FSMRL-based Feature Selection for Multi-Source Interval-Valued Decision Systems

doi:10.1016/j.ijar.2026.109849

FSMRL-based Feature Selection for Multi-Source Interval-Valued Decision Systems

2026-10-13

0

PRE

AI

DOI：10.1016/j.ijar.2026.109849

Original

Original request for help

Save

Abstract

En 中文

• An FSMRL framework is developed for multi‑source interval‑valued feature selection under label enhancement. • Neighborhood label evidence and Jaccard‑based prior are normalized to construct valid label distributions. • An interval‑aware neighborhood distance is defined to preserve inherent uncertainty of interval‑valued data. • A reliability‑driven feature‑level fusion strategy aggregates informative features from multiple sources. • An evaluation function optimizes feature‑label relevance and feature redundancy for multi‑source feature selection.

Keywords:

Feature Selection

Multi - Label Learning

Neighborhood Rough Set

Label Enhancement

Information Fusion

FSMRL

FSMRL-based Feature Selection for Multi-Source Interval-Valued Decision Systems

Abstract

Journal

Organization

Cited Papers