Can Foundation Models Help Us Achieve Perfect Secrecy?

05/27/2022
by   Simran Arora, et al.
8

A key promise of machine learning is the ability to assist users with personal tasks. Because the personal context required to make accurate predictions is often sensitive, we require systems that protect privacy. A gold standard privacy-preserving system will satisfy perfect secrecy, meaning that interactions with the system provably reveal no additional private information to adversaries. This guarantee should hold even as we perform multiple personal tasks over the same underlying data. However, privacy and quality appear to be in tension in existing systems for personal tasks. Neural models typically require lots of training to perform well, while individual users typically hold a limited scale of data, so the systems propose to learn from the aggregate data of multiple users. This violates perfect secrecy and instead, in the last few years, academics have defended these solutions using statistical notions of privacy – i.e., the probability of learning private information about a user should be reasonably low. Given the vulnerabilities of these solutions, we explore whether the strong perfect secrecy guarantee can be achieved using recent zero-to-few sample adaptation techniques enabled by foundation models. In response, we propose FOCUS, a framework for personal tasks. Evaluating on popular privacy benchmarks, we find the approach, satisfying perfect secrecy, competes with strong collaborative learning baselines on 6 of 7 tasks. We empirically analyze the proposal, highlighting the opportunities and limitations across task types, and model inductive biases and sizes.

READ FULL TEXT
research
09/11/2021

A Privacy-Preserving Logistics Information System with Traceability

Logistics Information System (LIS) is an interactive system that provide...
research
01/03/2023

Recent Trends on Privacy-Preserving Technologies under Standardization at the IETF

End-users are concerned about protecting the privacy of their personal d...
research
02/12/2020

Fundamental Limits of Biometric Identification Systems with Strong Secrecy

The fundamental limits of biometric identification systems under a stron...
research
03/14/2020

Decentralized Policy-Based Private Analytics

We are increasingly surrounded by applications, connected devices, servi...
research
03/30/2020

Private Colocation Discovery: Taming the Coronavirus While Preserving Privacy

Successful containment of the Coronavirus pandemic rests on the ability ...
research
02/26/2018

Privacy Things: Systematic Approach to Privacy and Personal Identifiable Information

Defining privacy and related notions such as Personal Identifiable Infor...

Please sign up or login with your details

Forgot password? Click here to reset