Summary: Stochastic Filtering in a Probabilistic Action Model
Hannaneh Hajishirzi and Eyal Amir
Computer Science Department
University of Illinois at Urbana­Champaign
Urbana, IL 61801. USA
{hajishir, eyal}uiuc.edu
Abstract
Stochastic filtering is the problem of estimating the state of
a dynamic system after time passes and given partial obser­
vations. It is fundamental to automatic tracking, planning,
and control of real­world stochastic systems such as robots,
programs, and autonomous agents. This paper presents a
novel sampling­based filtering algorithm. Its expected er­
ror is smaller than sequential Monte Carlo sampling tech­
niques given a fixed number of samples, as we prove and
show empirically. It does so by sampling deterministic ac­
tion sequences, and exact filtering of those sequences. These
results are promising for applications in stochastic planning,
natural language processing, and robot control.
1 Introduction

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

Collections: Computer Technologies and Information Sciences