Robust Auction Design in the Auto-bidding World
Authors: Santiago Balseiro, Yuan Deng, Jieming Mao, Vahab Mirrokni, Song Zuo
NeurIPS 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Finally, we complement our theoretical observations with an empirical study confirming the effectiveness of these ideas using data from online advertising auctions. In this section, we derive semi-synthetic data from real auction data of a major search engine to validate our theoretical findings concerning VCG auctions. Finally, we conduct empirical analyses with semi-synthetic data and validate our theoretical findings for the performance of our mechanisms in VCG auctions. |
| Researcher Affiliation | Collaboration | Santiago Balseiro Columbia University and Google srb2155@columbia.edu Dengyuan Li Google dengyuan@google.com Jieming Mao Google maojm@google.com Vahab Mirrokni Google mirrokni@google.com Song Zuo Google szuo@google.com |
| Pseudocode | No | The paper does not contain any structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide any concrete access information for open-source code for the methodology described. |
| Open Datasets | No | We derive semi-synthetic data from real auction data of a major search engine. Instead of using real return on ad spend targets for value maximizers, we generate artificial targets to exclude any practical noises from the real system. The paper does not provide specific access information (link, DOI, citation with authors/year) for this semi-synthetic dataset. |
| Dataset Splits | No | The paper does not provide specific train/validation/test dataset splits. It mentions simulating for '25 iterations' and '10 runs of repeated experiments' but no partitioning of data for validation purposes. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., GPU/CPU models, memory amounts, or cloud instance types) used for running its experiments. It mentions 'online advertising auctions' and 'major search engine' data, implying a large computational setup, but no specifics. |
| Software Dependencies | No | The paper does not provide specific software dependencies with version numbers. |
| Experiment Setup | Yes | We first pre-train the (uniform) bid multipliers δ for value maximizers in 25 iterations without reserve prices and boosts to obtain an equilibrium as a starting point. We then simulate the response of value maximizers by gradient descent on their bid multipliers in log space until convergence [Aggarwal et al., 2019, Nesterov, 2013]. Formally, let δi,t be the bid multiplier for value maximizer i in iteration t... log δi,t+1 = (1 t) log δi,t + t log Weli,t/Revi,t , where t 2 (0, 1) is a properly chosen learning rate for t-th iteration. After obtaining a starting point, we simulate another 25 iterations for auctions with reserve prices and/or boosts. We conduct 10 runs of repeated experiments. |