Pure exploration is one of the fundamental problems in multi-armed bandits (MAB). However, existing works mostly focus on specific pure exploration tasks, without a holistic view of the general pure exploration problem. This work fills this gap by introducing a versatile framework to study pure exploration, with a focus on identifying the pairwise relationships between targeted arm pairs. Moreover, unlike existing works that only optimize the stopping time (i.e., sample complexity), this work considers that arms are associated with potentially different costs and targets at optimizing the cumulative cost that occurred during learning. Under the general framework of pairwise pure exploration with arm-specific costs, a performance lower bound is derived. Then, a novel algorithm, termed CAET (Cost-Aware Pairwise Exploration Task), is proposed. CAET builds on the track-and-stop principle with a novel design to handle the arm-specific costs, which can potentially be zero and thus represent a very challenging case. Theoretical analyses prove that the performance of CAET approaches the lower bound asymptotically. Special cases are further discussed, including an extension to regret minimization, which is another major focus of MAB. The effectiveness and efficiency of CAET are also verified through experimental results under various settings.

Recent studies have shown that large language models' (LLMs) mathematical problem-solving capabilities can be enhanced by integrating external tools, such as code interpreters, and employing multi-turn Chain-of-Thought (CoT) reasoning. While current methods focus on synthetic data generation and Supervised Fine-Tuning (SFT), this paper studies the complementary direct preference learning approach to further improve model performance. However, existing direct preference learning algorithms are originally designed for the single-turn chat task, and do not fully address the complexities of multi-turn reasoning and external tool integration required for tool-integrated mathematical reasoning tasks. To fill in this gap, we introduce a multi-turn direct preference learning framework, tailored for this context, that leverages feedback from code interpreters and optimizes trajectory-level preferences. This framework includes multi-turn DPO and multi-turn KTO as specific implementations. The effectiveness of our framework is validated through training of various language models using an augmented prompt set from the GSM8K and MATH datasets. Our results demonstrate substantial improvements: a supervised fine-tuned Gemma-1.1-it-7B model's performance increased from 77.5% to 83.9% on GSM8K and from 46.1% to 51.2% on MATH. Similarly, a Gemma-2-it-9B model improved from 84.1% to 86.3% on GSM8K and from 51.0% to 54.5% on MATH.

Reinforcement Learning from Human Feedback (RLHF) has emerged as a key technique for post‑training large language models. Despite its empirical success, the theoretical understanding of RLHF is still limited, as learning the KL-regularized target with only preference feedback poses additional challenges compared with canonical RL. Existing works mostly study the reward-based Bradley-Terry (BT) preference model, and extend classical designs utilizing optimism or pessimism. This work, instead, considers the general preference model (whose practical relevance has been observed recently) and obtains performance guarantees with major, order-wise improvements over existing ones. Surprisingly, these results are derived from algorithms that directly use empirical estimates (i.e., greedy sampling), as opposed to constructing optimistic or pessimistic estimates in previous works. This insight has a deep root in the unique structural property of the optimal policy class under the KL-regularized target, and we further specialize it to the BT model, highlighting the surprising sufficiency of greedy sampling in RLHF.

This paper investigates a hybrid learning framework for reinforcement learning (RL) in which the agent can leverage both an offline dataset and online interactions to learn the optimal policy. We present a unified algorithm and analysis and show that augmenting confidence-based online RL algorithms with the offline dataset outperforms any pure online or offline algorithm alone and achieves state-of-the-art results under two learning metrics, i.e., sub-optimality gap and online learning regret. Specifically, we show that our algorithm achieves a sub-optimality gap $\tilde{O}( \sqrt{1/(N_0/ \mathtt{C}(\pi^\star| \rho)+N_1} ) )$, where $\mathtt{C}(\pi^\star|\rho)$ is a new concentrability coefficient, $N_0$ and $N_1$ are the numbers of offline and online samples, respectively. For regret minimization, we show that it achieves a constant $\tilde{O}( \sqrt{N_1/(N_0/\mathtt{C}(\pi^{-}|\rho)+N_1)} )$ speed-up compared to pure online learning, where $\mathtt{C}(\pi^-|\rho)$ is the concentrability coefficient over all sub-optimal policies. Our results also reveal an interesting separation on the desired coverage properties of the offline dataset for sub-optimality gap minimization and regret minimization. We further validate our theoretical findings in several experiments in special RL models such as linear contextual bandits and Markov decision processes (MDPs).

The remarkable instruction-following capability of large language models (LLMs) has sparked a growing interest in automatically finding good prompts, i.e., prompt optimization. Most existing works follow the scheme of selecting from a pre-generated pool of candidate prompts. However, these designs mainly focus on the generation strategy, while limited attention has been paid to the selection method. Especially, the cost incurred during the selection (e.g., accessing LLM and evaluating the responses) is rarely explicitly considered. To overcome this limitation, this work provides a principled framework, TRIPLE, to efficiently perform prompt selection under an explicit budget constraint. TRIPLE is built on a novel connection established between prompt optimization and fixed-budget best arm identification (BAI-FB) in multi-armed bandits (MAB); thus, it is capable of leveraging the rich toolbox from BAI-FB systematically and also incorporating unique characteristics of prompt optimization. Extensive experiments on multiple well-adopted tasks using various LLMs demonstrate the remarkable performance improvement of TRIPLE over baselines while satisfying the limited budget constraints. As an extension, variants of TRIPLE are proposed to efficiently select examples for few-shot prompts, also achieving superior empirical performance.

The in-context learning (ICL) capability of pre-trained models based on the transformer architecture has received growing interest in recent years. While theoretical understanding has been obtained for ICL in reinforcement learning (RL), the previous results are largely confined to the single-agent setting. This work proposes to further explore the in-context learning capabilities of pre-trained transformer models in competitive multi-agent games, i.e., in-context game-playing (ICGP). Focusing on the classical two-player zero-sum games, theoretical guarantees are provided to demonstrate that pre-trained transformers can provably learn to approximate Nash equilibrium in an in-context manner for both decentralized and centralized learning settings. As a key part of the proof, constructional results are established to demonstrate that the transformer architecture is sufficiently rich to realize celebrated multi-agent game-playing algorithms, in particular, decentralized V-learning and centralized VI-ULCB.

In-Context Learning (ICL) empowers Large Language Models (LLMs) to tackle various tasks by providing input-output examples as additional inputs, referred to as demonstrations. Nevertheless, the performance of ICL could be easily impacted by the quality of selected demonstrations. Existing efforts generally learn a retriever model to score each demonstration for selecting suitable demonstrations, however, the effect is suboptimal due to the large search space and the noise from unhelpful demonstrations. In this study, we introduce MoD, which partitions the demonstration pool into groups, each governed by an expert to reduce search space. We further design an expert-wise training strategy to alleviate the impact of unhelpful demonstrations when optimizing the retriever model. During inference, experts collaboratively retrieve demonstrations for the input query to enhance the ICL performance. We validate MoD via experiments across a range of NLP datasets and tasks, demonstrating its state-of-the-art performance and shedding new light on the future design of retrieval methods for ICL.

The recent development of reinforcement learning (RL) has boosted the adoption of online RL for wireless radio resource management (RRM). However, online RL algorithms require direct interactions with the environment, which may be undesirable given the potential performance loss due to the unavoidable exploration in RL. In this work, we first explore the use ofofflineRL algorithms in solving the RRM problem. We evaluate several state-of-the-art offline RL algorithms for a practical RRM problem that aims at maximizing a linear combination of total rates and 5-percentile rates via user scheduling. Our findings indicate that the performance of offline RL for the RRM problem is heavily contingent upon the behavior policy deployed for data collection. We propose an innovative offline RL approach utilizing heterogeneous datasets from various behavior policies. This method demonstrates that a strategic mixture of datasets enables near-optimal RL policy generation, even with suboptimal behavior policies. Additionally, we introduce two enhancements: an ensemble-based policy to augment dataset mixture training efficiency, and a novel offline-to-online strategy for seamless adaptation to new environments. Our data mixture approach achieves over 95% efficiency of an online RL agent in the absence of expert data. The ensemble algorithm notably reduces training duration by half compared to the data mixture method. Furthermore, our model, when applied with offline-to-online fine-tuning, surpasses existing benchmarks by approximately 5% in our user scheduling problem.

Existing theoretical studies on offline reinforcement learning (RL) mostly consider a dataset sampled directly from the target task. In practice, however, data often come from several heterogeneous but related sources. Motivated by this gap, this work aims at rigorously understanding offline RL with multiple datasets that are collected from randomly perturbed versions of the target task instead of from itself. An information-theoretic lower bound is derived, which reveals a necessary requirement on the number of involved sources in addition to that on the number of data samples. Then, a novel HetPEVI algorithm is proposed, which simultaneously considers the sample uncertainties from a finite number of data samples per data source and the source uncertainties due to a finite number of available data sources. Theoretical analyses demonstrate that HetPEVI can solve the target task as long as the data sources collectively provide a good data coverage. Moreover, HetPEVI is demonstrated to be optimal up to a polynomial factor of the horizon length. Finally, the study is extended to offline Markov games and offline robust RL, which demonstrates the generality of the proposed designs and theoretical analyses.

Most existing studies on linear bandits focus on a one-dimensional characterization of the overall system. While being representative, this formulation may fail to model applications with high-dimensional but favorable structures, such as the low-rank tensor representation for recommender systems. To address this limitation, this work studies a general tensor bandits model, where actions and system parameters are represented by tensors as opposed to vectors, and we particularly focus on the case that the unknown system tensor is low-rank. A novel bandit algorithm, coined TOFU (Tensor Optimism in the Face of Uncertainty), is developed. TOFU first leverages flexible tensor regression techniques to estimate low-dimensional subspaces associated with the system tensor. These estimates are then utilized to convert the original problem to a new one with norm constraints on its system parameters. Lastly, a norm-constrained bandit subroutine is adopted by TOFU, which utilizes these constraints to avoid exploring the entire high-dimensional parameter space. Theoretical analyses show that TOFU improves the best-known regret upper bound by a multiplicative factor that grows exponentially in the system order. A novel performance lower bound is also established, which further corroborates the efficiency of TOFU.

Most existing federated multi-armed bandits (FMAB) designs are based on the presumption that clients will implement the new design to collaborate with the server. In reality, however, it may not be possible to modify the client protocols. Motivated by this limitation, this work focuses on clients who always maximize their individual cumulative rewards, and introduces a novel idea of reward teaching, where the server guides the clients towards global optimality through implicit local reward adjustments. Under this framework, the server faces two tightly coupled tasks of bandit learning and target teaching, whose combination is non-trivial and challenging. A novel algorithm, called Teaching-After-Learning (TAL), is proposed, which encourages and discourages clients’ explorations separately. General performance analyses of TAL on regret and cost are first established when the clients’ strategies satisfy certain requirements. To particularize the results, clients with UCB or ε-greedy strategies are then considered, where novel technical approaches are developed to analyze their warm-start behaviors. The obtained guarantees concretely demonstrate that when facing these client strategies, TAL achieves logarithmic regrets while only incurring logarithmic adjustment costs, which is order-optimal w.r.t. a natural lower bound.

Existing studies on provably efficient algorithms for Markov games (MGs) almost exclusively build on the “optimism in the face of uncertainty” (OFU) principle. This work focuses on a distinct approach of posterior sampling, which is celebrated in many bandits and reinforcement learning settings but remains under-explored for MGs. Specifically, for episodic two-player zero-sum MGs, a novel posterior sampling algorithm is developed with general function approximation. Theoretical analysis demonstrates that the posterior sampling algorithm admits a $\sqrt{T}$-regret bound for problems with a low multi-agent decoupling coefficient, which is a new complexity measure for MGs, where $T$ denotes the number of episodes. When specializing to linear MGs, the obtained regret bound matches the state-of-the-art results. To the best of our knowledge, this is the first provably efficient posterior sampling algorithm for MGs with frequentist regret guarantees, which extends the toolbox for MGs and promotes the broad applicability of posterior sampling.

Federated multi-armed bandits (FMAB) is a new bandit paradigm that parallels the federated learning (FL) framework in supervised learning. It is inspired by practical applications in cognitive radio and recommender systems, and enjoys features that are analogous to FL. This paper proposes a general framework of FMAB and then studies two specific federated bandit models. We first study the approximate model where the heterogeneous local models are random realizations of the global model from an unknown distribution. This model introduces a new uncertainty of client sampling, as the global model may not be reliably learned even if the finite local models are perfectly known. Furthermore, this uncertainty cannot be quantified a priori without knowledge of the suboptimality gap. We solve the approximate model by proposing Federated Double UCB (Fed2-UCB), which constructs a novel “double UCB” principle accounting for uncertainties from both arm and client sampling. We show that gradually admitting new clients is critical in achieving an O(log(T)) regret while explicitly considering the communication loss. The exact model, where the global bandit model is the exact average of heterogeneous local models, is then studied as a special case. We show that, somewhat surprisingly, the order-optimal regret can be achieved independent of the number of clients with a careful choice of the update periodicity. Experiments using both synthetic and real-world datasets corroborate the theoretical analysis and demonstrate the effectiveness and efficiency of the proposed algorithms.

A general framework of personalized federated multi-armed bandits (PF-MAB) is proposed, which is a new bandit paradigm analogous to the federated learning (FL) framework in supervised learning and enjoys the features of FL with personalization. Under the PF-MAB framework, a mixed bandit learning problem that flexibly balances generalization and personalization is studied. A lower bound analysis for the mixed model is presented. We then propose the Personalized Federated Upper Confidence Bound (PF-UCB) algorithm, where the exploration length is chosen carefully to achieve the desired balance of learning the local model and supplying global information for the mixed learning objective. Theoretical analysis proves that PF-UCB achieves an O(log(T)) regret regardless of the degree of personalization, and has a similar instance dependency as the lower bound. Experiments using both synthetic and real-world datasets corroborate the theoretical analysis and demonstrate the effectiveness of the proposed algorithm.

In this work, we study the no-sensing adversarial multi-player multi-armed bandits problem. A new dimension of hardness, called attackability, is introduced, which is orthogonal to the hardness of multiple players. All adversaries can be categorized based on the attackability and we introduce Adversary-Adaptive Collision-Communication (A2C2), a family of algorithms with forced-collision communications among players. Information-theoretic tools of the Z-channel model, error-correction/detection coding, and randomized communication are utilized to address the challenge of implicit communication without collision information in an adversarial environment. Theoretical analysis proves that asymptotic attackability-dependent sublinear regrets can be achieved, which do not have an exponential dependence on the number of players and as a result reveal a fundamental tradeoff between the two dimensions of hardness in this problem.

Despite the significant interests and many progresses in decentralized multi-player multi-armed bandits (MP-MAB) problems in recent years, the regret gap to the natural centralized lower bound in the heterogeneous MP-MAB setting remains open. In this paper, we propose BEACON -- Batched Exploration with Adaptive COmmunicatioN -- that closes this gap. BEACON accomplishes this goal with novel contributions in implicit communication and efficient exploration. For the former, we propose a novel adaptive differential communication (ADC) design that significantly improves the implicit communication efficiency. For the latter, a carefully crafted batched exploration scheme is developed to enable incorporation of the combinatorial upper confidence bound (CUCB) principle. We then generalize the existing linear-reward MP-MAB problems, where the system reward is always the sum of individually collected rewards, to a new MP-MAB problem where the system reward is a general (nonlinear) function of individual rewards. We extend BEACON to solve this problem and prove a logarithmic regret. BEACON bridges the algorithm design and regret analysis of combinatorial MAB (CMAB) and MP-MAB, two largely disjointed areas in MAB, and the results in this paper suggest that this previously ignored connection is worth further investigation.

Incentivized exploration in multi-armed bandits (MAB) has witnessed increasing interests and many progresses in recent years, where a principal offers bonuses to agents to do explorations on her behalf. However, almost all existing studies are confined to temporary myopic agents. In this work, we break this barrier and study incentivized exploration with multiple and long-term strategic agents, who have more complicated behaviors that often appear in real-world applications. An important observation of this work is that strategic agents' intrinsic needs of learning benefit (instead of harming) the principal's explorations by providing "free pulls". Moreover, it turns out that increasing the population of agents significantly lowers the principal's burden of incentivizing. The key and somewhat surprising insight revealed from our results is that when there are sufficiently many learning agents involved, the exploration process of the principal can be (almost) free. Our main results are built upon three novel components which may be of independent interest: (1) a simple yet provably effective incentive-provision strategy; (2) a carefully crafted best arm identification algorithm for rewards aggregated under unequal confidences; (3) a high-probability finite-time lower bound of UCB algorithms. Experimental results are provided to complement the theoretical analysis.

The decentralized stochastic multi-player multi-armed bandit (MP-MAB) problem, where the collision information is not available to the players, is studied in this paper. Building on the seminal work of Boursier and Perchet (2019), we propose error correction synchronization involving communication (EC-SIC), whose regret is shown to approach that of the centralized stochastic MP-MAB with collision information. By recognizing that the communication phase without collision information corresponds to the Z-channel model in information theory, the proposed EC-SIC algorithm applies optimal error correction coding for the communication of reward statistics. A fixed message length, as opposed to the logarithmically growing one in Boursier and Perchet (2019), also plays a crucial role in controlling the communication loss. Experiments with practical Z-channel codes, such as repetition code, flip code and modified Hamming code, demonstrate the superiority of EC-SIC in both synthetic and real-world datasets.

Recent years have witnessed deep neural networks (DNNs) become the de facto tool in many applications such as image classification and speech recognition. But significant unmet needs remain in performing DNN inference tasks on mobile devices. Although edge computing enables complex DNN inference tasks to be performed in close proximity to the mobile device, performance optimization requires a carefully designed synergy between the edge and the mobile device. Moreover, the confidentiality of uploaded data to the possibly untrusted edge server is of great concern. In this paper, we investigate the impact of DNN partitioning on the inference latency performance and the privacy risks in edge computing. Based on the obtained insights, we design an offloading strategy that adaptively partitions the DNN in varying network environments to make the optimal tradeoff between performance and privacy for battery-powered mobile devices. This strategy is designed under the learning-aided Lyapunov optimization framework and has a provable performance guarantee. Finally, we build a small- scale testbed to demonstrate the efficacy of the proposed offloading scheme.