Efficient Attribute Reduction with Minimal Cost for Large-Scale Data

doi:10.1016/j.ijar.2026.109683

Efficient Attribute Reduction with Minimal Cost for Large-Scale Data

2026-04-16

0

PRE

AI

DOI：10.1016/j.ijar.2026.109683

Original

Original request for help

Save

Abstract

En 中文

• We propose a highly efficient minimum cost attribute reduction algorithm (MCDV) for test-cost-sensitive learning on large-scale datasets. • We develop an equidistant center granulation method to efficiently construct high-quality multi-granular information granules, avoiding redundant granulation and improving scalability. • We introduce the ∂-description vector to jointly evaluate the classification capability of attributes and test cost, transforming the problem into a single-objective optimization task. • The proposed method significantly reduces computational complexity from quadratic to linear, enabling efficient processing of large-scale datasets. • Extensive experiments demonstrate that MCDV achieves substantial reductions in attribute cost and runtime while maintaining high classification accuracy, outperforming existing methods.

Keywords:

Attribute Reduction

Test-Cost-Sensitive Learning

Multi-Granular Information

Minimum Cost

Large-Scale Data

Efficient Attribute Reduction with Minimal Cost for Large-Scale Data

Abstract

Journal

Organization