Label Leakage

Label leakage refers to the unintended exposure of sensitive training labels during the process of machine learning model training, particularly in privacy-preserving techniques like federated learning. Current research focuses on identifying and mitigating label leakage vulnerabilities in various model architectures, including tree-based models and those employing secure aggregation, through methods such as differential privacy and regularization techniques. Understanding and addressing label leakage is crucial for ensuring the privacy and security of sensitive data used in machine learning, impacting the development of trustworthy and ethically sound AI systems across diverse applications.

Papers

October 14, 2024

A few-shot Label Unlearning in Vertical Federated Learning
Hanlin Gu, Hong Xi Tae, Chee Seng Chan, Lixin Fan
Vertical Federated Learning Gradient Ascent Label Leakage

October 11, 2024

Training on Fake Labels: Mitigating Label Leakage in Split Learning via Secure Dimension Transformation
Yukun Jiang, Peiran Wang, Chengguo Lin, Ziyue Huang, Yong Cheng
Training Data Split Learning Preserving Transformation Model Splitting Label Inference Attack Label Leakage

June 22, 2024

Breaking Secure Aggregation: Label Leakage from Aggregated Gradients in Federated Learning
Zhibo Wang, Zhiwei Chang, Jiahui Hu, Xiaoyi Pang, Jiacheng Du, Yongle Chen, Kui Ren
Secure Aggregation Gradient Inversion Attack Aggregated Gradient Label Inference Attack Private Label Label Leakage

July 19, 2023

Eliminating Label Leakage in Tree-Based Vertical Federated Learning
Hideaki Takahashi, Jingjing Liu, Yang Liu
Differential Privacy Vertical Federated Learning Training Label Label Inference Attack Label Leakage

February 24, 2023

Don't be fooled: label leakage in explanation methods and the importance of their quantitative evaluation
Neil Jethani, Adriel Saporta, Rajesh Ranganath
Importance Aware Explanation Method Feature Attribution Method Quantitative Evaluation Label Leakage

March 2, 2022

Label Leakage and Protection from Forward Embedding in Vertical Federated Learning
Jiankai Sun, Xin Yang, Yuanshun Yao, Chong Wang
Differential Privacy Vertical Federated Learning PRIvacy Leakage Label Attack Private Label Label Leakage

December 16, 2021

PECO: Examining Single Sentence Label Leakage in Natural Language Inference Datasets through Progressive Evaluation of Cluster Outliers
Michael Saxon, Xinyi Wang, Wenda Xu, William Yang Wang
Natural Language Inference Parallel Text Label Leakage

November 25, 2021

ExPLoit: Extracting Private Labels in Split Learning
Sanjay Kariyappa, Moinuddin K Qureshi
Split Learning Adversarial Input Private Label Label Leakage

Label Leakage

Papers

A few-shot Label Unlearning in Vertical Federated Learning

Training on Fake Labels: Mitigating Label Leakage in Split Learning via Secure Dimension Transformation

Breaking Secure Aggregation: Label Leakage from Aggregated Gradients in Federated Learning

Eliminating Label Leakage in Tree-Based Vertical Federated Learning

Don't be fooled: label leakage in explanation methods and the importance of their quantitative evaluation

Label Leakage and Protection from Forward Embedding in Vertical Federated Learning

PECO: Examining Single Sentence Label Leakage in Natural Language Inference Datasets through Progressive Evaluation of Cluster Outliers

ExPLoit: Extracting Private Labels in Split Learning