A Data-Efficient Reinforcement Learning Agent for Automated Playtesting in Procedurally Generated Environments
收藏数据链接：
官方服务：
资源简介：
The thesis focuses on developing a reinforcement learning algorithm to create model that automates game playtests. The main goal of the model is to fully explore a randomly generated maze. The model is also expected to be able to continue exploration in the presence of a bug in the environment. The algorithm also needs less extensive training and has early convergence, while maintaining a high success rate in whole map exploration.
创建时间：
2026-08-02
