Saved in:
Bibliographic Details
Main Authors: Zhang, Hongwei, Ye, Ziqi, Wang, Xinyuan, Guo, Xin, Xu, Zenglin, Cheng, Yuan, Hu, Zixin, Qi, Yuan
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2506.12352
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909742512734208
author Zhang, Hongwei
Ye, Ziqi
Wang, Xinyuan
Guo, Xin
Xu, Zenglin
Cheng, Yuan
Hu, Zixin
Qi, Yuan
author_facet Zhang, Hongwei
Ye, Ziqi
Wang, Xinyuan
Guo, Xin
Xu, Zenglin
Cheng, Yuan
Hu, Zixin
Qi, Yuan
contents We propose Network Automatic Relevance Determination (NARD), an extension of ARD for linearly probabilistic models, to simultaneously model sparse relationships between inputs $X \in \mathbb R^{d \times N}$ and outputs $Y \in \mathbb R^{m \times N}$, while capturing the correlation structure among the $Y$. NARD employs a matrix normal prior which contains a sparsity-inducing parameter to identify and discard irrelevant features, thereby promoting sparsity in the model. Algorithmically, it iteratively updates both the precision matrix and the relationship between $Y$ and the refined inputs. To mitigate the computational inefficiencies of the $\mathcal O(m^3 + d^3)$ cost per iteration, we introduce Sequential NARD, which evaluates features sequentially, and a Surrogate Function Method, leveraging an efficient approximation of the marginal likelihood and simplifying the calculation of determinant and inverse of an intermediate matrix. Combining the Sequential update with the Surrogate Function method further reduces computational costs. The computational complexity per iteration for these three methods is reduced to $\mathcal O(m^3+p^3)$, $\mathcal O(m^3 + d^2)$, $\mathcal O(m^3+p^2)$, respectively, where $p \ll d$ is the final number of features in the model. Our methods demonstrate significant improvements in computational efficiency with comparable performance on both synthetic and real-world datasets.
format Preprint
id arxiv_https___arxiv_org_abs_2506_12352
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Efficient Network Automatic Relevance Determination
Zhang, Hongwei
Ye, Ziqi
Wang, Xinyuan
Guo, Xin
Xu, Zenglin
Cheng, Yuan
Hu, Zixin
Qi, Yuan
Artificial Intelligence
Machine Learning
We propose Network Automatic Relevance Determination (NARD), an extension of ARD for linearly probabilistic models, to simultaneously model sparse relationships between inputs $X \in \mathbb R^{d \times N}$ and outputs $Y \in \mathbb R^{m \times N}$, while capturing the correlation structure among the $Y$. NARD employs a matrix normal prior which contains a sparsity-inducing parameter to identify and discard irrelevant features, thereby promoting sparsity in the model. Algorithmically, it iteratively updates both the precision matrix and the relationship between $Y$ and the refined inputs. To mitigate the computational inefficiencies of the $\mathcal O(m^3 + d^3)$ cost per iteration, we introduce Sequential NARD, which evaluates features sequentially, and a Surrogate Function Method, leveraging an efficient approximation of the marginal likelihood and simplifying the calculation of determinant and inverse of an intermediate matrix. Combining the Sequential update with the Surrogate Function method further reduces computational costs. The computational complexity per iteration for these three methods is reduced to $\mathcal O(m^3+p^3)$, $\mathcal O(m^3 + d^2)$, $\mathcal O(m^3+p^2)$, respectively, where $p \ll d$ is the final number of features in the model. Our methods demonstrate significant improvements in computational efficiency with comparable performance on both synthetic and real-world datasets.
title Efficient Network Automatic Relevance Determination
topic Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2506.12352