Graduate Thesis Or Dissertation
 

Monte-Carlo planning for probabilistic domains

Public Deposited

Downloadable Content

Download PDF
https://ir.library.oregonstate.edu/concern/graduate_thesis_or_dissertations/3r075032q

Descriptions

Attribute NameValues
Creator
Abstract
  • This thesis presents a progression of novel planning algorithms that culminates in a new family of diverse Monte-Carlo methods for probabilistic planning domains. We provide a proof for performance guarantees and analyze how these algorithms can resolve some of the shortcomings of traditional probabilistic planning methods. The direct policy search methods we developed in collaboration with our local fire department resulted in counter-intuitive recommendations to improve response times for emergency responders. We describe a new family of planning algorithms that combines existing Monte-Carlo methods such as UCT and Sparse Sampling. These algorithms have resulted in groundbreaking empirical performance bounds for one of the most popular computer games of all time, Klondike Solitaire, solving more than 40% of randomly drawn games. These new results significantly improve over our own previously reported results, which represented the first published bound of any kind for Klondike. This builds upon our foundational work that improved performance bounds for a variant of Klondike where the identity of face down cards is known, solving more than 82% of games.
License
Resource Type
Date Available
Date Issued
Degree Level
Degree Name
Degree Field
Degree Grantor
Commencement Year
Advisor
Committee Member
Academic Affiliation
Non-Academic Affiliation
Subject
Rights Statement
Publisher
Peer Reviewed
Language
Replaces

Relationships

Parents:

This work has no parents.

In Collection:

Items

Thumbnail Title Date Uploaded Visibility Actions
file details BjarnasonRonaldV2009.pdf 2017-10-20 Public Download