Learning in zero-sum games studies a situation where multiple agents competitively learn their strategy. In such multi-agent learning, we often see that the strategies cycle around their optimum, i.e., Nash equilibrium. When a game periodically varies (called a ``periodic'' game), however, the Nash equilibrium moves generically. How learning dynamics behave in such periodic games is of interest but still unclear. Interestingly, we discover that the behavior is highly dependent on the relationship between the two speeds at which the game changes and at which players learn. We observe that when these two speeds synchronize, the learning dynamics diverge, and their time-average does not converge. Otherwise, the learning dynamics draw complicated cycles, but their time-average converges. Under some assumptions introduced for the dynamical systems analysis, we prove that this behavior occurs. Furthermore, our experiments observe this behavior even if these assumptions are removed. This study discovers a novel phenomenon, i.e., synchronization, and gains insight widely applicable to learning in periodic games.

This paper introduces state abstraction for two-player zero-sum Markov games (TZMGs) where the payoffs for the two players are determined by the state representing the environment and their respective actions, with state transitions following a Markov decision processes. For example, in games like soccer, the value of actions changes according to the state of play, we should describe them as Markov games. In TZMGs, the more the number of states becomes, the more difficult computing the equilibrium becomes. Therefore, we abstract the states of TZMGs and examine the performance. State abstraction reduces the number of states by treating multiple different states as a single state, and there is a substantial body of research on finding optimal policies for Markov decision processes using state abstraction. This study extends the state abstraction for MDPs to Markov games. In this case, the game with state abstraction may yield different equilibrium solutions from those of the ground game. To evaluate the equilibrium solutions of the game with state abstraction, we derived bounds on duality gap, which represents the distance from the equilibrium solutions of the ground game. Finally, we demonstrate our state abstraction with Markov Soccer, compute equilibrium policies, and examine the results.

This paper presents a payoff perturbation technique, introducing a strong convexity to players' payoff functions in games. This technique is specifically designed for first-order methods to achieve last-iterate convergence in games where the gradient of the payoff functions is monotone in the strategy profile space, potentially containing additive noise. Although perturbation is known to facilitate the convergence of learning algorithms, the magnitude of perturbation requires careful adjustment to ensure last-iterate convergence. Previous studies have proposed a scheme in which the magnitude is determined by the distance from a periodically re-initialized anchoring or reference strategy. Building upon this, we propose Gradient Ascent with Boosting Payoff Perturbation, which incorporates a novel perturbation into the underlying payoff function, maintaining the periodically re-initializing anchoring strategy scheme. This innovation empowers us to provide faster last-iterate convergence rates against the existing payoff perturbed algorithms, even in the presence of additive noise.

Yuu Jinnai, Tetsuro Morimura, Kaito Ariu, Kenshi Abe. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

This study raises and addresses the problem of time-delayed feedback in learning in games. Because learning in games assumes that multiple agents independently learn their strategies, a discrepancy in optimization often emerges among the agents. To overcome this discrepancy, the prediction of the future reward is incorporated into algorithms, typically known as Optimistic Follow-the-Regularized-Leader (OFTRL). However, the time delay in observing the past rewards hinders the prediction. Indeed, this study firstly proves that even a single-step delay worsens the performance of OFTRL from the aspects of social regret and convergence. This study proposes the weighted OFTRL (WOFTRL), where the prediction vector of the next reward in OFTRL is weighted $n$ times. We further capture an intuition that the optimistic weight cancels out this time delay. We prove that when the optimistic weight exceeds the time delay, our WOFTRL recovers the good performances that social regret is constant in general-sum normal-form games, and the strategies last-iterate converge to the Nash equilibrium in poly-matrix zero-sum games. The theoretical results are supported and strengthened by our experiments.

In the Lasry--Lions framework, Mean-Field Games (MFGs) model interactions among an infinite number of agents. However, existing algorithms either require strict monotonicity or only guarantee the convergence of averaged iterates, as in Fictitious Play in continuous time. We address this gap with the following theoretical result. First, we prove that the last-iterated policy of a proximal-point (PP) update with KL regularization converges to an MFG equilibrium under non-strict monotonicity. Second, we see that each PP update is equivalent to finding the equilibria of a KL-regularized MFG. We then prove that this equilibrium can be found using mirror descent with an exponential last-iterate convergence rate. Building on these insights, we propose the Approximate Proximal-Point ($\mathtt{APP}$) algorithm, which approximately implements the PP update via a small number of Mirror Descent steps. Numerical experiments on standard benchmarks confirm that the $\mathtt{APP}$ algorithm reliably converges to the unregularized mean-field equilibrium without time-averaging.

In online advertising, identifying the optimal creative is critical to maximizing performance. This study examines the application of top-two Thompson sampling (TTTS), an adaptive experimental design method, as an efficient alternative to traditional A/B testing for identifying the optimal creative. Our experiments on an online advertising platform highlight the effectiveness of TTTS in both accurately identifying the optimal creative and minimizing experimental costs, underscoring its potential as a promising approach to creative selection.

Learning in games considers how multiple agents maximize their own rewards through repeated games. Memory, an ability that an agent changes his/her action depending on the history of actions in previous games, is often introduced into learning to explore more clever strategies and discuss the decision-making of real agents like humans. However, such games with memory are hard to analyze because they exhibit complex phenomena like chaotic dynamics or divergence from Nash equilibrium. In particular, how asymmetry in memory capacities between agents affects learning in games is still unclear. In response, this study formulates a gradient ascent algorithm in games with asymmetry memory capacities. To obtain theoretical insights into learning dynamics, we first consider a simple case of zero-sum games. We observe complex behavior, where learning dynamics draw a heteroclinic connection from unstable fixed points to stable ones. Despite this complexity, we analyze learning dynamics and prove local convergence to these stable fixed points, i.e., the Nash equilibria. We identify the mechanism driving this convergence: an agent with a longer memory learns to exploit the other, which in turn endows the other's utility function with strict concavity. We further numerically observe such convergence in various initial strategies, action numbers, and memory lengths. This study reveals a novel phenomenon due to memory asymmetry, providing fundamental strides in learning in games and new insights into computing equilibria.

This work addresses learning online fair division under uncertainty, where a central planner sequentially allocates items without precise knowledge of agents’ values or utilities. Departing from conventional online algorithms, the planner here relies on noisy, estimated values obtained after allocating items. We introduce wrapper algorithms utilizing dual averaging, enabling gradual learning of both the type distribution of arriving items and agents’ values through bandit feedback. This approach enables the algorithms to asymptotically achieve optimal Nash social welfare in linear Fisher markets with agents having additive utilities. We also empirically verify the performance of the proposed algorithms across synthetic and empirical datasets.

Reinforcement learning from human feedback (RLHF) plays a crucial role in aligning language models with human preferences.While the significance of dataset quality is generally recognized, explicit investigations into its impact within the RLHF framework, to our knowledge, have been limited.This paper addresses the issue of text quality within the preference dataset by focusing on direct preference optimization (DPO), an increasingly adopted reward-model-free RLHF method.We confirm that text quality significantly influences the performance of models optimized with DPO more than those optimized with reward-modelbased RLHF.Building on this new insight, we propose an extension of DPO, termed filtered direct preference optimization (fDPO).fDPO uses a trained reward model to monitor the quality of texts within the preference dataset during DPO training.Samples of lower quality are discarded based on comparisons with texts generated by the model being optimized, resulting in a more accurate dataset.Experimental results demonstrate that fDPO enhances the final model performance.Our code is available at https://github.com/CyberAgentAILab/ filtered-dpo.

This paper proposes a payoff perturbation technique for the Mirror Descent (MD) algorithm in games where the gradient of the payoff functions is monotone in the strategy profile space, potentially containing additive noise. The optimistic family of learning algorithms, exemplified by optimistic MD, successfully achieves *last-iterate* convergence in scenarios devoid of noise, leading the dynamics to a Nash equilibrium. A recent re-emerging trend underscores the promise of the perturbation approach, where payoff functions are perturbed based on the distance from an anchoring, or *slingshot*, strategy. In response, we propose *Adaptively Perturbed MD* (APMD), which adjusts the magnitude of the perturbation by repeatedly updating the slingshot strategy at a predefined interval. This innovation empowers us to find a Nash equilibrium of the underlying game with guaranteed rates. Empirical demonstrations affirm that our algorithm exhibits significantly accelerated convergence.

Minimum Bayes Risk (MBR) decoding has been shown to be a powerful alternative to beam search decoding in a variety of text generation tasks. MBR decoding selects a hypothesis from a pool of hypotheses that has the least expected risk under a probability model according to a given utility function. Since it is impractical to compute the expected risk exactly over all possible hypotheses, two approximations are commonly used in MBR. First, it integrates over a sampled set of hypotheses rather than over all possible hypotheses. Second, it estimates the probability of each hypothesis using a Monte Carlo estimator. While the first approximation is necessary to make it computationally feasible, the second is not essential since we typically have access to the model probability at inference time. We propose model-based MBR (MBMBR), a variant of MBR that uses the model probability itself as the estimate of the probability distribution instead of the Monte Carlo estimate. We show analytically and empirically that the model-based estimate is more promising than the Monte Carlo estimate in text generation tasks. Our experiments show that MBMBR outperforms MBR in several text generation tasks, both with encoder-decoder models and with language models.

Typical recommendation and ranking methods aim to optimize the satisfaction of users, but they are often oblivious to their impact on the items (e.g., products, jobs, news, video) and their providers. However, there has been a growing understanding that the latter is crucial to consider for a wide range of applications, since it determines the utility of those being recommended. Prior approaches to fairness-aware recommendation optimize a regularized objective to balance user satisfaction and item fairness based on some notion such as exposure fairness. These existing methods have been shown to be effective in controlling fairness, however, most of them are computationally inefficient, limiting their applications to only unrealistically small-scale situations. This indeed implies that the literature does not yet provide a solution to enable a flexible control of exposure in the industry-scale recommender systems where millions of users and items exist. To enable a computationally efficient exposure control even for such large-scale systems, this work develops a scalable, fast, and fair method called exposure-aware ADMM (exADMM ). exADMM is based on implicit alternating least squares (iALS), a conventional scalable algorithm for collaborative filtering, but optimizes a regularized objective to achieve a flexible control of accuracy-fairness tradeoff. A particular technical challenge in developing exADMM is the fact that the fairness regularizer destroys the separability of optimization subproblems for users and items, which is an essential property to ensure the scalability of iALS. Therefore, we develop a set of optimization tools to enable yet scalable fairness control with provable convergence guarantees as a basis of our algorithm. Extensive experiments performed on three recommendation datasets demonstrate that exADMM enables a far more flexible fairness control than the vanilla version of iALS, while being much more computationally efficient than existing fairness-aware recommendation methods.

This paper proposes Mutation-Driven Multiplicative Weights Update (M2WU) for learning an equilibrium in two-player zero-sum normal-form games and proves that it exhibits the last-iterate convergence property in both full and noisy feedback settings. In the former, players observe their exact gradient vectors of the utility functions. In the latter, they only observe the noisy gradient vectors. Even the celebrated Multiplicative Weights Update (MWU) and Optimistic MWU (OMWU) algorithms may not converge to a Nash equilibrium with noisy feedback. On the contrary, M2WU exhibits the last-iterate convergence to a stationary point near a Nash equilibrium in both feedback settings. We then prove that it converges to an exact Nash equilibrium by iteratively adapting the mutation term. We empirically confirm that M2WU outperforms MWU and OMWU in exploitability and convergence rates.

Repeated games consider a situation where multiple agents are motivated by their independent rewards throughout learning. In general, the dynamics of their learning become complex. Especially when their rewards compete with each other like zero-sum games, the dynamics often do not converge to their optimum, i.e., the Nash equilibrium. To tackle such complexity, many studies have understood various learning algorithms as dynamical systems and discovered qualitative insights among the algorithms. However, such studies have yet to handle multi-memory games (where agents can memorize actions they played in the past and choose their actions based on their memories), even though memorization plays a pivotal role in artificial intelligence and interpersonal relationship. This study extends two major learning algorithms in games, i.e., replicator dynamics and gradient ascent, into multi-memory games. Then, we prove their dynamics are identical. Furthermore, theoretically and experimentally, we clarify that the learning dynamics diverge from the Nash equilibrium in multi-memory zero-sum games and reach heteroclinic cycles (sojourn longer around the boundary of the strategy space), providing a fundamental advance in learning in games.

Bandit algorithms for online learning to rank (OLTR) problems often aim to maximize long-term revenue by utilizing user feedback. From a practical point of view, however, such algorithms have a high risk of hurting user experience due to their aggressive exploration. Thus, there has been a rising demand for safe exploration in recent years. One approach to safe exploration is to gradually enhance the quality of an original ranking that is already guaranteed acceptable quality. In this paper, we propose a safe OLTR algorithm that efficiently exchanges one of the items in the current ranking with an item outside the ranking (i.e., an unranked item) to perform exploration. We select an unranked item optimistically to explore based on Kullback-Leibler upper confidence bounds (KL-UCB) and safely re-rank the items including the selected one. Through experiments, we demonstrate that the proposed algorithm improves long-term regret from baselines without any safety violation.

In this paper, we revisit the regret minimization problem in sparse stochastic contextual linear bandits, where feature vectors may be of large dimension $d$, but where the reward function depends on a few, say $s_0\ll d$, of these features only. We present Thresholded Lasso bandit, an algorithm that (i) estimates the vector defining the reward function as well as its sparse support, i.e., significant feature elements, using the Lasso framework with thresholding, and (ii) selects an arm greedily according to this estimate projected on its support. The algorithm does not require prior knowledge of the sparsity index $s_0$ and can be parameter-free under some symmetric assumptions. For this simple algorithm, we establish non-asymptotic regret upper bounds scaling as $\mathcal{O}( \log d + \sqrt{T} )$ in general, and as $\mathcal{O}( \log d + \log T)$ under the so-called margin condition (a probabilistic condition on the separation of the arm rewards). The regret of previous algorithms scales as $\mathcal{O}( \log d + \sqrt{T \log (d T)})$ and $\mathcal{O}( \log T \log d)$ in the two settings, respectively. Through numerical experiments, we confirm that our algorithm outperforms existing methods.

This paper considers the capacity expansion problem in two-sided matchings, where the policymaker is allowed to allocate some extra seats as well as the standard seats. In medical residency match, each hospital accepts a limited number of doctors. Such capacity constraints are typically given in advance. However, such exogenous constraints can compromise the welfare of the doctors; some popular hospitals inevitably dismiss some of their favorite doctors. Meanwhile, it is often the case that the hospitals are also benefited to accept a few extra doctors. To tackle the problem, we propose an anytime method that the upper confidence tree searches the space of capacity expansions, each of which has a resident-optimal stable assignment that the deferred acceptance method finds. Constructing a good search tree representation significantly boosts the performance of the proposed method. Our simulation shows that the proposed method identifies an almost optimal capacity expansion with a significantly smaller computational budget than exact methods based on mixed-integer programming.

In this study, we consider a variant of the Follow the Regularized Leader (FTRL) dynamics in two-player zero-sum games. FTRL is guaranteed to converge to a Nash equilibrium when time-averaging the strategies, while a lot of variants suffer from the issue of limit cycling behavior, i.e., lack the last-iterate convergence guarantee. To this end, we propose mutant FTRL (M-FTRL), an algorithm that introduces mutation for the perturbation of action probabilities. We then investigate the continuous-time dynamics of M-FTRL and provide the strong convergence guarantees toward stationary points that approximate Nash equilibria under full-information feedback. Furthermore, our simulation demonstrates that M-FTRL can enjoy faster convergence rates than FTRL and optimistic FTRL under full-information feedback and surprisingly exhibits clear convergence under bandit feedback.