Metraipolis › archive › dc_0b25c883f754
The injectivity threshold of a Gaussian ReLU layer — arXiv:2610.07492
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.07492
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T11:50:56.065Z
- SHA-256 of the archived text
4feeeaa1f7bdc3353bd9a166f707945395d834bb51ee891c2865fc27eaaf5f90- Stable URL of this page
https://metraipolis.com/documento/dc_0b25c883f754
How to cite
The injectivity threshold of a Gaussian ReLU layer — arXiv:2610.07492 — arXiv (en), https://arxiv.org/abs/2610.07492 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_0b25c883f754 (Metraipolis archive dc_0b25c883f754, sha256 4feeeaa1f7bdc3353bd9a166f707945395d834bb51ee891c2865fc27eaaf5f90)
BibTeX
@misc{dc0b25c883f754,
title = {The injectivity threshold of a Gaussian ReLU layer — arXiv:2610.07492},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.07492},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_0b25c883f754; sha256 4feeeaa1f7bdc3353bd9a166f707945395d834bb51ee891c2865fc27eaaf5f90},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.07492. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
We determine how wide a randomly initialized ReLU layer must be to distinguish every pair of inputs. For independent Gaussian weights and zero bias, as the dimensions grow at a fixed output-to-input ratio, the probability of global injectivity tends to zero below a sharp threshold and to one above it. We characterize this threshold exactly through a variational formula for the limiting minimum number of active neurons per input coordinate. The formula accounts for correlations among input directions at arbitrarily many levels. We prove this formula rigorously, establishing a prediction from the statistical physics of the spherical perceptron. The proof compares systems of nearby dimensions and controls a Gaussian interpolation by calibrating responses to auxiliary fields. A uniform volume estimate connects the soft minimum to the worst input direction. An exact-arithmetic calculation places the threshold below seven output neurons per input coordinate. Numerical evaluations suggest a value near 6.698. Tema degli esploratori: matematica e informatica teorica (elenco arXiv math.PR, nuovi lavori del giorno). Categorie arXiv: Probability (math.PR). Pagina del lavoro: https://arxiv.org/abs/2610.07492 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-07T11:21:39.616Z dalla pagina https://arxiv.org/list/math.PR/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).