Abstract:With the rapid development of deep learning, AI systems are being used more in complex and important domains and necessitates the simultaneous fulfillment of multiple constraints: accurate, robust, and fair. Accuracy measures how well a DNN can generalize to new data. Robustness demonstrates how well the network can withstand minor perturbations without changing the results. Fairness focuses on treating different groups equally. This survey provides an overview of the triangular trade-off among robustness, accuracy, and fairness in neural networks. This trade-off makes it difficult for AI systems to achieve true intelligence and is connected to generalization, robustness, and fairness in deep learning. The survey explores these trade-offs and their relationships to adversarial examples, adversarial training, and fair machine learning. The trade-offs between accuracy and robustness, accuracy and fairness, and robustness and fairness have been studied to different extents. However, there is a lack of taxonomy and analysis of these trade-offs. The accuracy-robustness trade-off is inherent in Gaussian models, but it varies when classes are not closely distributed. The accuracy-fairness and robustness-fairness trade-offs have been assessed empirically, but their theoretical nature needs more investigation. This survey aims to explore the origins, evolution, influencing factors, and future research directions of these trade-offs.

What problem does this paper attempt to address?

### Problems the paper attempts to solve The paper "The Tri - Trade - off between Robustness, Accuracy, and Fairness in Deep Neural Networks: A Survey" aims to explore and summarize the triangular trade - off among robustness, accuracy, and fairness in deep neural networks. Specifically, the paper attempts to solve the following key problems: 1. **Trade - off between robustness and accuracy**: - Robustness refers to the ability of a model to keep its prediction results unchanged when facing minor perturbations. - Accuracy refers to the generalization ability of a model to new data. - The paper explores why, while improving the accuracy of a model, its robustness often declines, and vice versa. This trade - off relationship is inherent in Gaussian models, but it will be different in cases where the class distribution is not compact. 2. **Trade - off between robustness and fairness**: - Fairness refers to the ability of a model to maintain consistent treatment when dealing with different groups. - The paper studies how to ensure that the fairness of a model is not affected while improving its robustness. This trade - off relationship has been empirically evaluated, but its theoretical properties still need further study. 3. **Trade - off between accuracy and fairness**: - The paper discusses how to avoid making unfair predictions or decisions while improving the accuracy of a model. This trade - off relationship also requires more theoretical analysis. 4. **Challenges of comprehensive trade - off**: - The paper points out that in practical applications, especially in safety - critical fields, it is very difficult to simultaneously achieve high accuracy, high robustness, and high fairness. This triangular trade - off relationship makes it difficult for AI systems to reach a truly intelligent level. 5. **Deficiencies in existing research**: - Although there have been some studies on these trade - off relationships, there is a lack of a systematic classification and analysis framework. The paper aims to fill this gap and provide a comprehensive survey to help researchers better understand the origin, evolution, influencing factors, and future research directions of these trade - off relationships. Through the exploration of these problems, the paper hopes to provide researchers with a comprehensive perspective to better understand and solve the complex trade - off problems among robustness, accuracy, and fairness in deep neural networks.

The Triangular Trade-off between Robustness, Accuracy and Fairness in Deep Neural Networks: A Survey

Robustness of Deep Learning Models on Graphs: A Survey

There is Limited Correlation Between Coverage and Robustness for Deep Neural Networks

An Empirical Study on Correlation between Coverage and Robustness for Deep Neural Networks

RobustFair: Adversarial Evaluation through Fairness Confusion Directed Gradient Search

To be Robust or to be Fair: Towards Fairness in Adversarial Training

Fairness in Deep Learning: A Survey on Vision and Language Research

A Survey of Neural Network Robustness Assessment in Image Recognition

How Robust is your Fair Model? Exploring the Robustness of Diverse Fairness Strategies

FaiR-N: Fair and Robust Neural Networks for Structured Data

The Double-Edged Sword of Input Perturbations to Robust Accurate Fairness

SoK: Certified Robustness for Deep Neural Networks

A Tale of Two Cities: Data and Configuration Variances in Robust Deep Learning

Technical Challenges for Training Fair Neural Networks

To be Robust and to be Fair: Aligning Fairness with Robustness

An Empirical Study of Accuracy, Fairness, Explainability, Distributional Robustness, and Adversarial Robustness

Fault Tolerance of Neural Networks in Adversarial Settings

Survey on Fairness Notions and Related Tensions

NeuronFair

Debiasing Methods for Fairer Neural Models in Vision and Language Research: A Survey

Opportunities and Challenges in Deep Learning Adversarial Robustness: A Survey