GNet: A scalable and flexible Gaussian process network with nonparametric neurons

Abstract

We build GNet, a scalable and flexible Gaussian process network with nonparametric activation functions. The key computational contribution is to make estimation of GP activations computationally scalable without forming covariance matrices explicitly. We provide theoretical studies comparing function classes with a fixed architecture, and derive predictive risk bounds for GNets. Using essentially the same architecture and optimization settings, GNet shows competitive performance across a diverse range of test problems, including predicting nonlinear functions, nonparametric regression of real-world data, and predicting one-body direct correlation functions with high-dimensional inputs in classical density functional theory. The strong performance of GNet suggests that flexible nonparametric neurons can substantially reduce the number of trainable parameters while remaining computationally scalable.

Dirichlet Process Mixtures of Trees with Gaussian Process Splits: A Bayesian Nonparametric Framework with Posterior Contraction Rate
Subhasish Basak, Anik Roy, Sourabh Bhattacharya
arXiv1 Sep 2026PreprintarXiv licenceFull textStatisticsarXiv
Bayesian nonparametric estimation of correlated gravitational wave detector network noise using matrix-gamma process priors
Yixuan Liu, Renate Meyer, Nelson Christensen +4
HAL1 Sep 2026PreprintCC BY 4.0Full textHigh-energy astrophysics Instrumentation & methodsHALarXiv
ISBO: Scalable Spatio-Temporal Bayesian Optimization with Log Gaussian Cox Process Models via the INLA-SPDE Approach
Kaichuang Yang, Håvard Rue, Jakob Zeitler
arXiv10 Oct 2026PreprintCC BY 4.0Full textStatisticsarXiv
Scalable Logistic Gaussian Process Density Regression with Kinetic Langevin Sampling
Daniel Paulin, Ádám Jung, András A. Benczúr
arXiv7 Oct 2026PreprintCC BY 4.0Full textStatisticsarXiv
Triply-Scalable Equivariant Gaussian Process Modeling
Tim Steinert, David Ginsbourger
arXiv17 Sep 2026PreprintCC BY 4.0Full textStatisticsarXiv

Cite

BibTeX RIS

BibTeX

@misc{gu2026gnet,
  title = {{GNet: A scalable and flexible Gaussian process network with nonparametric neurons}},
  author = {Mengyang Gu},
  howpublished = {arXiv},
  year = {2026},
  month = sep,
  doi = {10.48550/arxiv.2607.10735},
  eprint = {2607.10735},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2607.10735},
}

RIS

TY  - UNPB
TI  - GNet: A scalable and flexible Gaussian process network with nonparametric neurons
AU  - Mengyang Gu
PY  - 2026
DA  - 2026/09/21
PB  - arXiv
DO  - 10.48550/arxiv.2607.10735
UR  - https://arxiv.org/abs/2607.10735
AB  - We build GNet, a scalable and flexible Gaussian process network with nonparametric activation functions. The key computational contribution is to make estimation of GP activations computationally scalable without forming covariance matrices explicitly. We provide theoretical studies comparing function classes with a fixed architecture, and derive predictive risk bounds for GNets. Using essentially the same architecture and optimization settings, GNet shows competitive performance across a diverse range of test problems, including predicting nonlinear functions, nonparametric regression of real-world data, and predicting one-body direct correlation functions with high-dimensional inputs in classical density functional theory. The strong performance of GNet suggests that flexible nonparametric neurons can substantially reduce the number of trainable parameters while remaining computationally scalable.
ER  -

Abstract

Related

Cite