crites (U. Mass.)
  • Value Function Approximations and Job Shop Scheduling
    Tom Dietterich (Oregon State)
  • Discussion
    1:30-3:15: Afternoon C
  • Robust VFA Algorithms Not Based On Value Iteration
    Justin Boyan (CMU)
  • New Approaches to Dynamic Programming for Continuous Systems
    Chris Atkeson (Georgia Tech)
  • Learning Smooth Control by Fitting Value Function Derivatives
    Satinder Singh (MIT)
  • Fast convergence for value function approximation
    Leemon Baird (USAF)
  • Discussion
    3:45-5:15 pm: Afternoon D
  • AHC for Stochastic Minimum Cost Path Problems Using Neural Net Function Approximators
    Stuart Dreyfus (Berkeley IEOR)
  • Solving Partially Observable Markov Decision Processes Via Function Approximation
    Michael L. Littman (Brown)
  • Input Representation Engineering
    Paul Utgoff (U. Mass.)
  • Panel Discussion
    Evening
  • Informal brainstorming; details to be announced.
    • VFA workshop home -- Other ML95 workshops -- ML95 conference home

    Maintainer:
    Justin.Boyan@cs.cmu.edu