We investigate the extent to which offline demonstration data can improve online learning. It
is natural to expect some improvement, but the question is how, and by how much? We …

While learning in an unknown Markov Decision Process (MDP), an agent should trade off
exploration to discover new information about the MDP, and exploitation of the current …

We study stochastic structured bandits for minimizing regret. The fact that the popular
optimistic algorithms do not achieve the asymptotic instance-dependent regret optimality …

We study finite-armed stochastic bandits where the rewards of each arm might be correlated
to those of other arms. We introduce a novel phased algorithm that exploits the given …

Influence maximization, adaptive routing, and dynamic spectrum allocation all require
choosing the right action from a large set of alternatives. Thanks to the advances in …

Problem definition: Sequential experiments that are deployed in a broad range of practices
are characterized by an exploration-exploitation trade-off that is well understood when in …

In this work we consider a seller who sells an item via second-price auctions with a reserve
price. By controlling the reserve price, the seller can influence the revenue from the auction …

The demand for seamless Internet access under extreme user mobility, such as on high-
speed trains and vehicles, has become a norm rather than an exception. However, the …

This work considers training conditional probability distributions called policies, using
simulated environments via gradient-based optimization methods. It begins by investigating …

We consider the bandit-based framework for diversity-preserving recommendations
introduced by Celis et al.(2019), who approached it mainly by a reduction to the setting of …