Diffusive method for multiple class data classification

DOI10.48550/arXiv.2607.12313MaRDI QIDQ7284022arXivFDO

Authors Toyohiko Aiki, Takeshi Ohtsuka, Honoka Yano

Publication date 14 July 2026

Published in arXiv

Tldr This paper proposes a novel diffusive method for multiple class classification using the heat equation, where classification is determined by the sign of the difference between two solutions to the heat equation with different initial conditions; it shows promising performance on the MNIST dataset.

Full work available at URL https://doi.org/10.48550/arXiv.2607.12313

Mathematics Subject Classification ID

No records found.

arXiv classification

math.NA

Cites work

• A fast sweeping method for Eikonal equations
• Functional analysis, Sobolev spaces and partial differential equations
• Gradient-based learning applied to document recognition
• Linear Time Distances Between Fuzzy Sets With Applications to Pattern Matching and Classification
• Machine Learning
• Mathematics Subject Classification
• On behavior of signs for the heat equation and a diffusion method for data separation
• Rapid and accurate computation of the distance function using grids

This page was built for publication: Diffusive method for multiple class data classification

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7284022)