Metraipolis › archive › dc_2e8bd316d889
Approaching Backpropagation Accuracy with Weight Perturbation on MNIST — arXiv:2610.21665
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.21665
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-16T10:22:40.746Z
- SHA-256 of the archived text
f3073be879bc37e8a15c75a58fa7c1d3d1333c8b43eb6c6a44fcb6c4e56c3ae7- Stable URL of this page
https://metraipolis.com/documento/dc_2e8bd316d889
How to cite
Approaching Backpropagation Accuracy with Weight Perturbation on MNIST — arXiv:2610.21665 — arXiv (en), https://arxiv.org/abs/2610.21665 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_2e8bd316d889 (Metraipolis archive dc_2e8bd316d889, sha256 f3073be879bc37e8a15c75a58fa7c1d3d1333c8b43eb6c6a44fcb6c4e56c3ae7)
BibTeX
@misc{dc2e8bd316d889,
title = {Approaching Backpropagation Accuracy with Weight Perturbation on MNIST — arXiv:2610.21665},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.21665},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_2e8bd316d889; sha256 f3073be879bc37e8a15c75a58fa7c1d3d1333c8b43eb6c6a44fcb6c4e56c3ae7},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.21665. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Weight perturbation (WP) estimates learning directions from random parameter perturbations and a scalar change in loss, without backpropagating derivatives. We examine how batch size, training duration, and estimator stabilization affect its attainable accuracy in fully connected MNIST networks. In the shallow configuration, large-batch training reaches approximately 97\% test accuracy, compared with an early backpropagation (BP) reference near 97.5--98\%. The same comparison exposes a substantial efficiency gap: under a ten-epoch budget, shallow WP reaches 90.2\%, whereas the reported BP reference reaches 97.4\%. Epoch-indexed curves show diminishing returns from increasing batch size beyond 4,000. Learning-rate reduction and averaging independent perturbation-based updates also stabilize deeper networks, although their recorded results do not establish BP-level accuracy. We distinguish these empirical findings from a finite-horizon approximation argument: under stated regularity conditions, population WP can approximate a fixed gradient-descent trajectory arbitrarily closely at any fixed finite depth. Positive classification margins then suffice to preserve the reference predictions on a finite test set with arbitrarily high probability. Together, the results identify near-BP accuracy as a configuration- and budget-dependent possibility, rather than evidence of equal efficiency or an unconditional accuracy guarantee across depth. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno). Categorie arXiv: Machine Learning (cs.LG) ; Neural and Evolutionary Computing (cs.NE). Pagina del lavoro: https://arxiv.org/abs/2610.21665 — 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 2026-10-16T08:48:02.741Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).