analiza aiqbench
Gaussian Neural Networks introduce a novel regularization approach by imposing priors over activation-space rather than just weight-space, treating layer activities as signals with Gaussian noise. This method uses an unsupervised loss to penalize unexpected activations during training, demonstrating improved performance on classification and regression tasks while enhancing uncertainty quantification capabilities. The research represents a significant theoretical advancement in Bayesian neural network frameworks that could improve model robustness and generalization.
Gaussian Neural Networks
arXiv:2609.34825v1 Announce Type: new Abstract: Gaussian neural networks (GaNNs) are proposed as a novel regularization mechanism for neural networks. From a Bayesian perspective standard regularization techniques can be viewed as imposing priors over weight-space. Assuming priors over activation-space remains a largely unexplored possibility. GaNNs assume such priors. They do this by treating activities from earlier layers like signals with Gaussian noise and predicting the properties of the noise distribution using an additional unsupervised loss. While training, the unsupervised loss acts as a penalty on unexpected activities, allowing greater weight updates in less surprising directions. The paper demonstrates the superiority of Gaussian neural networks over standard neural networks on a variety of classification and regression tasks. We also investigate the ability of GaNNs to quantify uncertainty.
Abstract:Gaussian neural networks (GaNNs) are proposed as a novel regularization mechanism for neural networks. From a Bayesian perspective standard regularization techniques can be viewed as imposing priors over weight-space. Assuming priors over activation-space remains a largely unexplored possibility. GaNNs assume such priors. They do this by treating activities from earlier layers like signals with Gaussian noise and predicting the properties of the noise distribution using an additional unsupervised loss. While training, the unsupervised loss acts as a penalty on unexpected activities, allowing greater weight updates in less surprising directions. The paper demonstrates the superiority of Gaussian neural networks over standard neural networks on a variety of classification and regression tasks. We also investigate the ability of GaNNs to quantify uncertainty.
10 pages, 5 figures. Extended abstract and poster to be presented at ICONIP 2026
Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
arXiv:2609.34825 [cs.LG]
(or arXiv:2609.34825v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2609.34825 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 231 cuvinte
Subiecte conexe
Generalization and infinite width
LessWrong · 1 aug. 2026
A Mike's-Eye View of ARC's Research
AI Alignment Forum · 9 iun. 2026
Topology-Aware Gaussian Graph Repair for Robust Graph Neural Networks
arXiv · 3 iun. 2026
GNFormer: Gaussian Neural Field Transformer for Efficient Camera-based Occupancy Prediction
arXiv · 14 oct. 2026