de
en
Schliessen
Detailsuche
Bibliotheken
Projekt
Impressum
Datenschutz
Schliessen
Publizieren
Besondere Sammlungen
Digitalisierungsservice
Hilfe
Impressum
Datenschutz
zum Inhalt
Detailsuche
Schnellsuche:
OK
Ergebnisliste
Titel
Titel
Inhalt
Inhalt
Seite
Seite
Im Werk suchen
Learning shepherding behavior / Michael Baumann. Paderborn, 2016
Inhalt
Introduction
Background
(Multi-)Agent Systems
Single Agent Reinforcement Learning
Growing Neural Gas for Vector Quantization
Related Work
Shepherding Approaches
Discussion of Shepherding Tasks and Approaches
Approximations for Reinforcement Learning
Discussion of Approximation Approaches
The Shepherding Task
Motivation
Biological Background
Description of the Shepherding Task
Modeling the Shepherding Task as Multiagent System
Sheep Behavior
Complexity of the Shepherding Task
Conclusion
Single Agent Shepherding
Foundations
A Greedy Shepherding Algorithm
Analysis of the GCC Algorithm
Conclusion
Learning Shepherding Behavior
Adaptive State Aggregation
Motivation
Theoretical Model of State Space Abstraction
General Approach
From States to State Regions
Neighborhood Connections
Adapting the Approximation
Refining the Approximation
Stopping Criteria
Eligibility Traces for State Regions
Complete Algorithm
Analysis
Conclusion
Adaptive Function Approximation
Motivation
Function Approximation for Reinforcement Learning
Adjusting the Approximation
Smoothing the Approximation
Update Rule
Complete Algorithm
Computational Complexity
Comparison GNG-Q vs. I-GNG-Q
Conclusion
Evaluation
Experimental Results
Experimental Setup
Comparison of Base Configurations for GNG-Q and I-GNG-Q
Evaluation of GNG-Q
Evaluation of I-GNG-Q
Comparison to Other Approaches
Advantages of Adaptive Approximations in Unknown Environments
Shepherding
Conclusion
Conclusion and Future Work
Conclusions
Future Work
Bibliography
Die detaillierte Suchanfrage erfordert aktiviertes Javascript.