History-dependent graphical multiagent models

History-dependent graphical multiagent models,10.1145/1838206.1838364,Quang Duong,Michael P. Wellman,Satinder P. Singh,Yevgeniy Vorobeychik

History-dependent graphical multiagent models  
BibTex | RIS | RefWorks Download
A dynamic model of a multiagent system defines a proba- bility distribution over possible system behaviors over time. Alternative representations for such models present trade- os in expressive power, and accuracy and cost for inferen- tial tasks of interest. In a history-dependent representation, behavior at a given time is specified as a probabilistic func- tion of some portion of system history. Models may be fur- ther distinguished based on whether they specify individual or joint behavior. Joint behavior models are more expres- sive, but in general grow exponentially in number of agents. Graphical multiagent models (GMMs), introduced by Duong et al., provide a more compact representation of joint behav- ior, when agent interactions exhibit some local structure. We extend GMMs to condition on history, thus supporting inference about system dynamics. To evaluate this hGMM representation we study a voting consensus scenario, where agents on a network attempt to reach a preferred unani- mous vote through a process of smooth fictitious play. We induce hGMMs and individual behavior models from exam- ple traces, showing that the former provide better predic- tions, given limited history information. These hGMMs also provide advantages for answering general inference queries compared to sampling the true generative model.
Cumulative Annual
View Publication
The following links allow you to view full publications. These links are maintained by other sources not affiliated with Microsoft Academic Search.