Target task processing method, device and equipment based on reinforcement learning migration

Abstract

The invention provides a target task processing method, device and equipment based on reinforcement learning migration. The method comprises the following steps that a sub-target set in a preset source task database is obtained, source tasks are learned tasks, and the sub-target set comprises sub-target tasks corresponding to target tasks in the source tasks; extracting trajectory features of the sub-target tasks based on the sub-target set, and constructing a candidate item set; and screening a reuse candidate item from the candidate item set, and performing reinforcement learning on the target task based on the reuse candidate item to obtain a reinforcement learning result of the target task. The candidate item set of the learned task is migrated to the target task, so that the convergence speed of the target task in the reinforcement learning process is accelerated, and the task learning speed is greatly improved.

Description

technical field [0001] The present invention relates to the technical field of machine learning, in particular to a target task processing method, device and equipment based on reinforcement learning transfer. Background technique [0002] Reinforcement Learning (RL), also known as Reinforcement Learning, Evaluation Learning or Enhanced Learning, is one of the paradigms and methodologies of machine learning, which is used to describe and solve the problem of learning through the process of interaction between agents and the environment. Strategies to maximize rewards or achieve specific goals are an important class of machine learning techniques to solve sequential decision-making problems. After decades of development, they have been successfully applied to many fields such as automatic control, robotics, recommendation and retrieval. A reinforcement learning task can be expressed as a Markov decision process (MDP). However, in the process of reinforcement learning in the ...

AI Technical Summary

Problems solved by technology

[0003] Therefore, the technical problem to be solved by the present invention is to overcome the defects in the prior art that the convergence speed is slow and the learning speed cannot meet the growing demand, thereby providing a target task processing method, device and equipment based on reinforcement learning migration

Images