Optimizing Online Advertising with Multi-Armed Bandits: Mitigating the Cold Start Problem under Auction Dynamics

TL;DR

This paper introduces a UCB-like algorithm for multi-armed bandits tailored to auction-based online advertising, effectively addressing the cold start problem by balancing exploration and exploitation to improve long-term revenue.

Contribution

The paper presents a novel algorithm combining theoretical regret bounds with practical experiments, specifically designed for auction-based ad systems to mitigate cold start issues.

Findings

The algorithm reduces cold start effects in ad CTR estimation.

Experimental results on real data validate the method's effectiveness.

The approach enhances long-term profitability of ad platforms.

Abstract

Online advertising platforms often face a common challenge: the cold start problem. Insufficient behavioral data (clicks) makes accurate click-through rate (CTR) forecasting of new ads challenging. CTR for "old" items can also be significantly underestimated due to their early performance influencing their long-term behavior on the platform. The cold start problem has far-reaching implications for businesses, including missed long-term revenue opportunities. To mitigate this issue, we developed a UCB-like algorithm under multi-armed bandit (MAB) setting for positional-based model (PBM), specifically tailored to auction pay-per-click systems. Our proposed algorithm successfully combines theory and practice: we obtain theoretical upper estimates of budget regret, and conduct a series of experiments on synthetic and real-world data that confirm the applicability of the method on the real platform. In addition to increasing the platform's long-term profitability, we also propose a mechanism for maintaining short-term profits through controlled exploration and exploitation of items.