Privacy and efficiency in personalized decision-making and recommendation
Abstract/Contents
- Abstract
- This dissertation explores challenges concerning data privacy and efficiency in decision-making and recommendation systems. The first chapter discusses difficulties of model learning under misspecification in contextual bandits for adaptive decision-making. We propose a reduction of contextual bandits to general-purpose heterogeneous treatment effect estimation, and demonstrate that this approach improves data efficiency and robustness to model misspecification in contextual bandits. The second chapter explores heterogeneous domain adaptation in decision-making in a data-constrained setting. Specifically, we consider learning personalized decision policies on historical observational data from heterogeneous data sources in a federated setting. A federated policy learning algorithm is proposed for this setting and a novel regret analysis is presented that highlights the extent of benefits of heterogeneity in policy learning. The final chapter focuses on data privacy in recommender systems. It proposes a novel approach for developing privacy-preserving large-scale recommender systems with synthetic data generated using differentially private language models. We demonstrate this approach ensures query-level privacy guarantees without sacrificing on recommendation quality, a key step towards the realization of secure and efficient data-driven personalized recommendation.
Description
Type of resource | text |
---|---|
Form | electronic resource; remote; computer; online resource |
Extent | 1 online resource. |
Place | California |
Place | [Stanford, California] |
Publisher | [Stanford University] |
Copyright date | 2023; ©2023 |
Publication date | 2023; 2023 |
Issuance | monographic |
Language | English |
Creators/Contributors
Author | Carranza, Aldo Gael |
---|---|
Degree supervisor | Athey, Susan |
Thesis advisor | Athey, Susan |
Thesis advisor | Wager, Stefan |
Thesis advisor | Weintraub, Gabriel |
Degree committee member | Wager, Stefan |
Degree committee member | Weintraub, Gabriel |
Associated with | Stanford University, School of Engineering |
Associated with | Stanford University, Institute for Computational and Mathematical Engineering |
Subjects
Genre | Theses |
---|---|
Genre | Text |
Bibliographic information
Statement of responsibility | Aldo Gael Carranza. |
---|---|
Note | Submitted to the Institute for Computational and Mathematical Engineering. |
Thesis | Thesis Ph.D. Stanford University 2023. |
Location | https://purl.stanford.edu/ky391xq2715 |
Access conditions
- Copyright
- © 2023 by Aldo Gael Carranza
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...