Saved in:
Bibliographic Details
Main Authors: Zhu, Yunqin, Yuchi, Henry Shaowu, Xie, Yao
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2505.18526
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911420131573760
author Zhu, Yunqin
Yuchi, Henry Shaowu
Xie, Yao
author_facet Zhu, Yunqin
Yuchi, Henry Shaowu
Xie, Yao
contents Learning expressive kernels while retaining tractable inference remains a central challenge in scaling Gaussian processes (GPs) to large and complex datasets. We propose a scalable GP regressor based on deep basis kernels (DBKs). Our DBK is constructed from a small set of neural-network-parameterized basis functions with an explicit low-rank structure. This formulation immediately enables linear-complexity inference with respect to the number of samples, possibly without inducing points. DBKs provide a unifying perspective that recovers sparse deep kernel learning and Gaussian Bayesian last-layer methods as special cases. We further identify that naively maximizing the marginal likelihood can lead to oversimplified uncertainty and rank-deficient solutions. To address this, we introduce a mini-batch stochastic objective that directly targets the predictive distribution with decoupled regularization. Empirically, DBKs show advantages in predictive accuracy, uncertainty quantification, and computational efficiency across a range of large-scale regression benchmarks.
format Preprint
id arxiv_https___arxiv_org_abs_2505_18526
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Scalable Deep Basis Kernel Gaussian Processes
Zhu, Yunqin
Yuchi, Henry Shaowu
Xie, Yao
Machine Learning
Learning expressive kernels while retaining tractable inference remains a central challenge in scaling Gaussian processes (GPs) to large and complex datasets. We propose a scalable GP regressor based on deep basis kernels (DBKs). Our DBK is constructed from a small set of neural-network-parameterized basis functions with an explicit low-rank structure. This formulation immediately enables linear-complexity inference with respect to the number of samples, possibly without inducing points. DBKs provide a unifying perspective that recovers sparse deep kernel learning and Gaussian Bayesian last-layer methods as special cases. We further identify that naively maximizing the marginal likelihood can lead to oversimplified uncertainty and rank-deficient solutions. To address this, we introduce a mini-batch stochastic objective that directly targets the predictive distribution with decoupled regularization. Empirically, DBKs show advantages in predictive accuracy, uncertainty quantification, and computational efficiency across a range of large-scale regression benchmarks.
title Scalable Deep Basis Kernel Gaussian Processes
topic Machine Learning
url https://arxiv.org/abs/2505.18526