Fuzzy multi-label learning under veristic variables

Abstract

Multi-label learning is increasingly required by many applications where instances may belong to several classes at the same time. In this paper, we propose a fuzzy k-nearest neighbor method for multi-label classification using the veristic variable framework. Veristic variables are variables that can assume simultaneously multiple values with different degrees. In multi-label learning, class labels can be considered as veristic variables since each instance can belong simultaneously to more than one class. Several applications on benchmark datasets demonstrate the efficiency of our approach.

Multi‐label learning: a review of the state of the art and ongoing research

“…Evidential Multi‐Label kNN (EML‐kNN) used an evidence‐theoretic rule that extended the Dempster–Shafer framework to the MLL setting with only a moderate increase in complexity as compared to the classical case. Fuzzy Veristic kNN (FV‐kNN) used a fuzzy kNN rule for MLL based on the theory of Veristic Variables . It was able to generate fuzzy label sets for instances that have been originally labeled by crisp ones and obtained competitive results.…”

A new approach to uncertainty description through accomplishment membership functions

“…This common framework relaxes the mathematical restrictions imposed to uncertainty representations and allows the possibility theory, for instance, to be the starting point for further developments; e.g., the proposal of veristic variables which can manipulate not one but many solutions to a given proposition (Yager, 2000a(Yager, , 2000b. A veristic approach to classification has been reported by Younes, Abdallah, and Denoeux (2010).…”