Class Accuracy

Class accuracy, the performance of a classification model on individual classes, is a crucial aspect of model evaluation, going beyond overall accuracy metrics. Current research emphasizes addressing class imbalance, where some classes are significantly under- or over-represented, leading to biased performance estimates and inaccurate predictions. This involves developing novel algorithms and adapting existing methods, such as ensemble techniques and nonlinear integer programming, to improve class-specific accuracy and reduce bias. Improved understanding and mitigation of class-level accuracy issues are vital for building robust and reliable machine learning models across diverse applications, particularly in sensitive domains like medicine and security.

Papers

May 23, 2024

A Neighbor-Searching Discrepancy-based Drift Detection Scheme for Learning Evolving Data
Feng Gu, Jie Lu, Zhen Fang, Kun Wang, Guangquan Zhang
Drift Detection Concept Drift Detection Class Accuracy

May 13, 2024

COBias and Debias: Minimizing Language Model Pairwise Accuracy Bias via Nonlinear Integer Programming
Ruixi Lin, Yang You
Large Language Model Language Model Self Debiasing Integer Programming Class Accuracy

December 7, 2023

Understanding the Detrimental Class-level Effects of Data Augmentation
Polina Kirichenko, Mark Ibrahim, Randall Balestriero, Diane Bouchacourt, Ramakrishna Vedantam, Hamed Firooz, Andrew Gordon Wilson
Data Augmentation Supervised ImageNet Class Imbalance Class Accuracy

September 16, 2023

Tightening Classification Boundaries in Open Set Domain Adaptation through Unknown Exploitation
Lucas Fernando Alvarenga e Silva, Nicu Sebe, Jurandy Almeida
Convolutional Neural Network Domain Shift Open Set Domain Adaptation Adversarial Feature Class Accuracy

June 12, 2023

Machine Learning Approach on Multiclass Classification of Internet Firewall Log Files
Md Habibur Rahman, Taminul Islam, Md Masum Rana, Rehnuma Tasnim, Tanzina Rahman Mona, Md. Mamun Sakib
Machine Learning Approach Multiclass Classification Class Accuracy Web Application Firewall

March 24, 2023

Leveraging Old Knowledge to Continually Learn New Classes in Medical Images
Evelyn Chee, Mong Li Lee, Wynne Hsu
LeArning Abstract Medical Image Catastrophic Forgetting Medical Datasets Class Incremental Continual Learning Class Accuracy

February 21, 2023

Classy Ensemble: A Novel Ensemble Algorithm for Classification
Moshe Sipper
Classification Code Diverse Ensemble Novel Ensemble Ensemble Clustering Class Accuracy

November 16, 2022

Interclass Prototype Relation for Few-Shot Segmentation
Atsuro Okazawa
Segmentation Based Approach Segmentation Performance Shot Segmentation Class Accuracy Intra Domain Category Aware Prototype

September 20, 2022

Learning Acceptance Regions for Many Classes with Anomaly Detection
Zhou Wang, Xingye Qiao
Anomaly Detection Prediction Set Class Specific Set Valued Class Accuracy Classification Paradigm Learnable Region

July 20, 2022

Estimating Model Performance under Domain Shifts with Class-Specific Confidence Scores
Zeju Li, Konstantinos Kamnitsas, Mobarakol Islam, Chen Chen, Ben Glocker
Domain Shift Model Performance Model Confidence Estimation Performance Class Calibration Class Accuracy Accuracy Estimation Class Level Confidence

Class Accuracy

Papers

A Neighbor-Searching Discrepancy-based Drift Detection Scheme for Learning Evolving Data

COBias and Debias: Minimizing Language Model Pairwise Accuracy Bias via Nonlinear Integer Programming

Understanding the Detrimental Class-level Effects of Data Augmentation

Tightening Classification Boundaries in Open Set Domain Adaptation through Unknown Exploitation

Machine Learning Approach on Multiclass Classification of Internet Firewall Log Files

Leveraging Old Knowledge to Continually Learn New Classes in Medical Images

Classy Ensemble: A Novel Ensemble Algorithm for Classification

Interclass Prototype Relation for Few-Shot Segmentation

Learning Acceptance Regions for Many Classes with Anomaly Detection

Estimating Model Performance under Domain Shifts with Class-Specific Confidence Scores