Sensitive Attribute

Sensitive attributes, such as race, gender, or age, pose significant challenges in machine learning due to their potential to introduce bias and privacy violations. Current research focuses on mitigating these issues through various techniques, including data pre-processing methods (e.g., synthetic data generation, attribute disentanglement), model training modifications (e.g., adversarial training, fairness-aware regularization), and post-processing algorithms (e.g., demographic parity constraints). These efforts aim to create fairer and more privacy-preserving AI systems, impacting fields like healthcare, finance, and criminal justice by reducing discriminatory outcomes and protecting sensitive user information.

Papers

August 26, 2022

Selective manipulation of disentangled representations for privacy-aware facial image processing
Sander De Coninck, Wei-Cheng Wang, Sam Leroux, Pieter Simoens
Meaningful Representation Privacy Preservation Sensitive Attribute Image Sensor Selective Engagement

August 21, 2022

Inferring Sensitive Attributes from Model Explanations
Vasisht Duddu, Antoine Boutet
Model Explanation Sensitive Attribute Attribute Inference Attack Monotone Adversary

July 25, 2022

Estimating and Controlling for Equalized Odds via Sensitive Attribute Predictors
Beepul Bharti, Paul Yi, Jeremias Sulam
Machine Learning Model Procedural Fairness Estimation Task Sensitive Attribute Relative Probability Equalized Odds Fairness Issue

July 18, 2022

When Fairness Meets Privacy: Fair Classification with Semi-Private Sensitive Attributes
Canyu Chen, Yueqing Liang, Xiongxiao Xu, Shangyu Xie, Ashish Kundu, Ali Payani, Yuan Hong, Kai Shu
Procedural Fairness Privacy Policy Local Differential Privacy Sensitive Attribute Fair Classification Fair Prediction

June 30, 2022

Discrimination in machine learning algorithms
Roberta Pappadà, Francesco Pauli
Practical Algorithm Machine Learning Algorithm Sensitive Attribute Group DIscrimination Credit Scoring Potential Bias

May 31, 2022

A Reduction to Binary Approach for Debiasing Multiclass Datasets
Ibrahim Alabdulmohsin, Jessica Schrouff, Oluwasanmi Koyejo
Constructive Reduction Sensitive Attribute Multiclass Classification Multi Class

May 12, 2022

Fair NLP Models with Differentially Private Text Encoders
Gaurav Maheshwari, Pascal Denis, Mikaela Keller, Aurélien Bellet
Differential Privacy Adversarial Training Text Representation Sensitive Attribute Private Training NLP Fairness

March 30, 2022

Learning Fair Models without Sensitive Attributes: A Generative Approach
Huaisheng Zhu, Enyan Dai, Hui Liu, Suhang Wang
Generative Approach Probabilistic Generative Model Sensitive Attribute Fair Model FAir Classifier

February 4, 2022

Fairness for Text Classification Tasks with Identity Information Data Augmentation Methods
Mohit Wadhwa, Mohan Bhambhani, Ashvini Jindal, Uma Sawant, Ramanujam Madhavan
Data Augmentation Procedural Fairness Text Classification Task Sensitive Attribute Counterfactual Fairness Counterfactual Example

January 23, 2022

Are Your Sensitive Attributes Private? Novel Model Inversion Attribute Inference Attacks on Classification Models
Shagufta Mehnaz, Sayanton V. Dibbo, Ehsanul Kabir, Ninghui Li, Elisa Bertino
Classification Model Model Inversion Attack Sensitive Attribute Score Based Attribute Inference Attack Label Only Model Inversion Attack

January 15, 2022

FairIF: Boosting Fairness in Deep Learning via Influence Functions with Validation Set Sensitive Attributes
Haonan Wang, Ziwei Wu, Jingrui He
Deep Learning Gradient Descent Procedural Fairness Influence Function Sensitive Attribute Two Stage Training Sample Weight

Sensitive Attribute

Papers

Selective manipulation of disentangled representations for privacy-aware facial image processing

Inferring Sensitive Attributes from Model Explanations

Estimating and Controlling for Equalized Odds via Sensitive Attribute Predictors

When Fairness Meets Privacy: Fair Classification with Semi-Private Sensitive Attributes

Discrimination in machine learning algorithms

A Reduction to Binary Approach for Debiasing Multiclass Datasets

Fair NLP Models with Differentially Private Text Encoders

Learning Fair Models without Sensitive Attributes: A Generative Approach

Fairness for Text Classification Tasks with Identity Information Data Augmentation Methods

Are Your Sensitive Attributes Private? Novel Model Inversion Attribute Inference Attacks on Classification Models

FairIF: Boosting Fairness in Deep Learning via Influence Functions with Validation Set Sensitive Attributes