Metraipolis › archive › dc_6d0787f10c86
Convergent Evolution in Algorithmic Space — arXiv:2608.05985
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.05985
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-31T11:05:22.267Z
- SHA-256 of the archived text
2f25641d370b90649a3634e11761b4dab9fdd870635ec83d16ffe9160b0affb3- Stable URL of this page
https://metraipolis.com/documento/dc_6d0787f10c86
How to cite
Convergent Evolution in Algorithmic Space — arXiv:2608.05985 — arXiv (en), https://arxiv.org/abs/2608.05985 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_6d0787f10c86 (Metraipolis archive dc_6d0787f10c86, sha256 2f25641d370b90649a3634e11761b4dab9fdd870635ec83d16ffe9160b0affb3)
BibTeX
@misc{dc6d0787f10c86,
title = {Convergent Evolution in Algorithmic Space — arXiv:2608.05985},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.05985},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_6d0787f10c86; sha256 2f25641d370b90649a3634e11761b4dab9fdd870635ec83d16ffe9160b0affb3},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.05985. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
In evolutionary biology, unrelated organisms can independently evolve similar structures when exposed to similar functional demands. Here we ask whether an analogous form of convergent evolution occurs during neural network training: do networks with different random initializations develop similar internal weight structures when trained on the same task? This question is technically nontrivial because hidden neurons can be arbitrarily permuted without changing the represented function, making direct matrix comparisons misleading. We introduce a matching-based framework for comparing multilayer perceptrons in structural weight space. Hidden neurons are first coarsely aligned using permutation-invariant features and then refined by iterative Hungarian matching. After alignment, networks are compared with structural distance metrics designed to emphasize task-relevant weight patterns. Applying this approach to ensembles of small MLPs trained on MNIST, Fashion-MNIST, and KMNIST, we find that networks trained on the same task remain closer to one another than to networks trained on different tasks. Thus, task-specific training appears to guide initially random networks toward distinct regions, or attractors, in structural network space. The earliest phase of training reveals an additional and unexpected phenomenon. Classification accuracy rises rapidly before the matched structural distances show strong task-specific separation, and before the global weight distribution visibly changes. Nevertheless, individual weight entries already begin to drift in a coordinated manner. This suggests that early learning may first operate through subtle, distributed adjustments that strongly affect function while leaving coarse network morphology almost unchanged. We treat this early morphogenesis as a first glimpse of a richer dynamical process that will be investigated in future work. Tema degli esploratori: scienze della vita e salute (retro-catalogo arXiv: annunci del 2026-08-07 in q-bio.NC). Categorie arXiv: Neurons and Cognition (q-bio.NC). Pagina del lavoro: https://arxiv.org/abs/2608.05985 — 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 2026-10-31T10:55:45.823Z dalla pagina https://arxiv.org/catchup/q-bio.NC/2026-08-07?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).