Contextual bandits constitute a classical framework for decision-making under uncertainty. In this setting, the goal is to learn the arms of highest reward subject to contextual information, while the unknown reward parameters of each arm need to be learned by experimenting that specific arm. Accordingly, a fundamental problem is that of balancing exploration (i.e., pulling different arms to learn their parameters), versus exploitation (i.e., pulling the best arms to gain reward). To study this problem, the existing literature mostly considers perfectly observed contexts. However, the setting of partial context observations remains unexplored to date, despite being theoretically more general and practically more versatile. We study bandit policies for learning to select optimal arms based on the data of observations, which are noisy linear functions of the unobserved context vectors. Our theoretical analysis shows that the Thompson sampling policy successfully balances exploration and exploitation. Specifically, we establish the followings: (i) regret bounds that grow poly-logarithmically with time, (ii) square-root consistency of parameter estimation, and (iii) scaling of the regret with other quantities including dimensions and number of arms. Extensive numerical experiments with both real and synthetic data are presented as well, corroborating the efficacy of Thompson sampling. To establish the results, we introduce novel martingale techniques and concentration inequalities to address partially observed dependent random variables generated from unspecified distributions, and also leverage problem-dependent information to sharpen probabilistic bounds for time-varying suboptimality gaps. These techniques pave the road towards studying other decision-making problems with contextual information as well as partial observations.
Efficient exploration is a key challenge in contextual bandits due to the large size of their action space, where uninformed exploration can result in computational and statistical inefficiencies. Fortunately, the rewards of actions are often correlated and this can be leveraged to explore them efficiently. In this work, we capture such correlations using pre-trained diffusion models; upon which we design diffusion Thompson sampling (dTS). Both theoretical and algorithmic foundations are developed for dTS, and empirical evaluation also shows its favorable performance.
This work tackles the complexities of multi-player scenarios in \emph{unknown games}, where the primary challenge lies in navigating the uncertainty of the environment through bandit feedback alongside strategic decision-making. We introduce Thompson Sampling (TS)-based algorithms that exploit the information of opponents' actions and reward structures, leading to a substantial reduction in experimental budgets -- achieving over tenfold improvements compared to conventional approaches. Notably, our algorithms demonstrate that, given specific reward structures, the regret bound depends logarithmically on the total action space, significantly alleviating the curse of multi-player. Furthermore, we unveil the \emph{Optimism-then-NoRegret} (OTN) framework, a pioneering methodology that seamlessly incorporates our advancements with established algorithms, showcasing its utility in practical scenarios such as traffic routing and radar sensing in the real world.
We propose a new framework for contextual multi-armed bandits based on tree ensembles. Our framework adapts two widely used bandit methods, Upper Confidence Bound and Thompson Sampling, for both standard and combinatorial settings. As part of this framework, we propose a novel method of estimating the uncertainty in tree ensemble predictions. We further demonstrate the effectiveness of our framework via several experimental studies, employing XGBoost and random forests, two popular tree ensemble methods. Compared to state-of-the-art methods based on decision trees and neural networks, our methods exhibit superior performance in terms of both regret minimization and computational runtime, when applied to benchmark datasets and the real-world application of navigation over road networks.
Public health programs often provide interventions to encourage program adherence, and effectively allocating interventions is vital for producing the greatest overall health outcomes, especially in underserved communities where resources are limited. Such resource allocation problems are often modeled as restless multi-armed bandits (RMABs) with unknown underlying transition dynamics, hence requiring online reinforcement learning (RL). We present Bayesian Learning for Contextual RMABs (BCoR), an online RL approach for RMABs that novelly combines techniques in Bayesian modeling with Thompson sampling to flexibly model the complex RMAB settings present in public health program adherence problems, namely context and non-stationarity. BCoR's key strength is the ability to leverage shared information within and between arms to learn the unknown RMAB transition dynamics quickly in intervention-scarce settings with relatively short time horizons, which is common in public health applications. Empirically, BCoR achieves substantially higher finite-sample performance over a range of experimental settings, including a setting using real-world adherence data that was developed in collaboration with ARMMAN, an NGO in India which runs a large-scale maternal mHealth program, showcasing BCoR practical utility and potential for real-world deployment.
We present evidence of substantial benefit from efficient exploration in gathering human feedback to improve large language models. In our experiments, an agent sequentially generates queries while fitting a reward model to the feedback received. Our best-performing agent generates queries using double Thompson sampling, with uncertainty represented by an epistemic neural network. Our results demonstrate that efficient exploration enables high levels of performance with far fewer queries. Further, both uncertainty estimation and the choice of exploration scheme play critical roles.
We explore a stochastic contextual linear bandit problem where the agent observes a noisy, corrupted version of the true context through a noise channel with an unknown noise parameter. Our objective is to design an action policy that can approximate" that of an oracle, which has access to the reward model, the channel parameter, and the predictive distribution of the true context from the observed noisy context. In a Bayesian framework, we introduce a Thompson sampling algorithm for Gaussian bandits with Gaussian context noise. Adopting an information-theoretic analysis, we demonstrate the Bayesian regret of our algorithm concerning the oracle's action policy. We also extend this problem to a scenario where the agent observes the true context with some delay after receiving the reward and show that delayed true contexts lead to lower Bayesian regret. Finally, we empirically demonstrate the performance of the proposed algorithms against baselines.
Approximate Thompson sampling with Langevin Monte Carlo broadens its reach from Gaussian posterior sampling to encompass more general smooth posteriors. However, it still encounters scalability issues in high-dimensional problems when demanding high accuracy. To address this, we propose an approximate Thompson sampling strategy, utilizing underdamped Langevin Monte Carlo, where the latter is the go-to workhorse for simulations of high-dimensional posteriors. Based on the standard smoothness and log-concavity conditions, we study the accelerated posterior concentration and sampling using a specific potential function. This design improves the sample complexity for realizing logarithmic regrets from $\mathcal{\tilde O}(d)$ to $\mathcal{\tilde O}(\sqrt{d})$. The scalability and robustness of our algorithm are also empirically validated through synthetic experiments in high-dimensional bandit problems.
Industrial Internet of Things (IIoT) networks will provide Ultra-Reliable Low-Latency Communication (URLLC) to support critical processes underlying the production chains. However, standard protocols for allocating wireless resources may not optimize the latency-reliability trade-off, especially for uplink communication. For example, centralized grant-based scheduling can ensure almost zero collisions, but introduces delays in the way resources are requested by the User Equipments (UEs) and granted by the gNB. In turn, distributed scheduling (e.g., based on random access), in which UEs autonomously choose the resources for transmission, may lead to potentially many collisions especially when the traffic increases. In this work we propose DIStributed combinatorial NEural linear Thompson Sampling (DISNETS), a novel scheduling framework that combines the best of the two worlds. By leveraging a feedback signal from the gNB and reinforcement learning, the UEs are trained to autonomously optimize their uplink transmissions by selecting the available resources to minimize the number of collisions, without additional message exchange to/from the gNB. DISNETS is a distributed, multi-agent adaptation of the Neural Linear Thompson Sampling (NLTS) algorithm, which has been further extended to admit multiple parallel actions. We demonstrate the superior performance of DISNETS in addressing URLLC in IIoT scenarios compared to other baselines.
The flexibility of choosing the ad action as a function of the consumer state is critical for modern-day marketing campaigns. We study the problem of identifying the optimal sequential personalized interventions that maximize the adoption probability for a new product. We model consumer behavior by a conversion funnel that captures the state of each consumer (e.g., interaction history with the firm) and allows the consumer behavior to vary as a function of both her state and firm's sequential interventions. We show our model captures consumer behavior with very high accuracy (out-of-sample AUC of over 0.95) in a real-world email marketing dataset. However, it results in a very large-scale learning problem, where the firm must learn the state-specific effects of various interventions from consumer interactions. We propose a novel attribution-based decision-making algorithm for this problem that we call model-free approximate Bayesian learning. Our algorithm inherits the interpretability and scalability of Thompson sampling for bandits and maintains an approximate belief over the value of each state-specific intervention. The belief is updated as the algorithm interacts with the consumers. Despite being an approximation to the Bayes update, we prove the asymptotic optimality of our algorithm and analyze its convergence rate. We show that our algorithm significantly outperforms traditional approaches on extensive simulations calibrated to a real-world email marketing dataset.
Multi-agent multi-target tracking has a wide range of applications, including wildlife patrolling, security surveillance or environment monitoring. Such algorithms often make restrictive assumptions: the number of targets and/or their initial locations may be assumed known, or agents may be pre-assigned to monitor disjoint partitions of the environment, reducing the burden of exploration. This also limits applicability when there are fewer agents than targets, since agents are unable to continuously follow the targets in their fields of view. Multi-agent tracking algorithms additionally assume inter-agent synchronization of observations, or the presence of a central controller to coordinate joint actions. Instead, we focus on the setting of decentralized multi-agent, multi-target, simultaneous active search-and-tracking with asynchronous inter-agent communication. Our proposed algorithm DecSTER uses a sequential monte carlo implementation of the probability hypothesis density filter for posterior inference combined with Thompson sampling for decentralized multi-agent decision making. We compare different action selection policies, focusing on scenarios where targets outnumber agents. In simulation, we demonstrate that DecSTER is robust to unreliable inter-agent communication and outperforms information-greedy baselines in terms of the Optimal Sub-Pattern Assignment (OSPA) metric for different numbers of targets and varying teamsizes.
Sequential optimization methods are often confronted with the curse of dimensionality in high-dimensional spaces. Current approaches under the Gaussian process framework are still burdened by the computational complexity of tracking Gaussian process posteriors and need to partition the optimization problem into small regions to ensure exploration or assume an underlying low-dimensional structure. With the idea of transiting the candidate points towards more promising positions, we propose a new method based on Markov Chain Monte Carlo to efficiently sample from an approximated posterior. We provide theoretical guarantees of its convergence in the Gaussian process Thompson sampling setting. We also show experimentally that both the Metropolis-Hastings and the Langevin Dynamics version of our algorithm outperform state-of-the-art methods in high-dimensional sequential optimization and reinforcement learning benchmarks.
Anytime multi-agent path finding (MAPF) is a promising approach to scalable path optimization in large-scale multi-agent systems. State-of-the-art anytime MAPF is based on Large Neighborhood Search (LNS), where a fast initial solution is iteratively optimized by destroying and repairing a fixed number of parts, i.e., the neighborhood, of the solution, using randomized destroy heuristics and prioritized planning. Despite their recent success in various MAPF instances, current LNS-based approaches lack exploration and flexibility due to greedy optimization with a fixed neighborhood size which can lead to low quality solutions in general. So far, these limitations have been addressed with extensive prior effort in tuning or offline machine learning beyond actual planning. In this paper, we focus on online learning in LNS and propose Bandit-based Adaptive LArge Neighborhood search Combined with Exploration (BALANCE). BALANCE uses a bi-level multi-armed bandit scheme to adapt the selection of destroy heuristics and neighborhood sizes on the fly during search. We evaluate BALANCE on multiple maps from the MAPF benchmark set and empirically demonstrate cost improvements of at least 50% compared to state-of-the-art anytime MAPF in large-scale scenarios. We find that Thompson Sampling performs particularly well compared to alternative multi-armed bandit algorithms.
Many real-world bandit applications are characterized by sparse rewards, which can significantly hinder learning efficiency. Leveraging problem-specific structures for careful distribution modeling is recognized as essential for improving estimation efficiency in statistics. However, this approach remains under-explored in the context of bandits. To address this gap, we initiate the study of zero-inflated bandits, where the reward is modeled using a classic semi-parametric distribution known as the zero-inflated distribution. We develop algorithms based on the Upper Confidence Bound and Thompson Sampling frameworks for this specific structure. The superior empirical performance of these methods is demonstrated through extensive numerical studies.
We study the multi-agent multi-armed bandit (MAMAB) problem, where $m$ agents are factored into $\rho$ overlapping groups. Each group represents a hyperedge, forming a hypergraph over the agents. At each round of interaction, the learner pulls a joint arm (composed of individual arms for each agent) and receives a reward according to the hypergraph structure. Specifically, we assume there is a local reward for each hyperedge, and the reward of the joint arm is the sum of these local rewards. Previous work introduced the multi-agent Thompson sampling (MATS) algorithm \citep{verstraeten2020multiagent} and derived a Bayesian regret bound. However, it remains an open problem how to derive a frequentist regret bound for Thompson sampling in this multi-agent setting. To address these issues, we propose an efficient variant of MATS, the $\epsilon$-exploring Multi-Agent Thompson Sampling ($\epsilon$-MATS) algorithm, which performs MATS exploration with probability $\epsilon$ while adopts a greedy policy otherwise. We prove that $\epsilon$-MATS achieves a worst-case frequentist regret bound that is sublinear in both the time horizon and the local arm size. We also derive a lower bound for this setting, which implies our frequentist regret upper bound is optimal up to constant and logarithm terms, when the hypergraph is sufficiently sparse. Thorough experiments on standard MAMAB problems demonstrate the superior performance and the improved computational efficiency of $\epsilon$-MATS compared with existing algorithms in the same setting.
Recently multi-armed bandit problem arises in many real-life scenarios where arms must be sampled in batches, due to limited time the agent can wait for the feedback. Such applications include biological experimentation and online marketing. The problem is further complicated when the number of arms is large and the number of batches is small. We consider pure exploration in a batched multi-armed bandit problem. We introduce a general linear programming framework that can incorporate objectives of different theoretical settings in best arm identification. The linear program leads to a two-stage algorithm that can achieve good theoretical properties. We demonstrate by numerical studies that the algorithm also has good performance compared to certain UCB-type or Thompson sampling methods.
We consider the combinatorial volatile Gaussian process (GP) semi-bandit problem. Each round, an agent is provided a set of available base arms and must select a subset of them to maximize the long-term cumulative reward. We study the Bayesian setting and provide novel Bayesian cumulative regret bounds for three GP-based algorithms: GP-UCB, GP-BayesUCB and GP-TS. Our bounds extend previous results for GP-UCB and GP-TS to the infinite, volatile and combinatorial setting, and to the best of our knowledge, we provide the first regret bound for GP-BayesUCB. Volatile arms encompass other widely considered bandit problems such as contextual bandits. Furthermore, we employ our framework to address the challenging real-world problem of online energy-efficient navigation, where we demonstrate its effectiveness compared to the alternatives.
Mobile health leverages personalized and contextually tailored interventions optimized through bandit and reinforcement learning algorithms. In practice, however, challenges such as participant heterogeneity, nonstationarity, and nonlinear relationships hinder algorithm performance. We propose RoME, a Robust Mixed-Effects contextual bandit algorithm that simultaneously addresses these challenges via (1) modeling the differential reward with user- and time-specific random effects, (2) network cohesion penalties, and (3) debiased machine learning for flexible estimation of baseline rewards. We establish a high-probability regret bound that depends solely on the dimension of the differential-reward model, enabling us to achieve robust regret bounds even when the baseline reward is highly complex. We demonstrate the superior performance of the RoME algorithm in a simulation and two off-policy evaluation studies.
Transformer requires a fixed number of layers and heads which makes them inflexible to the complexity of individual samples and expensive in training and inference. To address this, we propose a sample-based Dynamic Hierarchical Transformer (DHT) model whose layers and heads can be dynamically configured with single data samples via solving contextual bandit problems. To determine the number of layers and heads, we use the Uniform Confidence Bound while we deploy combinatorial Thompson Sampling in order to select specific head combinations given their number. Different from previous work that focuses on compressing trained networks for inference only, DHT is not only advantageous for adaptively optimizing the underlying network architecture during training but also has a flexible network for efficient inference. To the best of our knowledge, this is the first comprehensive data-driven dynamic transformer without any additional auxiliary neural networks that implement the dynamic system. According to the experiment results, we achieve up to 74% computational savings for both training and inference with a minimal loss of accuracy.
This paper studies the one-shot behavior of no-regret algorithms for stochastic bandits. Although many algorithms are known to be asymptotically optimal with respect to the expected regret, over a single run, their pseudo-regret seems to follow one of two tendencies: it is either smooth or bumpy. To measure this tendency, we introduce a new notion: the sliding regret, that measures the worst pseudo-regret over a time-window of fixed length sliding to infinity. We show that randomized methods (e.g. Thompson Sampling and MED) have optimal sliding regret, while index policies, although possibly asymptotically optimal for the expected regret, have the worst possible sliding regret under regularity conditions on their index (e.g. UCB, UCB-V, KL-UCB, MOSS, IMED etc.). We further analyze the average bumpiness of the pseudo-regret of index policies via the regret of exploration, that we show to be suboptimal as well.