Výsledky vyhledávání

Report

An Unbiased Risk Estimator for Partial Label Learning with Augmented Classes

Autor: Hu, Jiayu, Shu, Senlin, Li, Beibei, Xiang, Tao, He, Zhongshi

Partial Label Learning (PLL) is a typical weakly supervised learning task, which assumes each training instance is annotated with a set of candidate labels containing the ground-truth label. Recent PLL methods adopt identification-based disambiguatio

Externí odkaz: http://arxiv.org/abs/2409.19600

Zobrazit plný text záznamu

Report

A Generalized Unbiased Risk Estimator for Learning with Augmented Classes

Autor: Shu, Senlin, He, Shuo, Wang, Haobo, Wei, Hongxin, Xiang, Tao, Feng, Lei

In contrast to the standard learning paradigm where all classes can be observed in training data, learning with augmented classes (LAC) tackles the problem where augmented classes unobserved in the training data may emerge in the test phase. Previous

Externí odkaz: http://arxiv.org/abs/2306.06894

Zobrazit plný text záznamu

Report

Multi-Class Classification from Single-Class Data with Confidences

Autor: Cao, Yuzhou, Feng, Lei, Shu, Senlin, Xu, Yitian, An, Bo, Niu, Gang, Sugiyama, Masashi

Can we learn a multi-class classifier from only data of a single class? We show that without any assumptions on the loss functions, models, and optimizers, we can successfully learn a multi-class classifier from only data of a single class with a rig

Externí odkaz: http://arxiv.org/abs/2106.08864

Zobrazit plný text záznamu

Report

Pointwise Binary Classification with Pairwise Confidence Comparisons

Autor: Feng, Lei, Shu, Senlin, Lu, Nan, Han, Bo, Xu, Miao, Niu, Gang, An, Bo, Sugiyama, Masashi

To alleviate the data requirement for training effective binary classifiers in binary classification, many weakly supervised learning settings have been proposed. Among them, some consider using pairwise but not pointwise labels, when pointwise label

Externí odkaz: http://arxiv.org/abs/2010.01875

Zobrazit plný text záznamu

Report

Incorporating Multiple Cluster Centers for Multi-Label Learning

Autor: Shu, Senlin, Lv, Fengmao, Yan, Yan, Li, Li, He, Shuo, He, Jun

Multi-label learning deals with the problem that each instance is associated with multiple labels simultaneously. Most of the existing approaches aim to improve the performance of multi-label learning by exploiting label correlations. Although the da

Externí odkaz: http://arxiv.org/abs/2004.08113

Zobrazit plný text záznamu