Metraipolis › archive › dc_dc753f293dae

LocalProp: Neuro-Localized Memory-Efficient Backpropagation — arXiv:2609.32517

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.32517
License (as archived): CC0-1.0
Archived: 2026-10-07T20:55:37.535Z
SHA-256 of the archived text: 524d6daae7a204d45931311b11ba3a9d42575ae951a5bcc1b2882af583a6832e
Stable URL of this page: https://metraipolis.com/documento/dc_dc753f293dae

How to cite

LocalProp: Neuro-Localized Memory-Efficient Backpropagation — arXiv:2609.32517 — arXiv (en), https://arxiv.org/abs/2609.32517 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_dc753f293dae (Metraipolis archive dc_dc753f293dae, sha256 524d6daae7a204d45931311b11ba3a9d42575ae951a5bcc1b2882af583a6832e)

BibTeX

@misc{dcdc753f293dae,
  title = {LocalProp: Neuro-Localized Memory-Efficient Backpropagation — arXiv:2609.32517},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.32517},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_dc753f293dae; sha256 524d6daae7a204d45931311b11ba3a9d42575ae951a5bcc1b2882af583a6832e},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.32517. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

The current deep learning training paradigm employs end-to-end backpropagation, regardless of the training stage, i.e. pre-training or fine-tuning. However, backpropagating through the entire model is neither biologically plausible nor memory efficient, since learning inside the brain is highly localized. Therefore, we propose LocalProp, a training procedure that locally updates the weights of a model. Our neuro-localized weight updates follow the "pre-training then fine-tuning" paradigm, where the pre-training is based on I-JEPA. After locally updating the weights, a pruning operation is performed, followed by a short final fine-tuning phase. Pruning helps by sending the learning signal from higher blocks to lower blocks. We perform experiments on several datasets, including large-scale benchmarks such as ImageNet, and empirically show that LocalProp reaches good performance at a fraction of GPU peak memory. By varying the number of jointly optimized blocks, we identify gradient-propagation span as a practical control over the accuracy-memory trade-off.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Optimization and Control (math.OC).
Pagina del lavoro: https://arxiv.org/abs/2609.32517 — 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 2026-10-07T20:29:00.843Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).