• 12 September 2007
Abstract
The principles of statistical mechanics and information theory play an important role in learning and have been widely studied. The new aspect here is a focus on integrating feedback from the observer. A quantitative approach to interactive learning and adaptive behavior is proposed, integrating model- and decision-making into one theoretical framework. Following simple principles, requiring that the observer's world model and action policy should result in maximal predictive power at minimal complexity, an objective function is proposed which reflects this trade-off between prediction and complexity. A fundamental consequence of the feedback is that the optimal action policy balances exploration and control. The optimal model reflects the process's causal organization.

This publication has 0 references indexed in Scilit: