An approach to online optimization of heuristic coordination algorithms

Due to computational intractability, large scale coordination algorithms are necessarily heuristic and hence require tuning for particular environments. In domains where characteristics of the environment vary dramatically from scenario to scenario, it is desirable to have automated techniques for appropriately configuring the coordination. This paper presents an approach that takes performance data from a simulator to train a stochastic neural network that concisely models the complex, probabilistic relationship between configurations, environments and performance metrics. The stochastic neural network is used as the core of a tool that allows rapid online or offline configuration of coordination algorithms to particular scenarios and user preferences. The overall system allows rapid adaptation of coordination, leading to better performance in new scenarios. Categories and Subject Descriptors I.2.6 [Learning]: Connectionism and neural nets; I.6.5 [Model Development]: Modeling methodo...

• Type: Conference
• Year: 2008
• Where: ATAL
• Authors: Jumpol Polvichai, Paul Scerri, Michael Lewis
• Keywords: ATAL 2008, Coordination Algorithms, Intelligent Agents, Scale Coordination Algorithms, Stochastic Neural Network