# Learning to Play Two-Player Perfect-Information Games without Knowledge

arXiv: https://arxiv.org/abs/2008.01188 · HTML: https://mlsanity.com/paper/2008.01188/

## Key facts

- Cited by 11 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 3 Aug 2020.

## Built on by (most cited first, 11 of 11)

- [Deep Learning for General Game Playing with Ludii and Polygames](https://mlsanity.com/paper/2101.09562/) (2021)
- [Completeness of Unbounded Best-First Game Algorithms](https://mlsanity.com/paper/2109.09468/) (2021)
- [Spatial State-Action Features for General Games](https://mlsanity.com/paper/2201.06401/) (2022)
- [On some improvements to Unbounded Minimax](https://mlsanity.com/paper/2505.04525/) (2025)
- [Learning to Play Stochastic Two-player Perfect-Information Games without Knowledge](https://mlsanity.com/paper/2302.04318/) (2023)
- [Assessing Policy, Loss and Planning Combinations in Reinforcement Learning using a New Modular Architecture](https://mlsanity.com/paper/2201.02874/) (2022)
- [Deep Reinforcement Learning for 5*5 Multiplayer Go](https://mlsanity.com/paper/2405.14265/) (2024)
- [Minibal: Balanced Game-Playing Without Opponent Modeling](https://mlsanity.com/paper/2603.23059/) (2026)
- [Completeness of Unbounded Best-First Minimax and Descent Minimax](https://mlsanity.com/paper/2603.24572/) (2026)
- [Study and Improvement of Search Algorithms in Multi-Player Perfect-Information Games](https://mlsanity.com/paper/2604.17378/) (2026)
- [The Surprising Effectiveness of Approximate Value Iteration in Self-Play](https://mlsanity.com/paper/2609.09094/) (2026)

## Abstract (arXiv)

In this paper, several techniques for learning game state evaluation functions by reinforcement are proposed. The first is a generalization of tree bootstrapping (tree learning): it is adapted to the context of reinforcement learning without knowledge based on non-linear functions. With this technique, no information is lost during the reinforcement learning process. The second is a modification of minimax with unbounded depth extending the best sequences of actions to the terminal states. This modified search is intended to be used during the learning process. The third is to replace the classic gain of a game (+1 / -1) with a reinforcement heuristic. We study particular reinforcement heuristics such as: quick wins and slow defeats ; scoring ; mobility or presence. The four is a new action selection distribution. The conducted experiments suggest that these techniques improve the level of play. Finally, we apply these different techniques to design program-players to the game of Hex (size 11 and 13) surpassing the level of Mohex 3HNN with reinforcement learning from self-play without knowledge.