arXiv research

Develops parameter-free online mirror descent for optimal dynamic regret.

New algorithms achieve high-probability parameter-free regret in online convex optimization with heavy-tailed data.

We introduce several new black-box reductions that significantly improve the design of adaptive and parameter-free online learning algorithms by simplifying analysis, improving regret guarantees, and sometimes even improving runtime. We reduce parameter-free online learning to online exp-concave optimization, we reduce…

We consider the problem of unconstrained online convex optimization (OCO) with sub-exponential noise, a strictly more general problem than the standard OCO. In this setting, the learner receives a subgradient of the loss functions corrupted by sub-exponential noise and strives to achieve optimal regret guarantee, witho…

Develops a parameter-free SGD algorithm with optimal convergence rate.

New algorithms reduce online learning regret by tracking gradient variation.

A parameter-free PGD algorithm for convex optimization.

New RL algorithm tackles nonstationary MDPs with linear approximations and varying rewards.

New method tracks shifts in infinite-armed bandits without prior knowledge.

New algorithm for online learning with noisy side observations.

Efficient methods reduce projections in non-stationary online learning.

A new algorithm reduces online exp-concave optimization runtime.

We propose the first contextual bandit algorithm that is parameter-free, efficient, and optimal in terms of dynamic regret. Specifically, our algorithm achieves dynamic regret $\mathcal{O}(\min\{\sqrt{ST}, Δ^{\frac{1}{3}}T^{\frac{2}{3}}\})$ for a contextual bandit problem with $T$ rounds, $S$ switches and $Δ$ total var…

We show how to take any two parameter-free online learning algorithms with different regret guarantees and obtain a single algorithm whose regret is the minimum of the two base algorithms. Our method is embarrassingly simple: just add the iterates. This trick can generate efficient algorithms that adapt to many norms s…

This paper describes a new parameter-free online learning algorithm for changing environments. In comparing against algorithms with the same time complexity as ours, we obtain a strongly adaptive regret bound that is a factor of at least $\sqrt{\log(T)}$ better, where $T$ is the time horizon. Empirical results show tha…

New algorithm minimizes cumulative loss in dynamic linear bandits without prior knowledge of comparator switches.

Paper studies continuous prediction with experts' advice using differential equations.

New algorithm reduces dynamic regret in time-varying movement costs.

New algorithms minimize regret in SSP with optimal sparse updates.

A new UCB algorithm for heavy-tailed bandits with near-optimal regret.

We uncover a fairly general principle in online learning: If regret can be (approximately) expressed as a function of certain "sufficient statistics" for the data sequence, then there exists a special Burkholder function that 1) can be used algorithmically to achieve the regret bound and 2) only depends on these suffic…

We consider the problem of online linear regression on arbitrary deterministic sequences when the ambient dimension d can be much larger than the number of time rounds T. We introduce the notion of sparsity regret bound, which is a deterministic online counterpart of recent risk bounds derived in the stochastic setting…

Most contextual bandit algorithms minimize regret against the best fixed policy, a questionable benchmark for non-stationary environments that are ubiquitous in applications. In this work, we develop several efficient contextual bandit algorithms for non-stationary environments by equipping existing methods for i.i.d. …

We develop a novel and generic algorithm for the adversarial multi-armed bandit problem (or more generally the combinatorial semi-bandit problem). When instantiated differently, our algorithm achieves various new data-dependent regret bounds improving previous work. Examples include: 1) a regret bound depending on the …

New method achieves optimal performance without needing problem parameters.

Algorithm learns from changing zero-sum games with no regret.

Thresholded Lasso bandit minimizes regret in sparse linear bandits.

Develops a new method for online conformal prediction without manual tuning.

We consider a variation on the problem of prediction with expert advice, where new forecasters that were unknown until then may appear at each round. As often in prediction with expert advice, designing an algorithm that achieves near-optimal regret guarantees is straightforward, using aggregation of experts. However, …

In this paper, we investigate the non-stationary combinatorial semi-bandit problem, both in the switching case and in the dynamic case. In the general case where (a) the reward function is non-linear, (b) arms may be probabilistically triggered, and (c) only approximate offline oracle exists \cite{wang2017improving}, o…

Proposes a tuning-free dynamic pricing method for linear valuation models.

New RL method tackles dynamic MDPs with evolving rewards and states.

New algorithm closes empirical gap in PFSGD performance.

New algorithm tackles dynamic query routing to multiple embedding models.

Simpler, parameter-free AdaGrad and Adam variants with convergence guarantees.

New algorithm reduces online learning regret in uninformed Markov games.

A key challenge in online learning is that classical algorithms can be slow to adapt to changing environments. Recent studies have proposed "meta" algorithms that convert any online learning algorithm to one that is adaptive to changing environments, where the adaptivity is analyzed in a quantity called the strongly-ad…

New geometric metrics improve Bayesian optimization performance evaluation.

New algorithm provides robust uncertainty quantification without parameter tuning.

We investigate the piecewise-stationary combinatorial semi-bandit problem. Compared to the original combinatorial semi-bandit problem, our setting assumes the reward distributions of base arms may change in a piecewise-stationary manner at unknown time steps. We propose an algorithm, \texttt{GLR-CUCB}, which incorporat…

A method learns common bias for multiple low-variance tasks without hyper-parameter tuning.

Adaptive clustering and personalization algorithms minimize regret in multi-agent stochastic linear bandits.

New algorithms adapt to both gradient norms and comparator norms in online learning.

We introduce data-driven decision-making algorithms that achieve state-of-the-art \emph{dynamic regret} bounds for non-stationary bandit settings. These settings capture applications such as advertisement allocation, dynamic pricing, and traffic network routing in changing environments. We show how the difficulty posed…

Cascading bandit (CB) is a popular model for web search and online advertising, where an agent aims to learn the $K$ most attractive items out of a ground set of size $L$ during the interaction with a user. However, the stationary CB model may be too simple to apply to real-world problems, where user preferences may ch…

We study the problem of optimizing a function under a \emph{budgeted number of evaluations}. We only assume that the function is \emph{locally} smooth around one of its global optima. The difficulty of optimization is measured in terms of 1) the amount of \emph{noise} $b$ of the function evaluation and 2) the local smo…

We study small-loss bounds for adversarial multi-armed bandits with graph feedback, that is, adaptive regret bounds that depend on the loss of the best arm or related quantities, instead of the total number of rounds. We derive the first small-loss bound for general strongly observable graphs, resolving an open problem…

We consider un-discounted reinforcement learning (RL) in Markov decision processes (MDPs) under temporal drifts, ie, both the reward and state transition distributions are allowed to evolve over time, as long as their respective total variations, quantified by suitable metrics, do not exceed certain variation budgets. …