What Doubling Tricks Can and Can't Do for Multi-Armed Bandits

Abstract

An online reinforcement learning algorithm is anytime if it does not need to know in advance the horizon T of the experiment. A well-known technique to obtain an anytime algorithm from any non-anytime algorithm is the "Doubling Trick". In the context of adversarial or stochastic multi-armed bandits, the performance of an algorithm is measured by its regret, and we study two families of sequences of growing horizons (geometric and exponential) to generalize previously known results that certain doubling tricks can be used to conserve certain regret bounds. In a broad setting, we prove that a geometric doubling trick can be used to conserve (minimax) bounds in $R\_T=O(\sqrt{T})$ but cannot conserve (distribution-dependent) bounds in $R\_T=O(\log T)$. We give insights as to why exponential doubling tricks may be better, as they conserve bounds in $R\_T=O(\log T)$, and are close to conserving bounds in $R\_T=O(\sqrt{T})$.

References in corpus (5)

• A Contextual-Bandit Approach to Personalized News Article Recommendation
• Analysis of Thompson Sampling for the multi-armed bandit problem
• Risk-Aversion in Multi-armed Bandits
• On Explore-Then-Commit Strategies
• On the Complexity of A/B Testing

Cited by in corpus (16)

• Reinforcement Learning in Feature Space: Matrix Bandit, Kernels, and Regret Bound
• A Practical Algorithm for Multiplayer Bandits when Arm Means Vary Among Players
• Provably Efficient Reinforcement Learning for Discounted MDPs with Feature Mapping
• Logarithmic Regret for Reinforcement Learning with Linear Function Approximation
• Multi-player Multi-armed Bandits with Collision-Dependent Reward Distributions
• Tsallis-INF: An Optimal Algorithm for Stochastic and Adversarial Bandits
• Autonomous Drug Design with Multi-Armed Bandits
• Efficient Pure Exploration for Combinatorial Bandits with Semi-Bandit Feedback
• Functional Sequential Treatment Allocation with Covariates
• Be Greedy in Multi-Armed Bandits
• Online Sub-Sampling for Reinforcement Learning with General Function Approximation
• Generalized non-stationary bandits
• Ballooning Multi-Armed Bandits
• Constant or logarithmic regret in asynchronous multiplayer bandits
• DART: aDaptive Accept RejecT for non-linear top-K subset identification
• Sequential Choice Bandits with Feedback for Personalizing users' experience