Merging of Neural Networks
Neural Processing Letters · 06.02.2024 · 56 (1)
DOI: 10.1007/s11063-024-11445-y ↗Kurz gefasst
Fachartikel in der Zeitschrift Neural Processing Letters, online erschienen am 06.02.2024. Verfasst von Martin Pašen und 1 weiteren Autor:innen. Die Arbeit wurde bisher 4-mal zitiert (Crossref) und nennt 23 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Comenius University in Bratislava.
Abstract (Originalsprache)
We propose a simple scheme for merging two neural networks trained with different starting initialization into a single one with the same size as the original ones. We do this by carefully selecting channels from each input network. Our procedure might be used as a finalization step after one tries multiple starting seeds to avoid an unlucky one. We also show that training two networks and merging them leads to better performance than training a single network for an extended period of time.
Förderung
- Comenius University in Bratislava
Zitieren
APA
Pašen, M., & Boža, V. (2024). Merging of Neural Networks. Neural Processing Letters, 56(1). https://doi.org/10.1007/s11063-024-11445-y
Vancouver
Pašen M, Boža V. Merging of Neural Networks. Neural Processing Letters. 2024;56(1). doi:10.1007/s11063-024-11445-y
BibTeX
@article{Paen2024,
author = {Pašen, Martin and Boža, Vladimír},
title = {Merging of Neural Networks},
journal = {Neural Processing Letters},
year = {2024},
volume = {56},
number = {1},
doi = {10.1007/s11063-024-11445-y}
}
EUVIMED