Summary: Learning Partially Observable Deterministic Action Models
Eyal Amir
Computer Science Department
University of Illinois, Urbana­Champaign
Urbana, IL 61801, USA
eyal@cs.uiuc.edu
Abstract
We present the first tractable, exact solution for
the problem of identifying actions' effects in
partially observable STRIPS domains. Our al­
gorithms resemble Version Spaces and Logical
Filtering, and they identify all the models that
are consistent with observations. They apply in
other deterministic domains (e.g., with condi­
tional effects), but are inexact (may return false
positives) or inefficient (we could not bound
the representation size). Our experiments ver­
ify the theoretical guarantees, and show that
we learn STRIPS actions efficiently, with time
that is significantly better than approaches for

Source: Amir, Eyal - Department of Computer Science, University of Illinois at Urbana-Champaign

Collections: Computer Technologies and Information Sciences