Augmented Class

Augmented Class Learning (ACL) addresses the challenge of machine learning models encountering previously unseen classes during testing, a common occurrence in real-world applications. Current research focuses on developing unbiased risk estimators, often incorporating unlabeled data to estimate the distribution of these new classes and adapting existing loss functions for robust performance. These methods aim to improve the generalization ability of classifiers trained on limited labeled data, thereby enhancing the reliability and adaptability of machine learning systems in dynamic environments. The impact of this research lies in creating more robust and practical machine learning solutions for various domains where new classes are expected to emerge over time.

Papers

September 29, 2024

An Unbiased Risk Estimator for Partial Label Learning with Augmented Classes
Jiayu Hu, Senlin Shu, Beibei Li, Tao Xiang, Zhongshi He
Partial Label Learning Regularization Penalty Unbiased Risk Estimator Augmented Class

February 16, 2024

Collaborative Learning with Different Labeling Functions
Yuyang Deng, Mingda Qiao
Multi Task Learning Efficient Learning Collaborative Learning Sample Efficient PAC Learning Labeling Function Augmented Class

June 12, 2023

A Generalized Unbiased Risk Estimator for Learning with Augmented Classes
Senlin Shu, Shuo He, Haobo Wang, Hongxin Wei, Tao Xiang, Lei Feng
Loss Function Multi Class Classification Regularization Penalty Unbiased Risk Estimator Augmented Class

March 9, 2023

$\pi$-augmented pregroups and applications to linguistics
Valentin Boboc
Financial Application Machine Permutation Linguistic Study Harm Free Pronoun Use Algebraic Structure Data Normalisation Augmented Class

November 19, 2022

Complementary Labels Learning with Augmented Classes
Zhongnian Li, Jian Zhang, Mengting Xu, Xinzheng Xu, Daoqiang Zhang
Supervised Learning Complementary Label Complementary Label Learning Augmented Class

July 27, 2022

Learning from Positive and Unlabeled Data with Augmented Classes
Zhongnian Li, Liutao Yang, Zhongchen Ma, Tongfeng Sun, Xinzheng Xu, Daoqiang Zhang
LeArning Abstract Unlabeled Data Binary Classifier Positive Emotional Language Unbiased Risk Estimator Augmented Class