Metraipolis › archive › dc_76030d1e2fcf
Sparse Bayesian Neural Networks via Sticky Piecewise Deterministic Markov Processes — arXiv:2610.32753
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.32753
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-26T20:14:50.086Z
- SHA-256 of the archived text
80512aaf9ca6fba2c2df32a374f2559fa76f3789b3974eceefc1b5630e88b748- Stable URL of this page
https://metraipolis.com/documento/dc_76030d1e2fcf
How to cite
Sparse Bayesian Neural Networks via Sticky Piecewise Deterministic Markov Processes — arXiv:2610.32753 — arXiv (en), https://arxiv.org/abs/2610.32753 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_76030d1e2fcf (Metraipolis archive dc_76030d1e2fcf, sha256 80512aaf9ca6fba2c2df32a374f2559fa76f3789b3974eceefc1b5630e88b748)
BibTeX
@misc{dc76030d1e2fcf,
title = {Sparse Bayesian Neural Networks via Sticky Piecewise Deterministic Markov Processes — arXiv:2610.32753},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.32753},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_76030d1e2fcf; sha256 80512aaf9ca6fba2c2df32a374f2559fa76f3789b3974eceefc1b5630e88b748},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.32753. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
There is growing evidence that the performance of a neural network depends largely on a small subset of its weights, which are typically identified by training a dense network and pruning it afterwards. Within a Bayesian framework, pruning can be directly incorporated through a spike-and-slab prior, providing a proper treatment of the uncertainties involved, both with respect to pruning and to parameter uncertainty. In Bayesian neural networks (BNNs), sampling by MCMC is notoriously hard. We demonstrate that modern Monte Carlo methods based on piecewise deterministic Markov processes (PDMPs) can be used to sample posteriors of BNNs asymptotically exact whenever numerically constructed rate bounds hold. Spike-and-slab priors can be incorporated through sticky versions of the PDMPs, setting weights exactly to zero during sampling. We examine the predictive performance and convergence of the samplers, and show that sparse BNNs sampled this way can be more robust to input noise than dense ones. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno). Categorie arXiv: Machine Learning (stat.ML) ; Machine Learning (cs.LG); Computation (stat.CO). Pagina del lavoro: https://arxiv.org/abs/2610.32753 — 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 2026-10-26T19:49:03.503Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).