| | |
Summary: Sampling First Order Logical Particles
Hannaneh Hajishirzi and Eyal Amir
Department of Computer Science
University of Illinois at Urbana-Champaign
{hajishir, eyal}@uiuc.edu
Abstract
Approximate inference in dynamic systems is
the problem of estimating the state of the sys-
tem given a sequence of actions and partial ob-
servations. High precision estimation is funda-
mental in many applications like diagnosis, natu-
ral language processing, tracking, planning, and
robotics. In this paper we present an algorithm
that samples possible deterministic executions of
a probabilistic sequence. The algorithm takes ad-
vantage of a compact representation (using first
order logic) for actions and world states to im-
prove the precision of its estimation. Theoretical
and empirical results show that the algorithm's
expected error is smaller than propositional sam-
|
