NeurIPS 2020

Towards Playing Full MOBA Games with Deep Reinforcement Learning

Meta Review

This paper demonstrates an application of RL and search to a challenging MOBA game-playing task, leading to AI agents able to defeat top professional human players. Three out of four reviewers consider that although this is an application-oriented paper with a strong engineering focus, it is still relevant enough for publication at NeurIPS. Only R2 is advocating for rejection, based essentially on the lack of scientific novelty. I believe that such impressive large scale applications of RL are well worth pushing forward and I am thus recommending acceptance. The general algorithms being used may not be novel, but their instantiation to solve this specific task largely is. In addition, although the exact techniques developed here may be specific to MOBA games, I expect them to potentially inspire further applications both within and outside of video games (I do not believe, like R2 suggested, that there is a need for a second application in this submission, since at such scale each application domain is worth a paper on its own IMHO). The authors’ feedback helped clarify many points raised by the reviewers, and the authors committed to improve the paper accordingly (please do so! This will be very helpful to future readers of your paper)