Adaptive Machine Reinforcement Learning

• Autorzy: Poliscuk J.
• Języki publikacji: EN

Abstrakty

EN

In this article is defined a reinforcement learning method, in which a subject of learning is analyzed. The essence of this method is the selection of activities by a try and fail process and awarding deferred rewards. If an environment is characterized by the Markov property, then step-by-step dynamics will enable forecasting of subsequent conditions and awarding subsequent rewards on the basis of the present known conditions and actions, relatively to the Markov decision making process. The relationship between the present conditions and values and the potential future conditions is defined by the Bellman equation. The article discusses also a method of temporal difference learning, mechanism of eligibility traces, as well as their algorithms TD(0) and TD(Lambda). Theoretical analyses were supplemented by the practical studies, with reference to all implementation of the Sarsa(Lambda) algorithm, with replacing eligibility traces and the Epsilon greedy policy.

Słowa kluczowe

EN

algorithm TD(0); algorithm TD(Lambda); Bellman equation; Markov decision making process; mechanism of eligibility traces; method of temporal difference learning; reinforcement learning method

• Wydawca: Wydawnictwo Uniwersytetu Jagiellońskiego
• Czasopismo: Schedae Informaticae
• Rocznik: 2002
• Tom: Vol. 11
• Strony: 57--74

Twórcy

autor

Poliscuk J.

• Department of Electrical Engineering, Poggorica, University of Montenegro, Yugoslavia,