|
| |
|
|
| Home People Projects Publications Misc | |
| |
 |
|
|
||||
Solving Factored POMDPs with Linear Value Functions (2001)Carlos Guestrin, Daphne Koller, and Ronald ParrAbstract -- Partially Observable Markov Decision Processes (POMDPs) provide a coherent mathematical framework for planning under uncertainty when the state of the system cannot be fully observed. However, the problem of finding an exact POMDP solution is intractable. Computing such solution requires the manipulation of a piecewise linear convex value function, which specifies a value for each possible belief state. This value function can be represented by a set of vectors, each one with dimension equal to the size of the state space. In nontrivial problems, however, these vectors are too large for such a representation to be feasible, preventing the use of exact POMDP algorithms. We propose an approximation scheme where each vector is represented as a linear combination of basis functions to provide a compact approximation to the value function. We also show that this representation can be exploited to allow for efficient computations in approximate value and policy iteration algorithms in the context of factored POMDPs, where the transition model is specified using a dynamic Bayesian network. download information
bibtex citation@inproceedings{Guestrin+al:2001d,author = {Carlos Guestrin and Daphne Koller and Ronald Parr}, title = {Solving Factored {POMDP}s with Linear Value Functions}, year = {2001}, booktitle = "Seventeenth International Joint Conference on Artificial Intelligence (IJCAI-01) workshop on Planning under Uncertainty and Incomplete Information", address = "Seattle, Washington", month = "August", pages = {67 -- 75}, } |
|
|
|
Copyright © 2002-2024 
DAGS - Daphne's Approximate Group of Students. All rights reserved.
| ||