Knowledge strategy selection method and device based on reinforcement learning

Abstract

The invention discloses a knowledge strategy selection method and device based on reinforcement learning, and the method comprises a lower layer and an upper layer: taking n decision-making systems of different sources of the lower layer as first-level strategy sources, respectively inputting the first-level strategy sources, and generating n different decision-making results; n different decision results are fused through an upper meta-learning agent trained based on a reinforcement learning algorithm, and a final decision is generated and output; and the meta-learning agent explores and selects a relatively optimal decision which can be generated in the first-level strategy sources under different inputs from the decisions generated by the first-level strategy sources. According to the strategy selection method and selection device provided by the invention, higher-quality decisions and stronger generalization are realized.

Description

technical field [0001] The invention relates to the field of artificial intelligence, in particular to a method and device for selecting a knowledge strategy based on reinforcement learning. Background technique [0002] With the rapid development of deep learning technology in the field of artificial intelligence, deep reinforcement learning technology combining deep learning and reinforcement learning has become a new research hotspot. Traditional reinforcement learning methods cannot solve tasks with high-dimensional state and action space, while deep reinforcement learning utilizes the powerful perception and fitting capabilities of deep learning, and realizes intelligent control from original input to output through end-to-end learning , is considered to be an important path to general artificial intelligence. Deep reinforcement learning training mainly consists of two steps: collecting training data and updating decision-making strategies. Collecting training data re...

AI Technical Summary

Problems solved by technology

[0007] Aiming at the problems existing in the prior art, the purpose of the present invention is to solve the problem that the learning method in the prior art uses a single policy source to cause low decision quality, and cannot be applied to non-sequential decision-making problems and sequential decision-making problems at the same time; and it needs to be divided into Subtasks lead to low efficiency. The present invention proposes a knowledge policy selection method and selection device that can achieve higher quality decision-making and stronger generalization and can integrate multiple policy sources.

Images