A reinforcement learning method, device, electronic device and storage medium

Abstract

An embodiment of the present invention provides a reinforcement learning method, device, electronic equipment, and storage medium. The method includes: determining the observation state of the agent after executing the current execution action as the next observation state; The mapping relationship between the state and the rule state determines the next rule state corresponding to the next observation state; based on the preset agent execution strategy and the next rule state, determines the next execution action of the agent, and based on the next rule state and the next execution action to determine the next action utility value; based on the next action utility value, update the current action utility value of the agent until the preset interaction termination condition is met. The reinforcement learning method, device, electronic device, and storage medium provided by the embodiments of the present invention greatly reduce the amount of calculation of action utility values, reduce the calculation dimension, and improve the efficiency of reinforcement learning.

Description

technical field [0001] The present invention relates to the technical field of artificial intelligence, in particular to a reinforcement learning method, device, electronic equipment and storage medium. Background technique [0002] Reinforcement learning mainly studies the decision-making and actions of agents of varying numbers based on their own and external information. The agent perceives the surrounding environment through the interaction with the external environment, and obtains the evaluation of the effectiveness of the action by the environment by executing an action or command, so as to adjust its own strategy. With the complexity of the problem, such as multi-agent, delayed return, sparse return, etc., the training process of reinforcement learning is often time-consuming, and it is difficult to obtain an ideal action strategy. [0003] The existing reinforcement learning method is based on the deep neural network to fit the utility value, and this method is mor...

AI Technical Summary

Problems solved by technology

[0004] Embodiments of the present invention provide a reinforcement learning method, device, electronic equipment, and storage medium to solve the technical problems of the existing reinforcement learning methods, which have a large amount of calculation and low learning efficiency

Images