Neural network model training method and device, equipment and storage medium

Abstract

The invention discloses a neural network model training method realized based on an artificial intelligence technology. The method comprises the following steps: acquiring first global information and first unit information at a first moment; on the basis of the first global information and the first unit information, action distribution of each agent at the first moment is obtained through a behavior prediction network; acquiring second global information and second unit information according to the first action distribution; determining reward information according to the first global information, the first unit information, the second global information and the second unit information; based on the global information and the unit information, obtaining a target own value and a target team value through a value network; and training the network according to the action distribution, the reward information, the target own value and the target team value. The invention further provides a device. The method encourages the agents under the same camp to execute the action capable of maximizing the own value and the team value, so that the combined action of the team is better, and a higher agent simulation level is achieved.

Description

technical field [0001] The present application relates to the technical field of artificial intelligence, and in particular to a training method, device, equipment and storage medium of a neural network model. Background technique [0002] With the development of Internet technology, there are more and more types of online games, and Multiplayer Online Battle Arena (MOBA) is one of them. In this type of game, players are usually divided into two or more opposing camps and compete with each other on a scattered game map, and each player controls a selected game character through the interface to fight against the opponent. [0003] The game character can not only be controlled by the player, but also can be controlled by an artificial intelligence (AI) model for battle. At present, the reinforcement learning algorithm is mainly used in the existing technology, so that the agent can learn strategies to maximize rewards or achieve specific goals in the process of interacting w...

AI Technical Summary

Problems solved by technology

Therefore, the model optimization method will cause the values ​​​​of the agent to deviate, and lack of team awareness, resulting in a low simulation level of the agent

Images