Multi-channel Autobidding with Budget and ROI Constraints

Authors: Yuan Deng, Negin Golrezaei, Patrick Jaillet, Jason Cheuk Nam Liang, Vahab Mirrokni

ICML 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical we present an efficient learning algorithm that produces per-channel budgets whose resulting conversion approximates that of the global optimal problem. ... In Theorem 3.2 of Section 3, we show that solely optimizing for per-channel ROIs is inadequate... In contrast, in Theorem 3.3 and Corollary 3.4 we show solely optimizing for per-channel budgets allows advertisers to achieve the global optimal. ... our proposed algorithm is the first to handle such a setting... we obtain the main result Theorem 4.8 whose proof we detail in Appendix C.6
Researcher Affiliation Collaboration Yuan Deng 1 Negin Golrezaei 2 Patrick Jaillet 2 Jason Cheuk Nam Liang 2 Vahab Mirrokni 1 1Google Research 2Massachusetts Institute of Technology.
Pseudocode Yes Algorithm 1 SGD-UCB
Open Source Code No The paper does not provide any statement or link indicating the release of open-source code for the methodology described.
Open Datasets No The paper is theoretical and analyzes an algorithm; it does not mention training, validation, or test dataset splits in the context of empirical evaluation.
Dataset Splits No The paper is theoretical and analyzes an algorithm; it does not mention training, validation, or test dataset splits in the context of empirical evaluation.
Hardware Specification No The paper is theoretical and focuses on algorithm analysis, thus no hardware specifications for running experiments are mentioned.
Software Dependencies No The paper is theoretical and focuses on algorithm design and proofs, thus it does not mention specific software dependencies with version numbers.
Experiment Setup No The paper mentions parameters like 'step size η', 'discretization width δ', and 'dual variable upper bound CF' as part of the algorithm's analysis, but these are not presented as an 'experiment setup' in the context of running empirical tests or training models with specific hyperparameters.