Supervised Learning Achieves Human-Level Performance in MOBA Games: A Case Study of Honor of Kings

TL;DR

This paper introduces JueWu-SL, a supervised learning AI that achieves human-level performance in MOBA games by integrating macro and micro strategies into neural networks, demonstrated on Honor of Kings.

Contribution

It presents the first supervised learning approach that combines macro and micro gameplay strategies end-to-end for MOBA games.

Findings

AI performs at High King level in Honor of Kings

Supervised learning effectively captures complex MOBA gameplay

End-to-end neural network integration is successful

Abstract

We present JueWu-SL, the first supervised-learning-based artificial intelligence (AI) program that achieves human-level performance in playing multiplayer online battle arena (MOBA) games. Unlike prior attempts, we integrate the macro-strategy and the micromanagement of MOBA-game-playing into neural networks in a supervised and end-to-end manner. Tested on Honor of Kings, the most popular MOBA at present, our AI performs competitively at the level of High King players in standard 5v5 games.