Risk-sensitive methods in deception games

We introduce the application of semi-field analysis to stochastic dynamic games of information. Such problems typically have a high curse-of-dimensionality; hence a method of managing this issue is developed by formulating the payoff function as a min-log sum whose structure is preserved. This technique is applied to a problem of deception and search with a risk-sensitive payoff function where the information is represented using a log-plus probability space.