Prediction Bias

Prediction bias, the systematic error in a model's predictions due to biases in training data or model architecture, is a significant concern across machine learning applications. Current research focuses on identifying and mitigating these biases in various contexts, including federated learning, long-tailed object detection, and natural language processing, often employing Bayesian approaches, resampling techniques, or debiasing methods applied to model outputs (e.g., logits). Understanding and addressing prediction bias is crucial for ensuring fairness, reliability, and trustworthiness in AI systems, impacting fields ranging from healthcare and finance to education and social sciences.

Papers

November 14, 2024

Everyone deserves their voice to be heard: Analyzing Predictive Gender Bias in ASR Models Applied to Dutch Speech Data
Rik Raes, Saskia Lensink, Mykola Pechenizkiy
Automatic Speech Recognition Absolute Stance Bias Human VOICE ASR Model Bilingual Data Gender Classification Prediction Bias Performance Disparity Common Voice

September 6, 2024

Generalization vs. Memorization in the Presence of Statistical Biases in Transformers
John Mitros, Damien Teney
Strong Generalization Transformer Megatron Decepticons Transformer Model Distribution Data Speech Presence Limited Memorization Prediction Bias

May 30, 2024

Estimating before Debiasing: A Bayesian Approach to Detaching Prior Bias in Federated Semi-Supervised Learning
Guogang Zhu, Xuefeng Liu, Xinghao Wu, Shaojie Tang, Chao Tang, Jianwei Niu, Hao Su
Semi Supervised Estimation Task Bayesian Approach Prediction Bias

May 29, 2024

MANO: Exploiting Matrix Norm for Unsupervised Accuracy Estimation Under Distribution Shifts
Renchunzi Xie, Ambroise Odonnat, Vasilii Feofanov, Weijian Deng, Jianfeng Zhang, Bo An
Self Supervised Distribution Shift Generalization Performance P$ Norm Unsupervised Algorithm Score Estimation Prediction Bias

February 8, 2024

Hidden in Plain Sight: Undetectable Adversarial Bias Attacks on Vulnerable Patient Populations
Pranav Kulkarni, Andrew Chan, Nithya Navarathna, Skylar Chan, Paul H. Yi, Vishwa S. Parekh
Hidden Knowledge Plain Sight Prediction Bias Vulnerable Population Bias Neuron Adversarial Bias

January 29, 2024

Rectify the Regression Bias in Long-Tailed Object Detection
Ke Zhu, Minghao Fu, Jie Shao, Tianyu Liu, Jianxin Wu
Long Tailed Class Distribution Classification Head Long Tailed Object Detection Prediction Bias

November 14, 2023

Selecting Shots for Demographic Fairness in Few-Shot Learning with Large Language Models
Carlos Aguirre, Kuleen Sasse, Isabel Cachola, Mark Dredze
LeArning Abstract Model Fairness Fairness Evaluation Prediction Bias Demographic Fairness Natural Language Processing Classifier

October 9, 2023

Efficient Hybrid Oversampling and Intelligent Undersampling for Imbalanced Big Data Classification
Carla Vairetti, José Luis Assadi, Sebastián Maldonado
Imbalanced Classification Large Datasets Oversampling Method Prediction Bias SMOTE Integrated Resampling Strategy

September 29, 2023

Assessing Look-Ahead Bias in Stock Return Predictions Generated By GPT Sentiment Analysis
Paul Glasserman, Caden Lin
Large Language Model Absolute Stance Bias Stock Return Prediction Prediction Bias Publication Bias

May 24, 2023

Is Your Model "MADD"? A Novel Metric to Evaluate Algorithmic Fairness for Predictive Student Models
Mélina Verger, Sébastien Lallé, François Bouchet, Vanda Luengo
Full Model Algorithmic Fairness Fairness Metric Student Model Novel Metric Fair Predictor Prediction Bias

May 11, 2023

Think Twice: Measuring the Efficiency of Eliminating Prediction Shortcuts of Question Answering Models
Lukáš Mikula, Michal Štefánik, Marek Petrovič, Petr Sojka
High Efficiency Adversarial Sample Question Answering Model Biased Feature Prediction Bias Prediction Speed

March 23, 2023

Fairness-guided Few-shot Prompting for Large Language Models
Huan Ma, Changqing Zhang, Yatao Bian, Lemao Liu, Zhirui Zhang, Peilin Zhao, Shu Zhang, Huazhu Fu, Qinghua Hu, Bingzhe Wu
Large Language Model Context Learning Different Context Prediction Bias

February 23, 2023

Auditing for Spatial Fairness
Dimitris Sacharidis, Giorgos Giannopoulos, George Papastefanatos, Kostas Stefanidis
Algorithmic Fairness Audit Evidence Algorithmic Decision Prediction Bias Spatial Fairness

November 21, 2022

Parametric Classification for Generalized Category Discovery: A Baseline Study
Xin Wen, Bingchen Zhao, Xiaojuan Qi
Baseline Result Generalized Category Discovery Non Parametric Prediction Bias Parametric Learning

May 31, 2022

Social Bias Meets Data Bias: The Impacts of Labeling and Measurement Errors on Fairness Criteria
Yiqiao Liao, Parinaz Naghizadeh
Social Bias Dataset Bias Manual Label Biased Data Fairness Criterion Prediction Bias Measurement Error

February 25, 2022

Language technology practitioners as language managers: arbitrating data bias and predictive bias in ASR
Nina Markl, Stephen Joseph McNulty
Automatic Speech Recognition Speech Recognition System Algorithmic Bias ASR System Language Technology Prediction Bias

January 30, 2022

Debiased-CAM to mitigate systematic error with faithful visual explanations of machine learning
Wencan Zhang, Mariella Dimiccoli, Brian Y. Lim
Machine Learning Saliency Map Model Explanation Visual Explanation Faithful Explanation Prediction Bias Systematic Error

Prediction Bias

Papers

Everyone deserves their voice to be heard: Analyzing Predictive Gender Bias in ASR Models Applied to Dutch Speech Data

Generalization vs. Memorization in the Presence of Statistical Biases in Transformers

Estimating before Debiasing: A Bayesian Approach to Detaching Prior Bias in Federated Semi-Supervised Learning

MANO: Exploiting Matrix Norm for Unsupervised Accuracy Estimation Under Distribution Shifts

Hidden in Plain Sight: Undetectable Adversarial Bias Attacks on Vulnerable Patient Populations

Rectify the Regression Bias in Long-Tailed Object Detection

Selecting Shots for Demographic Fairness in Few-Shot Learning with Large Language Models

Efficient Hybrid Oversampling and Intelligent Undersampling for Imbalanced Big Data Classification

Assessing Look-Ahead Bias in Stock Return Predictions Generated By GPT Sentiment Analysis

Is Your Model "MADD"? A Novel Metric to Evaluate Algorithmic Fairness for Predictive Student Models

Think Twice: Measuring the Efficiency of Eliminating Prediction Shortcuts of Question Answering Models

Fairness-guided Few-shot Prompting for Large Language Models

Auditing for Spatial Fairness

Parametric Classification for Generalized Category Discovery: A Baseline Study

Social Bias Meets Data Bias: The Impacts of Labeling and Measurement Errors on Fairness Criteria

Language technology practitioners as language managers: arbitrating data bias and predictive bias in ASR

Debiased-CAM to mitigate systematic error with faithful visual explanations of machine learning