arXiv subjects

Yunfeng Luo

Publications and source records attributed to Yunfeng Luo.

5 recordsLinked to original sources

arXiv2024

Preference-CFR$\:$ Beyond Nash Equilibrium for Better Game Strategies

Artificial intelligence (AI) has surpassed top human players in a variety of games. In imperfect information games, these achievements have primarily been driven by Counterfactual Regret Minimization (CFR) and its variants for computing Nash equilibrium. However, most existing research has focused on maximizing payoff, while largely neglecting the importance of strategic diversity and the need for varied play styles, thereby limiting AI's adaptability to different user preferences. To address this gap, we propose Preference-CFR (Pref-CFR), a novel method that incorporates two key parameters: preference degree and vulnerability degree. These parameters enable the AI to adjust its strategic distribution within an acceptable performance loss threshold, thereby enhancing its adaptability to a wider range of strategic demands. In our experiments with Texas Hold'em, Pref-CFR successfully trained Aggressive and Loose Passive styles that not only match original CFR-based strategies in performance but also display clearly distinct behavioral patterns. Notably, for certain hand scenarios, Pref-CFR produces strategies that diverge significantly from both conventional expert heuristics and original CFR outputs, potentially offering novel insights for professional players.

cs.GT↗

arXiv2024

From First-Order to Second-Order Rationality: Advancing Game Convergence with Dynamic Weighted Fictitious Play

Constructing effective algorithms to converge to Nash Equilibrium (NE) is an important problem in algorithmic game theory. Prior research generally posits that the upper bound on the convergence rate for games is $O\left(T^{-1/2}\right)$. This paper introduces a novel perspective, positing that the key to accelerating convergence in game theory is rationality. Based on this concept, we propose a Dynamic Weighted Fictitious Play (DW-FP) algorithm. We demonstrate that this algorithm can converge to a NE and exhibits a convergence rate of $O(T^{-1})$ in experimental evaluations.

cs.GT↗

arXiv2024

ELO-Rated Sequence Rewards: Advancing Reinforcement Learning Models

Reinforcement Learning (RL) heavily relies on the careful design of the reward function. However, accurately assigning rewards to each state-action pair in Long-Term Reinforcement Learning (LTRL) tasks remains a significant challenge. As a result, RL agents are often trained under expert guidance. Inspired by the ordinal utility theory in economics, we propose a novel reward estimation algorithm: ELO-Rating based Reinforcement Learning (ERRL). This approach features two key contributions. First, it uses expert preferences over trajectories rather than cardinal rewards (utilities) to compute the ELO rating of each trajectory as its reward. Second, a new reward redistribution algorithm is introduced to alleviate training instability in the absence of a fixed anchor reward. In long-term scenarios (up to 5000 steps), where traditional RL algorithms struggle, our method outperforms several state-of-the-art baselines. Additionally, we conduct a comprehensive analysis of how expert preferences influence the results.

cs.LG↗

arXiv2024

Beyond Nash Equilibrium: Achieving Bayesian Perfect Equilibrium with Belief Update Fictitious Play

In the domain of machine learning and game theory, the quest for Nash Equilibrium (NE) in extensive-form games with incomplete information is challenging yet crucial for enhancing AI's decision-making support under varied scenarios. Traditional Counterfactual Regret Minimization (CFR) techniques excel in navigating towards NE, focusing on scenarios where opponents deploy optimal strategies. However, the essence of machine learning in strategic game play extends beyond reacting to optimal moves; it encompasses aiding human decision-making in all circumstances. This includes not only crafting responses to optimal strategies but also recovering from suboptimal decisions and capitalizing on opponents' errors. Herein lies the significance of transitioning from NE to Bayesian Perfect Equilibrium (BPE), which accounts for every possible condition, including the irrationality of opponents. To bridge this gap, we propose Belief Update Fictitious Play (BUFP), which innovatively blends fictitious play with belief to target BPE, a more comprehensive equilibrium concept than NE. Specifically, through adjusting iteration stepsizes, BUFP allows for strategic convergence to both NE and BPE. For instance, in our experiments, BUFP(EF) leverages the stepsize of Extensive Form Fictitious Play (EFFP) to achieve BPE, outperforming traditional CFR by securing a 48.53\% increase in benefits in scenarios characterized by dominated strategies.

cs.GT↗

arXiv2023

Accelerating Nash Equilibrium Convergence in Monte Carlo Settings Through Counterfactual Value Based Fictitious Play

Counterfactual Regret Minimization (CFR) and its variants are widely recognized as effective algorithms for solving extensive-form imperfect information games. Recently, many improvements have been focused on enhancing the convergence speed of the CFR algorithm. However, most of these variants are not applicable under Monte Carlo (MC) conditions, making them unsuitable for training in large-scale games. We introduce a new MC-based algorithm for solving extensive-form imperfect information games, called MCCFVFP (Monte Carlo Counterfactual Value-Based Fictitious Play). MCCFVFP combines CFR's counterfactual value calculations with fictitious play's best response strategy, leveraging the strengths of fictitious play to gain significant advantages in games with a high proportion of dominated strategies. Experimental results show that MCCFVFP achieved convergence speeds approximately 20\%$\sim$50\% faster than the most advanced MCCFR variants in games like poker and other test games.

cs.AI↗