Ge Liu

Title: Ge Liu and Privacy-Preserving Data Collaboration.

Introduction

Ge Liu is an inventor associated with Palo Alto, California, and Amazon Technologies, Inc. His patent work addresses ways to support collaboration across datasets while protecting sensitive information.

Latest Patents

Liu is associated with 1 patent: “Data Clean Room with Exportable, Privacy-Preserving Labels.” The invention describes systems, devices, and methods for enabling data clean rooms to export privacy-preserving labels. Code in the clean room can select and train a differential privacy machine-learning model, identify users of interest in a multi-party collaboration, and infer noisy labels for those users. The resulting exportable dataset contains noisy labels rather than private labels that could reveal sensitive information.

Career Highlights

The patent reflects Liu’s work on applying differential privacy and machine learning to data collaboration. Its approach is designed to make useful labels available for joint efforts without exporting the underlying private information.

Collaborations

Liu’s listed coworker is Shuai Tang. The patent’s multi-party framework also describes collaboration among parties contributing datasets to a data clean room.

Conclusion

Ge Liu’s patent focuses on enabling data-driven collaboration while limiting exposure of sensitive information through differential privacy and noisy, exportable labels.