Recovering from Privacy-Preserving Masking with Large Language Models

TL;DR

This paper explores using large language models to replace sensitive tokens in textual data, enabling privacy-preserving model adaptation with performance comparable to models trained on original data.

Contribution

It introduces LLM-based methods for token masking and empirically evaluates their effectiveness in privacy-preserving NLP model adaptation.

Findings

Models trained on obfuscated data perform comparably to those trained on original data.

LLM-based token substitution effectively balances privacy and model performance.

Empirical studies validate the proposed masking approaches across various datasets.

Abstract

Model adaptation is crucial to handle the discrepancy between proxy training data and actual users data received. To effectively perform adaptation, textual data of users is typically stored on servers or their local devices, where downstream natural language processing (NLP) models can be directly trained using such in-domain data. However, this might raise privacy and security concerns due to the extra risks of exposing user information to adversaries. Replacing identifying information in textual data with a generic marker has been recently explored. In this work, we leverage large language models (LLMs) to suggest substitutes of masked tokens and have their effectiveness evaluated on downstream language modeling tasks. Specifically, we propose multiple pre-trained and fine-tuned LLM-based approaches and perform empirical studies on various datasets for the comparison of these methods. Experimental results show that models trained on the obfuscation corpora are able to achieve comparable performance with the ones trained on the original data without privacy-preserving token masking.