Safe Linear Bandits over Unknown Polytopes

TL;DR

This paper introduces a new optimistic strategy for safe linear bandits over unknown polytopes, achieving near-optimal regret and safety violation bounds despite the challenge of unknown constraints.

Contribution

It proposes DOSS, a doubly-optimistic algorithm, with tight regret and safety bounds, and develops a novel dual analysis framework for the problem.

Findings

DOSS achieves $O( ext{log}^2 T)$ regret bounds.

DOSS maintains $ ilde O( ext{sqrt}(T))$ safety violations.

The analysis introduces dual notions of gaps based on sensitivity analysis.

Abstract

The safe linear bandit problem (SLB) is an online approach to linear programming with unknown objective and unknown roundwise constraints, under stochastic bandit feedback of rewards and safety risks of actions. We study the tradeoffs between efficacy and smooth safety costs of SLBs over polytopes, and the role of aggressive doubly-optimistic play in avoiding the strong assumptions made by extant pessimistic-optimistic approaches. We first elucidate an inherent hardness in SLBs due the lack of knowledge of constraints: there exist `easy' instances, for which suboptimal extreme points have large `gaps', but on which SLB methods must still incur $\Omega(\sqrt{T})$ regret or safety violations, due to an inability to resolve unknown optima to arbitrary precision. We then analyse a natural doubly-optimistic strategy for the safe linear bandit problem, DOSS, which uses optimistic estimates of both reward and safety risks to select actions, and show that despite the lack of knowledge of constraints or feasible points, DOSS simultaneously obtains tight instance-dependent $O(\log^2 T)$ bounds on efficacy regret, and $\tilde O(\sqrt{T})$ bounds on safety violations. Further, when safety is demanded to a finite precision, violations improve to $O(\log^2 T).$ These results rely on a novel dual analysis of linear bandits: we argue that \algoname proceeds by activating noisy versions of at least $d$ constraints in each round, which allows us to separately analyse rounds where a `poor' set of constraints is activated, and rounds where `good' sets of constraints are activated. The costs in the former are controlled to $O(\log^2 T)$ by developing new dual notions of gaps, based on global sensitivity analyses of linear programs, that quantify the suboptimality of each such set of constraints. The latter costs are controlled to $O(1)$ by explicitly analysing the solutions of optimistic play.