Oregon State University

We’d like your feedback: Calendar User Survey – Event Creator Survey



Event Details

PhD Oral Preliminary Examination – Atil Iscen

Friday, November 30, 2012 11:00 AM - 1:00 PM

Reward Shaping for Structural Credit Assignment Problem in Multiagent Learning
Coordinating learning agents to achieve a global task is a key research problem in multiagent systems. Since the agents learn according to the reward that they get from the environment, reward shaping is one of the proven coordination methods to achieve successful coordination. As a reward shaping method, difference rewards provides successful coordination, but is open to improvements from different perspectives. First, we address the amount of information the difference reward requires to provide coordination. Second, we use difference rewards to develop a distributed task decomposition method to provide learning on multiagent multitask problems. Third, we apply difference rewards and multiagent learning algorithms on a novel tensegrity robot system that is hard to control by classical control methods.

Major Advisor: Kagan Tumer
Committee: Prasad Tadepalli
Committee: Alan Fern
Committee: David Porter
Committee: Bill Smart
GCR: Eric Skyllingstad

Kelley Engineering Center (campus map)
Nicole Thompson
1 541 737 5556
Nicole.Thompson at oregonstate.edu
Sch Elect Engr/Comp Sci
This event appears on the following calendars: