DFReg: A Physics-Inspired Framework for Global Weight Distribution Regularization in Neural Networks

Giovanni Ruggieri
6/30/2025
cs.LG

No confident code implementation found for this paper.

Abstract

We introduce DFReg, a physics-inspired regularization method for deep neural networks that operates on the global distribution of weights. Drawing from Density Functional Theory (DFT), DFReg applies a functional penalty to encourage smooth, diverse, and well-distributed weight configurations. Unlike traditional techniques such as Dropout or L2 decay, DFReg imposes global structural regularity without architectural changes or stochastic perturbations.

View on arXivView PDF

Code Implementations

No confident code match yet

We couldn't find an author-owned or strongly-evidenced community implementation for this paper. Any repos shown below are weak matches — verify before relying on them.

No code implementations found yet.

Know of an implementation? Let us know in the comments below!

Similar Papers

Weight-Sharing Regularization

83% match1 repo42023

Weight Compander: A Simple Weight Reparameterization for Regularization

83% match2023

Deep Learning Meets Sparse Regularization: A Signal Processing Perspective

81% match2023

Heavy-Tailed Regularization of Weight Matrices in Deep Neural Networks

80% match2023

Data Diversity as Implicit Regularization: How Does Diversity Shape the Weight Space of Deep Neural Networks?

80% match2024

Achieving Constraints in Neural Networks: A Stochastic Augmented Lagrangian Approach

79% match2023

Cite this paper

@article{ruggieri2025dfreg,
  title  = {DFReg: A Physics-Inspired Framework for Global Weight Distribution Regularization in Neural Networks},
  author = {Giovanni Ruggieri},
  year   = {2025},
  eprint = {2507.00101},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2507.00101v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2507.00101.svg)](https://codeofpaper.com/paper/2507.00101)

Discussion