JMLR

Interpretable Global Minima of Deep ReLU Neural Networks on Sequentially Separable Data

Authors
Thomas Chen Patrícia Muñoz Ewald
Research Topics
Machine Learning
Paper Information
  • Journal:
    Journal of Machine Learning Research
  • Added to Tracker:
    Sep 08, 2025
Abstract

We explicitly construct zero loss neural network classifiers. We write the weight matrices and bias vectors in terms of cumulative parameters, which determine truncation maps acting recursively on input space. The configurations for the training data considered are $(i)$ sufficiently small, well separated clusters corresponding to each class, and $(ii)$ equivalence classes which are sequentially linearly separable. In the best case, for $Q$ classes of data in $\mathbb{R}^{M}$, global minimizers can be described with $Q(M+2)$ parameters.

Author Details
Thomas Chen
Author
Patrícia Muñoz Ewald
Author
Research Topics & Keywords
Machine Learning
Research Area
Citation Information
APA Format
Thomas Chen & Patrícia Muñoz Ewald . Interpretable Global Minima of Deep ReLU Neural Networks on Sequentially Separable Data. Journal of Machine Learning Research .
BibTeX Format
@article{paper482,
  title = { Interpretable Global Minima of Deep ReLU Neural Networks on Sequentially Separable Data },
  author = { Thomas Chen and Patrícia Muñoz Ewald },
  journal = { Journal of Machine Learning Research },
  url = { https://www.jmlr.org/papers/v26/24-1516.html }
}