An Efficient Deep Distribution Network for Bid Shading in First-Price Auctions

@article{Zhou2021AnED,
  title={An Efficient Deep Distribution Network for Bid Shading in First-Price Auctions},
  author={Tian Zhou and Hao He and Shengjun Pan and Niklas Karlsson and Bharatbhushan Shetty and Brendan Kitts and Djordje Gligorijevic and San Gultekin and Tingyu Mao and Junwei Pan and Jianlong Zhang and Aaron Flores},
  journal={Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery \& Data Mining},
  year={2021}
}
  • Tian Zhou, Hao He, Aaron Flores
  • Published 12 July 2021
  • Computer Science
  • Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining
Since 2019, most ad exchanges and sell-side platforms (SSPs), in the online advertising industry, shifted from second to first price auctions. Due to the fundamental difference between these auctions, demand-side platforms (DSPs) have had to update their bidding strategies to avoid bidding unnecessarily high and hence overpaying. Bid shading was proposed to adjust the bid price intended for second-price auctions, in order to balance cost and winning probability in a first-price auction setup… 

Figures and Tables from this paper

Learning to Bid with AuctionGym

Online advertising opportunities are sold through auctions, billions of times every day across the web. Advertisers who participate in those auctions need to decide on a bidding strategy: how much

Simultaneous Advertiser Profit and Ad Platform Revenue Maximization in Programmatic Advertising via Feedback Control

  • N. Karlsson
  • Business
    2022 American Control Conference (ACC)
  • 2022
The majority of online advertising is served through real-time bidding, and advertising campaigns are often defined as optimization problems. This paper deals with simultaneous advertiser profit and

Arbitrary Distribution Modeling with Censorship in Real-Time Bidding Advertising

TLDR
A novel loss function, Neighborhood Likelihood Loss (NLL), collaborating with a proposed framework, Arbitrary Distribution Modeling (ADM), to predict the winning price distribution under censorship with no pre-assumption required, demonstrating its great capability in modeling sophisticated price landscapes.

A Unified Solution to Constrained Bidding in Online Display Advertising

TLDR
This paper forms the demand as a constrained bidding problem, and deduce a unified optimal bidding function on behalf of an advertiser as Unified Solution to Constrained Bidding (USCB), and proposes a reinforcement learning method to dynamically adjust parameters to achieve the optimum.

References

SHOWING 1-10 OF 41 REFERENCES

Bid Shading by Win-Rate Estimation and Surplus Maximization

TLDR
A new win-rate based bid shading algorithm (WR) that does not rely on the minimum-bid-to-win feedback from a Sell-Side Platform (SSP) and uses a modified logistic regression to predict the profit from each possible shaded bid price.

Sold for more than you should have paid

  • 2018

Bid Shading in The Brave New World of First-Price Auctions

TLDR
A machine learning approach of modeling optimal bid shading for non-censored online first-price ad auctions is proposed and the results demonstrate the superiority and robustness of the new approach as compared to the existing approaches across a range of performance metrics.

A Unified Solution to Constrained Bidding in Online Display Advertising

TLDR
This paper forms the demand as a constrained bidding problem, and deduce a unified optimal bidding function on behalf of an advertiser as Unified Solution to Constrained Bidding (USCB), and proposes a reinforcement learning method to dynamically adjust parameters to achieve the optimum.

Adaptive Bid Shading Optimization of First-Price Ad Inventory

TLDR
This paper proposes an adaptive scheme for online learning of optimal bid shading that involves segmentation, a two-parametric nonlinear shading mechanism, and an online learning algorithm for parameter optimization.

DeepLight: Deep Lightweight Feature Interactions for Accelerating CTR Predictions in Ad Serving

TLDR
DeepLight is presented, a framework to accelerate the CTR predictions in three aspects: accelerate the model inference via explicitly searching informative feature interactions in the shallow component; prune redundant parameters at the inter-layer level in the DNN component; and prune the dense embedding vectors to make them sparse in the embedding matrix.

Predicting Different Types of Conversions with Multi-Task Learning in Online Advertising

TLDR
This paper forms conversion prediction as a multi-task learning problem, so that the prediction models for different types of conversions can be learned together, and proposes Multi-Task Field-weighted Factorization Machine (MT-FwFM) to solve these tasks jointly.

Bid Optimization by Multivariable Control in Display Advertising

TLDR
This paper converts the common case where advertisers aim to maximize the quantity of conversions, and set cost-per-click (CPC) as a KPI constraint, and leverage the primal-dual method to derive the optimal bidding strategy to address the applicability issue.

Auctions and Bidding

ONE PARTY TO AN EXCHANGE often knows something relevant to the transaction that the other party does not know. Such asymmetries of information are pervasive in economic activity: for example, in the

Deep Censored Learning of the Winning Price in the Real Time Bidding

TLDR
The main advantage of the proposed generalized deep learning model is to provide more flexibility to model the winning price and improve the performance in consideration of the possibly various winning price distributions and various model structures in practice.