Please use this identifier to cite or link to this item: https://hdl.handle.net/10356/42406
Title: Self-organizing neural architectures and multi-agent cooperative reinforcement learning
Authors: Xiao, Dan
Keywords: DRNTU::Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence
Issue Date: 2010
Source: Xiao, D. (2010). Self-organizing neural architectures and multi-agent cooperative reinforcement learning. Doctoral thesis, Nanyang Technological University, Singapore.
Abstract: Multi-agent system, wherein multiple agents work to perform tasks jointly through their interaction, is a fairly well studied problem. Many approaches to multi-agent learning exist, among which, reinforcement learning is widely used, as it does not require an explicit model of the environment. However, limitations remain in current multi-agent reinforcement learning approaches, including adaptability and scalability in complex and specialized multi-agent domains. In any multi-agent reinforcement learning system, two major considerations are the reinforcement learning methods used and the cooperative strategies among agents. In this research work, we propose to adopt a self-organizing neural network model, named Temporal Difference - Fusion Architecture for Learning, COgnition, and Navigation (TD-FALCON), for multi-agent reinforcement learning. TD-FALCON performs online and incremental learning in real-time with and without immediate reward signals. It thus enables an agent to learn effectively in a dynamic environment.
URI: https://hdl.handle.net/10356/42406
DOI: 10.32657/10356/42406
Fulltext Permission: open
Fulltext Availability: With Fulltext
Appears in Collections:SCSE Theses

Files in This Item:
File Description SizeFormat 
XiaoDan2010.pdfReport955.62 kBAdobe PDFThumbnail
View/Open

Google ScholarTM

Check

Altmetric


Plumx

Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.