Learning and information theory.

摘要: A general statistical description of learning processes is given, using an extension classical information theory which able to characterize a system's subjective uncertainty (subjective entropy) about set data and the relevance associated with them. Learning defined as process in entropy or, equivalently, its missing decreases time. This definition implies that enables system optimize responses upon external stimuli such expected profit increases. physical model discussed where probabilities, enter into information, can be considered internal parameters. It conjectured same might true models brain memory. Two forms associative memory are briefly discussed.