Integrating On-policy Reinforcement Learning with Multi-agent Techniques for Adaptive Service Composition.
Hongbing WangXin ChenQin WuQi YuZibin ZhengAthman BouguettayaPublished in: ICSOC (2014)
Keyphrases
- service composition
- reinforcement learning
- multi agent
- optimal policy
- web service composition
- web services
- qos aware
- actor critic
- petri net
- action selection
- function approximation
- goal driven
- policy search
- web services composition
- service oriented
- partially observable markov decision processes
- composition of web services
- partially observable
- state space
- reinforcement learning algorithms
- policy gradient
- asynchronous communication
- markov decision process
- service oriented computing
- function approximators
- service selection
- service integration
- multi agent environments
- business requirements
- single agent
- composite services
- ai planning
- markov decision processes
- action space
- description language
- adaptive learning
- petri net model
- composite web services
- ws bpel
- control policy
- infinite horizon
- dynamic programming
- learning algorithm
- service oriented architecture
- learning process