Public Data Assisted Differentially Private In-Context Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Joo, Seongho, Koh, Hyukhun, Jung, Kyomin
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911152931340288
author Joo, Seongho
Koh, Hyukhun
Jung, Kyomin
author_facet Joo, Seongho
Koh, Hyukhun
Jung, Kyomin
contents In-context learning (ICL) in Large Language Models (LLMs) has shown remarkable performance across various tasks without requiring fine-tuning. However, recent studies have highlighted the risk of private data leakage through the prompt in ICL, especially when LLMs are exposed to malicious attacks. While differential privacy (DP) provides strong privacy guarantees, it often significantly reduces the utility of in-context learning (ICL). To address this challenge, we incorporate task-related public data into the ICL framework while maintaining the DP guarantee. Based on this approach, we propose a private in-context learning algorithm that effectively balances privacy protection and model utility. Through experiments, we demonstrate that our approach significantly improves the utility of private ICL with the assistance of public data. Additionally, we show that our method is robust against membership inference attacks, demonstrating empirical privacy protection.
format Preprint
id arxiv_https___arxiv_org_abs_2509_10932
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Public Data Assisted Differentially Private In-Context Learning
Joo, Seongho
Koh, Hyukhun
Jung, Kyomin
Artificial Intelligence
Computation and Language
In-context learning (ICL) in Large Language Models (LLMs) has shown remarkable performance across various tasks without requiring fine-tuning. However, recent studies have highlighted the risk of private data leakage through the prompt in ICL, especially when LLMs are exposed to malicious attacks. While differential privacy (DP) provides strong privacy guarantees, it often significantly reduces the utility of in-context learning (ICL). To address this challenge, we incorporate task-related public data into the ICL framework while maintaining the DP guarantee. Based on this approach, we propose a private in-context learning algorithm that effectively balances privacy protection and model utility. Through experiments, we demonstrate that our approach significantly improves the utility of private ICL with the assistance of public data. Additionally, we show that our method is robust against membership inference attacks, demonstrating empirical privacy protection.
title Public Data Assisted Differentially Private In-Context Learning
topic Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2509.10932