Fairness in graph-based semi-supervised learning

Article Computer Science, Artificial Intelligence

Fair Representation: Guaranteeing Approximate Multiple Group Fairness for Unknown Tasks

Xudong Shen et al.

Summary: This study investigates whether fair representation can ensure fairness for unknown tasks and multiple fairness notions. It is found that fair representation guarantees fairness for an important subset of tasks where the representation is discriminative. Experimental results show that the learned representation indeed leads to fairer downstream predictions that are unknown during the representation learning stage.

IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE (2023)

添加到收藏夹

Article Computer Science, Artificial Intelligence

Fairness in Semi-Supervised Learning: Unlabeled Data Help to Reduce Discrimination

Tao Zhang et al.

Summary: This paper explores the use of semi-supervised learning to address fairness issues in machine learning, including predicting labels for unlabeled data, resampling to obtain multiple fair datasets, and using ensemble learning to improve accuracy and reduce discrimination. Theoretical analysis and experiments demonstrate that this method achieves a better trade-off between accuracy and fairness.

IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING (2022)

添加到收藏夹

Article Computer Science, Artificial Intelligence

Dual constraints and adversarial learning for fair recommenders

Haifeng Liu et al.

Summary: Recommender systems have a profound impact on people's lifestyles, but fairness problems have been identified. The presence of sensitive information in user behavior data leads to unfairness. To address this, a fairness-aware recommender model with dual fairness constraints is proposed, utilizing an adversarial graph neural network and fairness constraints to improve the fairness of recommendations.

KNOWLEDGE-BASED SYSTEMS (2022)

添加到收藏夹

Proceedings Paper Computer Science, Artificial Intelligence

Learning Fair Node Representations with Graph Counterfactual Fairness

Jing Ma et al.

Summary: This study aims to eliminate biases in machine learning models regarding sensitive attributes, such as race and gender. It introduces a novel fairness notion - graph counterfactual fairness, which uses counterfactual data augmentation to learn node representations and reduce biases in predictions.

WSDM'22: PROCEEDINGS OF THE FIFTEENTH ACM INTERNATIONAL CONFERENCE ON WEB SEARCH AND DATA MINING (2022)

添加到收藏夹

Article Computer Science, Artificial Intelligence

A Comprehensive Survey on Graph Neural Networks

Zonghan Wu et al.

Summary: This article provides a comprehensive overview of graph neural networks (GNNs) in data mining and machine learning fields. It discusses the taxonomy of GNNs, their applications, and summarizes open-source codes, benchmark data sets, and model evaluation. The article also proposes potential research directions in this rapidly growing field.

IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS (2021)

添加到收藏夹

Article Computer Science, Theory & Methods

A Survey on Bias and Fairness in Machine Learning

Ninareh Mehrabi et al.

Summary: With the widespread use of AI systems in everyday life, fairness in design has become crucial. Researchers have developed methods to address biases in different subdomains and established a taxonomy for fairness definitions. Existing work shows biases in AI applications, and researchers are working on solutions to mitigate these problems.

ACM COMPUTING SURVEYS (2021)

添加到收藏夹

Proceedings Paper Computer Science, Artificial Intelligence

Persistent Anti-Muslim Bias in Large Language Models

Abubakar Abid et al.

Summary: Research found that the GPT-3 model exhibits anti-Muslim bias, which is varied and severe, and can be partially alleviated with positive text prompts.

AIES '21: PROCEEDINGS OF THE 2021 AAAI/ACM CONFERENCE ON AI, ETHICS, AND SOCIETY (2021)

添加到收藏夹

Proceedings Paper Computer Science, Artificial Intelligence

Fair Bayesian Optimization

Valerio Perrone et al.

Summary: The article introduces a general constrained Bayesian optimization framework for optimizing the performance of machine learning models while enforcing fairness constraints. By operating solely on hyperparameters, accurate and fair solutions can be obtained. Compared to specialized fairness techniques, this method demonstrates competitiveness in practice.

AIES '21: PROCEEDINGS OF THE 2021 AAAI/ACM CONFERENCE ON AI, ETHICS, AND SOCIETY (2021)

添加到收藏夹

Proceedings Paper Computer Science, Hardware & Architecture