learn commun relat reinforc learn relat reinforc learn rrl emerg machin learn commun promis subfield reinforc learn rl upgrad rl techniqu relat represent state action learn valuefunct learn commun relat reinforc learn relat reinforc learn rrl emerg machin learn commun promis subfield reinforc learn rl upgrad rl techniqu relat represent state action learn valuefunct bay drelat learn oppon model incomplet inform dlimit poker propos oppon model approach nolimit texa hold em poker start learn prior gener expect oppon behavior learn relat regress tree function adapt prior specif oppon import asset approach learn incomplet inform know player hand train game decis tree answer network tempor differ dnetwork state represent intellig agent continu challeng abstract unavoid larg state space predict represent offer obtain state abstract replac state set predict futur interact gener order dsearch learn direct probabilist logic model abstract recent increas interest direct probabilist logic model varieti languag describ model propos author provid high level argument show principl model languag learn data propos learn algorithm studi detail introduc algorithm gener order search learn structur condit probabl distribut cpd direct probabilist logic model algorithm upgrad order search algorithm bayesian network relat probabl tree represent cpd present experi block world domain gene domain cora dataset onlin learn exploit relat model reinforc learn recent year grow interest rich represent relat languag reinforc learn express languag advantag term gener reason extend exist approach relat set trivial problem paper present step onlin learn exploit relat model propos represent transit reward function learn onlin present method exploit model augment relat reinforc learn algorithm plan techniqu benefit robust approach evalu experiment learn relat option induct transfer relat reinforc learn abstract reinforc learn problem agent task learn good optim strategi interact environ start learn task agent inform face complex problem larg state space learn good strategi infeas slow work practic overcom problem guidanc suppli agent trace reason polici lot case hard user suppli polici paper investig transfer learn relat reinforc learn goal transfer learn acceler learn target task train relat sourc task specif introduc extens option framework relat set show learn skill transfer similar domain present experi show benefit relat option transfer learn gener order dsearch learn direct probabilist logic model abstract recent increas interest direct probabilist logic model varieti languag describ model propos author provid high level argument show principl model languag learn data propos learn algorithm studi detail introduc algorithm gener order search learn structur condit probabl distribut cpd direct probabilist logic model algorithm upgrad order search algorithm bayesian network relat probabl tree represent cpd present experi block world domain gene domain cora dataset multi dagent relat reinforc learn abstract paper studi relat reinforc learn multi agent set grow evid reinforc learn research commun relat represent state space benefit proposit complex task plan inform retriev web repres natur relat form relat structur exploit multi agent reinforc learn task studi singl agent context paper attempt bridg gap relat reinforc learn rrl multi agent system ma precis explor relat structur state space multi agent reinforc learn context 