From: Christophe Bisciglia (chrisrb_at_cs.washington.edu)
Date: Fri May 09 2003 - 11:26:21 PDT
Anthony Cassandra, Michael L. Littman, Nevin L. Zhang Incremental Pruning:
A Simple, Fast, Exact Method for Partially Observable Markov Decision
Processes
This paper presents an improvement to the Witness algorithm for solving
POMDPs.
The main innovation in this paper was proving that Purge(a + b + c) =
Purge(a + Purge(b + Purge(c))). This equality assists in pruning.
I hate to call this a flaw, but the paper assumes a lot of understanding
that we (well, at least I) don.t necessarily have. It wasn.t until reading
the paper twice, and skimming the reviews that I was able to pull anything
meaningful out.
Open research: Um...yeah...
This archive was generated by hypermail 2.1.6 : Fri May 09 2003 - 11:26:22 PDT