Kevin B. Korb, Ann E. Nicholson and Nathalie Jitnah [17] at Monash University are working in the Bayesian Poker Program (BPP). BPP plays two-player five-card stud poker using a Bayesian network structure to represent the relationships between current hand type, final hand type (after the five cards have been dealt) and the behaviour of the opponent. Given evidence for BPP's current hand type and the observed cards and actions of the opponent, BPP obtains its posterior probability of winning the game. BPP uses this estimated probability of winning the game to randomly select its action based on probabilistic curves for each betting action.

BPP performs opponent modeling. It uses the relative frequencies of the opponent's betting actions to update the conditional probabilities per round of passing or calling versus betting or raising given the opponent's current hand type.

BPP is work in progress as pointed out by Korb *et al*. The authors state
that poker appears to be an ideal domain for investigating the application of Bayesian networks,
and report positive results of BPP playing against a simple probabilistic program, a rule-based
program and non-expert amateur human players.