Targeting Makes Sample Efficiency in Auction Design

TL;DR

This paper proposes a targeted sampling model for auction design that significantly reduces sample complexity compared to traditional i.i.d. sampling, especially when the seller can specify quantile intervals.

Contribution

It introduces a new targeted sampling framework with a parameter  that interpolates between existing models, improving sample efficiency in auction learning.

Findings

Targeted sampling reduces sample complexity from (n  ) to () for bounded valuations.

Even mild targeting power (=o(1)) yields better sample complexity than i.i.d. sampling.

The approach applies across the full range of targeting power,   [0,1).

Abstract

This paper introduces the targeted sampling model in optimal auction design. In this model, the seller may specify a quantile interval and sample from a buyer's prior restricted to the interval. This can be interpreted as allowing the seller to, for example, examine the top $40$ percents bids from previous buyers with the same characteristics. The targeting power is quantified with a parameter $\Delta \in [0, 1]$ which lower bounds how small the quantile intervals could be. When $\Delta = 1$, it degenerates to Cole and Roughgarden's model of i.i.d. samples; when it is the idealized case of $\Delta = 0$, it degenerates to the model studied by Chen et al. (2018). For instance, for $n$ buyers with bounded values in $[0, 1]$, $\tilde{O}(\epsilon^{-1})$ targeted samples suffice while it is known that at least $\tilde{\Omega}(n \epsilon^{-2})$ i.i.d. samples are needed. In other words, targeted sampling with sufficient targeting power allows us to remove the linear dependence in $n$, and to improve the quadratic dependence in $\epsilon^{-1}$ to linear. In this work, we introduce new technical ingredients and show that the number of targeted samples sufficient for learning an $\epsilon$-optimal auction is substantially smaller than the sample complexity of i.i.d. samples for the full spectrum of $\Delta \in [0, 1)$. Even with only mild targeting power, i.e., whenever $\Delta = o(1)$, our targeted sample complexity upper bounds are strictly smaller than the optimal sample complexity of i.i.d. samples.