Robustness Bias

Robustness bias refers to the uneven susceptibility of machine learning models, particularly deep learning models like Transformers and neural networks, to adversarial attacks or noisy inputs across different data subgroups. Current research focuses on identifying and mitigating this bias through methods like dataset diversification, improved model architectures (e.g., RBFormer), and novel robustness metrics that go beyond simple worst-case analysis (e.g., average-case robustness). Understanding and addressing robustness bias is crucial for building trustworthy and fair AI systems, ensuring consistent performance across diverse populations and preventing discriminatory outcomes in real-world applications.

Papers

November 29, 2023

ROBBIE: Robust Bias Evaluation of Large Generative Language Models
David Esiobu, Xiaoqing Tan, Saghar Hosseini, Megan Ung, Yuchen Zhang, Jude Fernandes, Jane Dwivedi-Yu, Eleonora Presani, Adina Williams, Eric Michael Smith
Generative Large Language Model Model Bias Robustness Bias Prompt Bias

September 23, 2023

RBFormer: Improve Adversarial Robustness of Transformer by Robust Bias
Hao Cheng, Jinhao Duan, Hui Li, Lyutianyang Zhang, Jiahang Cao, Ping Wang, Jize Zhang, Kaidi Xu, Renjing Xu
Transformer Based Adversarial Robustness Convolutional Structure Transformer Based Structure Robustness Bias Geometry Biased Transformer

July 26, 2023

Characterizing Data Point Vulnerability via Average-Case Robustness
Tessa Han, Suraj Srinivas, Himabindu Lakkaraju
Adversarial Attack Native Robustness Adversarial Robustness Software Vulnerability Robustness Bias Average Case Robustness

May 22, 2023

Keeping Up with the Language Models: Systematic Benchmark Extension for Bias Auditing
Ioana Baldini, Chhavi Yadav, Manish Nagireddy, Payel Das, Kush R. Varshney
Language Model Full Model Model Bias Bias Benchmark NLI Datasets Bias Auditing Robustness Bias

March 13, 2023

Enhancing LiDAR performance: Robust De-skewing Exclusively Relying on Range Measurements
Omar Salem, Emanuele Giacomini, Leonardo Brizi, Luca Di Giammarino, Giorgio Grisetti
Light Detection and Ranging LiDAR Measurement Robustness Bias Range Measurement

February 24, 2023

UnbiasedNets: A Dataset Diversification Framework for Robustness Bias Alleviation in Neural Networks
Mahum Naseer, Bharath Srinivas Prabakaran, Osman Hasan, Muhammad Shafique
Neural Network Balanced Dataset Robustness Bias Dataset Multiplicity

November 12, 2022

A Generalized Doubly Robust Learning Framework for Debiasing Post-Click Conversion Rate Prediction
Quanyu Dai, Haoxuan Li, Peng Wu, Zhenhua Dong, Xiao-Hua Zhou, Rui Zhang, Rui zhang, Jie Sun
Generalization Performance Generalization Bound Doubly Robust Conversion Rate Prediction Robustness Bias Post Click

June 7, 2022

Certifying Data-Bias Robustness in Linear Regression
Anna P. Meyer, Aws Albarghouthi, Loris D'Antoni
Training Data Linear Regression Modeling Assumption Linear Model Robustness Bias

May 5, 2022

Holistic Approach to Measure Sample-level Adversarial Vulnerability and its Utility in Building Trustworthy Systems
Gaurav Kumar Nayak, Ruchit Rawal, Rohit Lal, Himanshu Patil, Anirban Chakraborty
Adversarial Attack Adversarial Training Task Utility Adversarial Vulnerability Holistic Approach Trustworthy Computing Robustness Bias

Robustness Bias

Papers

ROBBIE: Robust Bias Evaluation of Large Generative Language Models

RBFormer: Improve Adversarial Robustness of Transformer by Robust Bias

Characterizing Data Point Vulnerability via Average-Case Robustness

Keeping Up with the Language Models: Systematic Benchmark Extension for Bias Auditing

Enhancing LiDAR performance: Robust De-skewing Exclusively Relying on Range Measurements

UnbiasedNets: A Dataset Diversification Framework for Robustness Bias Alleviation in Neural Networks

A Generalized Doubly Robust Learning Framework for Debiasing Post-Click Conversion Rate Prediction

Certifying Data-Bias Robustness in Linear Regression

Holistic Approach to Measure Sample-level Adversarial Vulnerability and its Utility in Building Trustworthy Systems