Decoupling Corruption and Horizon in Robust Contextual Pricing

2026-07-13

AI Summary

This work addresses the robust repeated contextual pricing problem under adversarial corruption, where user valuations depend linearly on contextual features and only binary sales feedback—potentially corrupted in up to $C$ rounds—is observed per round. The authors propose a novel robust pricing algorithm that integrates adaptive confidence intervals with a corruption detection mechanism. Under the linear contextual model, this approach achieves the first regret bound that decouples the corruption budget $C$ from the time horizon $T$, attaining a regret of $O(Cd + d^2 \log T)$, where $d$ denotes the context dimension. Both theoretical analysis and empirical experiments demonstrate significant improvements over existing methods, thereby resolving an open problem posed by Gupta et al. (2025).

Abstract

We study robust repeated contextual pricing, where valuations depends linearly on the features. At each round $t\in[T]$, a seller observes a context, posts a price, and receives only a possibly corrupted binary sale feedback. The seller knows an upper bound $C$ on the number of corrupted rounds. We design an algorithm with regret $\mathcal O(Cd+d^2\log T)$, where $d$ is the context dimension. This is the first guarantee for robust contextual pricing that separates the dependence on the corruption budget $C$ from the horizon $T$, closing the problem left open by Gupta, Guruganesh, Paes Leme, and Schneider (2025).

Problem

• robust contextual pricing
• corruption
• regret bound
• horizon
• contextual bandits

Innovation

• robust contextual pricing
• corruption budget
• regret bound
• decoupling
• online learning