Already, algorithms, which have surpassed human skill, have been developed for games such as chess, checkers, go, backgammon, and more. A team of researchers seek to develop an algorithm that can beat world class poker players in a game that is termed a symmetric. That game is heads-up no-limit Texas hold’em. This game combines a public state of probability between the hands of each player with the strategy of wit and deception. Previous iterations of algorithms have tried and failed to challenge players at major tournaments. The previous renditions used an abstract set of states to predict the how the entire game will unfold. They do this by condensing the available possible states to a fraction of options. However, the loss of information has set these algorithms to fail. A new algorithm, known as DeepStack, is lauded to have the ability to beat these algorithms and players. It uses a modified version of learning and combines past experience with reason and probabilistic states. In essence, DeepStack is beginning to model a human mind, and its success shows that these techniques can be effective in predicting asymmetric situations outside of games like poker.
Link to Article: http://science.sciencemag.org/content/356/6337/508.full