cs.LGNov 4, 2026

Neural Compression over Equivalence Classes: Exact Gauges and Functional Rate-Distortion

Authors: Toshiaki Koike-Akino

Abstract

Neural networks admit many parameterizations of the same or nearly the same function, but deployment typically stores whichever parameter vector training happens to produce. We show that this representational freedom can itself be used for compression. Rather than redesigning the codec, we search over exact function-preserving transformations and controlled functional perturbations, using the unchanged physical compressor itself to select cheaper representations. Exact gauge families provide a zero-distortion reference under the strict audit; allowing small functional changes exposes much larger rate-distortion gains. Across SmolLM2 and Qwen3, our method reduces Zstandard-compressed checkpoint size by up to 23.4%, with similar relative reductions under gzip, Brotli, and ZipNN, establishing trained-network representational freedom as a practical storage resource.

Figures & tables

Appendix figures & tables20 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. NeuralZip: Reusable Setup for Fast Lossless Compression

    Oct 7, 2026Martín Bravo, Samuel Horváth, Gonzalo Navarro +1Structured Neural Network CompressionLossless Data Compression

  2. Neural Network Compression by Approximate Differential Equivalence

    May 31, 2026Ravi Dhiman, Andrea Passarella, Mirco Tribastone +1Neural Network Compression

  3. Width-Independent Compressibility of Deep Neural Networks

    Aug 22, 2026Hong-Yi Wang, Mingze Wang, Liu ZiyinNeural Network CompressionNeural Network Approximation Theory