Private Label

Private label research focuses on protecting sensitive information, particularly labels in machine learning datasets, during collaborative training or data sharing. Current research emphasizes developing and analyzing attacks that infer private labels from seemingly anonymized data, such as aggregated gradients or intermediate model representations, across various model architectures including federated learning and split learning. This work is crucial for ensuring the privacy and security of sensitive data in collaborative machine learning applications and for developing effective defenses against label inference attacks. The ultimate goal is to enable secure and collaborative data utilization while maintaining strong privacy guarantees.

Papers

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

November 17, 2023

When a Language Question Is at Stake. A Revisited Approach to Label Sensitive Content
Stetsenko Daria
Offensive Language Detection Data Annotation Natural Language Question Private Label English Tweet

October 16, 2023

Label Differential Privacy via Aggregation
Anand Brahmbhatt, Rishi Saket, Shreyas Havaldar, Anshul Nasery, Aravindan Raghuveer
Label Noise Data Aggregation Label Proportion Private Label Bag Level DP Guarantee

September 18, 2023

Drawing the Same Bounding Box Twice? Coping Noisy Annotations in Object Detection with Repeated Labels
David Tschirschwitz, Christian Benz, Morris Florek, Henrik Norderhus, Benno Stein, Volker Rodehorst
Ground Truth Human Annotation Noisy Annotation Private Label

July 4, 2023

FREEDOM: Target Label & Source Data & Domain Information-Free Multi-Source Domain Adaptation for Unsupervised Personalization
Eunju Yang, Gyusang Cho, Chan-Hyun Youn
Domain Adaptation Internet Service Domain Multi Source Source Domain Multi Source Domain Adaptation Data Source Private Label Multi Source Free Domain Adaptation Free Personalization

April 20, 2023

Censoring chemical data to mitigate dual use risk
Quintina L. Campbell, Jonathan Herington, Andrew D. White
Deep Neural Network Machine Learning Application Chemical Data Censored Time to Event Private Label Dual Use

January 18, 2023

Label Inference Attack against Split Learning under Regression Setting
Shangyu Xie, Xin Yang, Yuanshun Yao, Tianyi Liu, Taiqing Wang, Jiankai Sun
Novel Regression Vertical Federated Learning Split Learning Label Inference Attack Private Label Learning Based Attack

January 1, 2023

Mutual Information Regularization for Vertical Federated Learning
Tianyuan Zou, Yang Liu, Ya-Qin Zhang
Mutual Information Vertical Federated Learning Label Inference Attack Private Label Mutual Information Regularization

March 10, 2022

Similarity-based Label Inference Attack against Training and Inference of Split Learning
Junlin Liu, Xinchen Lyu, Qimei Cui, Xiaofeng Tao
Training Data Differential Privacy Scientific Inference Split Learning Label Inference Attack Label Attack Private Label

March 4, 2022

Differentially Private Label Protection in Split Learning
Xin Yang, Jiankai Sun, Yuanshun Yao, Junyuan Xie, Chong Wang
Differential Privacy Split Learning Private Label

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

February 25, 2022

Does Label Differential Privacy Prevent Label Inference Attacks?
Ruihan Wu, Jin Peng Zhou, Kilian Q. Weinberger, Chuan Guo
Differential Privacy Private Machine Learning Label Inference Attack Private Label

January 20, 2022

Classic Graph Structural Features Outperform Factorization-Based Graph Embedding Methods on Community Labeling
Andrew Stolman, Caleb Levy, C. Seshadhri, Aneesh Sharma
Graph Drawing NCD Method Jina Embeddings Graph Representation Learning Graph Embeddings Graph Feature Private Label Community Detection Problem

November 25, 2021

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

Private Label

Papers

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

When a Language Question Is at Stake. A Revisited Approach to Label Sensitive Content

Label Differential Privacy via Aggregation

Drawing the Same Bounding Box Twice? Coping Noisy Annotations in Object Detection with Repeated Labels

FREEDOM: Target Label & Source Data & Domain Information-Free Multi-Source Domain Adaptation for Unsupervised Personalization

Censoring chemical data to mitigate dual use risk

Label Inference Attack against Split Learning under Regression Setting

Mutual Information Regularization for Vertical Federated Learning

Similarity-based Label Inference Attack against Training and Inference of Split Learning

Differentially Private Label Protection in Split Learning

Label Leakage and Protection from Forward Embedding in Vertical Federated Learning

Does Label Differential Privacy Prevent Label Inference Attacks?

Classic Graph Structural Features Outperform Factorization-Based Graph Embedding Methods on Community Labeling

ExPLoit: Extracting Private Labels in Split Learning