The Surprising Effectiveness of Approximate Value Iteration in Self-Play

• URL: http://arxiv.org/abs/2609.09094v1
• Date: Tue, 08 Sep 2026 17:37:19 GMT
• Title: The Surprising Effectiveness of Approximate Value Iteration in Self-Play
• Abstract summary: We train a minimal self-play implementation of Approximate Value Iteration (AVI)<n>Preliminary experiments on Othello and Go(9x9) show that AVI trains stably on larger games and learns effective value functions.
• Score: 0.18479558716666356
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Combining search with function approximation has driven major advances in game-playing programs, making self-play algorithms more competitive than ever. Still, the computational overhead of the most popular methods, based on Monte Carlo Tree Search (MCTS), can be substantial. In this work, we investigate whether simpler methods remain competitive in non-trivial, moderately sized games such as Connect Four, Hex(7x7) and synthetic games. We train a minimal self-play implementation of Approximate Value Iteration (AVI) and use ground-truth oracles for exact evaluation. Contrary to expectations, our results demonstrate the surprising effectiveness of AVI: it learns more accurate value functions than those learned by AlphaZero, while its one-step-lookahead greedy policies remain competitive with MCTS-based policies at substantially lower training and inference costs. Preliminary experiments on Othello and Go(9x9) show that AVI trains stably on larger games and learns effective value functions. These findings suggest that the success of MCTS-based methods may have eclipsed simpler approaches that have become increasingly practical with modern deep-learning tools.

Related papers

• Non-Adversarial Inverse Reinforcement Learning via Successor Feature Matching [23.600285251963395]
  In inverse reinforcement learning (IRL), an agent seeks to replicate expert demonstrations through interactions with the environment.<n>Traditionally, IRL is treated as an adversarial game, where an adversary searches over reward models, and a learner optimize the reward through repeated RL procedures.<n>We propose a novel approach to IRL by direct policy optimization, exploiting a linear factorization of the return as the inner product of successor features and a reward vector.
  arXiv  Detail & Related papers  (2024-11-11T14:05:50Z)
• Monte Carlo Tree Search Boosts Reasoning via Iterative Preference Learning [55.96599486604344]
  We introduce an approach aimed at enhancing the reasoning capabilities of Large Language Models (LLMs) through an iterative preference learning process. We use Monte Carlo Tree Search (MCTS) to iteratively collect preference data, utilizing its look-ahead ability to break down instance-level rewards into more granular step-level signals. The proposed algorithm employs Direct Preference Optimization (DPO) to update the LLM policy using this newly generated step-level preference data.
  arXiv  Detail & Related papers  (2024-05-01T11:10:24Z)
• Targeted Search Control in AlphaZero for Effective Policy Improvement [93.30151539224144]
  We introduce Go-Exploit, a novel search control strategy for AlphaZero. Go-Exploit samples the start state of its self-play trajectories from an archive of states of interest. Go-Exploit learns with a greater sample efficiency than standard AlphaZero.
  arXiv  Detail & Related papers  (2023-02-23T22:50:24Z)
• Spending Thinking Time Wisely: Accelerating MCTS with Virtual Expansions [89.89612827542972]
  This paper proposes a variant of Monte-Carlo tree search (MCTS) that spends more search time on harder states and less search time on simpler states adaptively. We evaluate the performance and computations on $9 times 9$ Go board games and Atari games. Experiments show that our method can achieve comparable performances to the original search algorithm while requiring less than $50%$ search time on average.
  arXiv  Detail & Related papers  (2022-10-23T06:39:20Z)
• Planning for Sample Efficient Imitation Learning [52.44953015011569]
  Current imitation algorithms struggle to achieve high performance and high in-environment sample efficiency simultaneously. We propose EfficientImitate, a planning-based imitation learning method that can achieve high in-environment sample efficiency and performance simultaneously. Experimental results show that EI achieves state-of-the-art results in performance and sample efficiency.
  arXiv  Detail & Related papers  (2022-10-18T05:19:26Z)
• A Ranking Game for Imitation Learning [22.028680861819215]
  We treat imitation as a two-player ranking-based Stackelberg game between a $textitpolicy$ and a $textitreward$ function. This game encompasses a large subset of both inverse reinforcement learning (IRL) methods and methods which learn from offline preferences. We theoretically analyze the requirements of the loss function used for ranking policy performances to facilitate near-optimal imitation learning at equilibrium.
  arXiv  Detail & Related papers  (2022-02-07T19:38:22Z)
• Combining Off and On-Policy Training in Model-Based Reinforcement Learning [77.34726150561087]
  We propose a way to obtain off-policy targets using data from simulated games in MuZero. Our results show that these targets speed up the training process and lead to faster convergence and higher rewards.
  arXiv  Detail & Related papers  (2021-02-24T10:47:26Z)
• Single-Agent Optimization Through Policy Iteration Using Monte-Carlo Tree Search [8.22379888383833]
  Combination of Monte-Carlo Tree Search (MCTS) and deep reinforcement learning is state-of-the-art in two-player perfect-information games. We describe a search algorithm that uses a variant of MCTS which we enhanced by 1) a novel action value normalization mechanism for games with potentially unbounded rewards, 2) defining a virtual loss function that enables effective search parallelization, and 3) a policy network, trained by generations of self-play, to guide the search.
  arXiv  Detail & Related papers  (2020-05-22T18:02:36Z)
• Provable Self-Play Algorithms for Competitive Reinforcement Learning [48.12602400021397]
  We study self-play in competitive reinforcement learning under the setting of Markov games. We show that a self-play algorithm achieves regret $tildemathcalO(sqrtT)$ after playing $T$ steps of the game. We also introduce an explore-then-exploit style algorithm, which achieves a slightly worse regret $tildemathcalO(T2/3)$, but is guaranteed to run in time even in the worst case.
  arXiv  Detail & Related papers  (2020-02-10T18:44:50Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.