Emerging topics and challenges of learning from noisy data in nonstandard classification: a survey beyond binary class noise

Author(s)

Prati, Ronaldo C. ^[1] ; Luengo, Julian ^[2] ; Herrera, Francisco ^[2]

Affiliation

^[1] Fed Univ ABC UFABC, Ctr Math Comp Sci & Cognit CMCC, Santo Andre, SP - Brazil

^[2] Univ Granada UGR, Dept Comp Sci & AI DECSAI, Granada - Spain

• Document type: Journal article
• Source: KNOWLEDGE AND INFORMATION SYSTEMS; v. 60, n. 1, p. 63-97, JUL 2019.
• Web of Science Citations: 0

Abstract

The problem of class noisy instances is omnipresent in different classification problems. However, most of research focuses on noise handling in binary classification problems and adaptations to multiclass learning. This paper aims to contextualize noise labels in the context of non-binary classification problems, including multiclass, multilabel, multitask, multi-instance ordinal and data stream classification. Practical considerations for analyzing noise under these classification problems, as well as trends, open-ended problems and future research directions are analyzed. We believe this paper could help expand research on class noise handling and help practitioners to better identify the particular aspects of noise in challenging classification scenarios. (AU)

• FAPESP's process: 15/20606-6 - Noise Labels in Machine Learning: Evaluation measures and machine learning algorithms
• Grantee: Ronaldo Cristiano Prati
• Support Opportunities: Scholarships abroad - Research