Deep reinforcement learning method and device based on environment state prediction

Abstract

The invention discloses a deep reinforcement learning method and device based on environment state prediction. The method comprises the following steps that: establishing a deep reinforcement learningnetwork based on the environment prediction, and selecting a proper strategy decision method according to the characteristics of tasks; initializing network parameters, and establishing a storage area which meets a storage condition as an experience replaying area; according to the output of a strategy decision network, selecting a proper strategy to interact with environment, and continuously storing the interaction information of an interaction process into the experience replaying area; sampling a first sample sequence from the experience replaying area, utilizing a supervised learning method to train an environment prediction part, and repeating a first preset frequency; sampling a second sample sequence from the experience replaying area, fixing the parameter of the environment prediction part to be constant, utilizing a reinforcement learning method to train the strategy decision part, and repeating a second preset frequency; when network convergence meets a preset condition, obtaining a reinforcement learning network. By use of the method, learning efficiency can be effectively improved.

Description

technical field [0001] The invention relates to the technical field of artificial intelligence, in particular to a deep reinforcement learning method and device based on environmental state prediction. Background technique [0002] Learning in the process of interacting with the environment is a research hotspot in the field of artificial intelligence, and an important mathematical and theoretical tool to solve such problems is reinforcement learning. By solving the Markov decision process, reinforcement learning can learn a goal-oriented behavior strategy through the interaction of the unknown environment. Moreover, since reinforcement learning does not require explicit supervisory signals, and its learning process is similar to the learning behavior of animals, reinforcement learning is also considered to be a trend in solving general artificial intelligence problems. [0003] Reinforcement learning algorithms in related technologies are limited by time, space, and sample...

AI Technical Summary

Problems solved by technology

On the one hand, the basis of reinforcement learning algorithms is Markov, and a necessary condition to satisfy Markov is that the state is required to be fully observed. Therefore, for partially observable Markov decision problems, current algorithms usually use A good strategy cannot be obtained; on the other hand, most of the current reinforcement learning algorithms can only solve one task in the environment. When the task is switched, the network often needs to be retrained, which does not have good versatility and Migration needs to be solved urgently

Images