You are in : Home » Results & submissions » Select UOA » 11 - Computer Science and Informatics » View submission: University of York » Outputs » Detail

Output details

11 - Computer Science and Informatics

Return to search Previous output Next output

Output 0 of 0 in the submission

Article title

Online learning of shaping rewards in reinforcement learning

Type

D - Journal article

DOI

Title of journal

Neural Networks

Article number

Volume number

Issue number

First page of article

541

ISSN of journal

0893-6080

Year of publication

2010

URL

Number of additional authors

Additional information

<12>This paper presents for the first time how potential-based reward shaping can be

automatically computed in the absence of explicit domain knowledge. In addition, the

proposed method preserves the complexity guarantees of the underlying reinforcement learning

techniques. The empirical study shows rigour, having been performed on two representative,

but qualitatively highly different evaluation domains.

Interdisciplinary

Cross-referral requested

Research group

G - Artificial Intelligence

Citation count

Proposed double-weighted

Double-weighted statement

Reserve for a double-weighted output

Non-English

English abstract