Reduced-Order Modeling of Deep Neural Networks

被引：2

作者：

Gusak, J. ^{[1
]}

Daulbaev, T. ^{[1
]}

Ponomarev, E. ^{[1
]}

Cichocki, A. ^{[1
]}

Oseledets, I ^{[1
]}

机构：

[1] Skolkovo Inst Sci & Technol, Moscow, Russia

来源：

COMPUTATIONAL MATHEMATICS AND MATHEMATICAL PHYSICS | 2021年 / 61卷 / 05期

关键词：

acceleration of neural networks; MaxVol; machine learning; component analysis;

D O I：

10.1134/S0965542521050109

中图分类号：

O29 [应用数学];

学科分类号：

070104 ;

摘要：

We introduce a new method for speeding up the inference of deep neural networks. It is somewhat inspired by the reduced-order modeling techniques for dynamical systems. The cornerstone of the proposed method is the maximum volume algorithm. We demonstrate efficiency on neural networks pre-trained on different datasets. We show that in many practical cases it is possible to replace convolutional layers with much smaller fully-connected layers with a relatively small drop in accuracy.

引用

页码：774 / 785

页数：12

共 50 条

[1] Reduced-Order Modeling of Deep Neural Networks
J. Gusak
T. Daulbaev
E. Ponomarev
A. Cichocki
I. Oseledets
[J]. Computational Mathematics and Mathematical Physics, 2021, 61 : 774 - 785
[2] Reduced-Order Modeling of Steady and Unsteady Flows with Deep Neural Networks
Barraza, Bryan
Gross, Andreas
[J]. AEROSPACE, 2024, 11 (07)
[3] HiDeNN-TD: Reduced-order hierarchical deep learning neural networks
Zhang, Lei
Lu, Ye
Tang, Shaoqiang
Liu, Wing Kam
[J]. COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING, 2022, 389
[4] Bifidelity Data-Assisted Neural Networks in Nonintrusive Reduced-Order Modeling
Lu, Chuan
Zhu, Xueyu
[J]. JOURNAL OF SCIENTIFIC COMPUTING, 2021, 87 (01)
[5] Bifidelity Data-Assisted Neural Networks in Nonintrusive Reduced-Order Modeling
Chuan Lu
Xueyu Zhu
[J]. Journal of Scientific Computing, 2021, 87
[6] Reduced-order modeling of conductive polymer pressure sensors using finite element simulations and deep neural networks
Ashouri, Majid
Khaleghian, Seyedmeysam
Emami, Anahita
[J]. STRUCTURAL AND MULTIDISCIPLINARY OPTIMIZATION, 2022, 65 (05)
[7] Reduced-order modeling of conductive polymer pressure sensors using finite element simulations and deep neural networks
Majid Ashouri
Seyedmeysam Khaleghian
Anahita Emami
[J]. Structural and Multidisciplinary Optimization, 2022, 65
[8] Construction of reduced-order models for fluid flows using deep feedforward neural networks
Lui, Hugo F. S.
Wolf, William R.
[J]. JOURNAL OF FLUID MECHANICS, 2019, 872 : 963 - 994
[9] Reduced-Order Modeling of Subsurface Multi-phase Flow Models Using Deep Residual Recurrent Neural Networks
Kani, J. Nagoor
Elsheikh, Ahmed H.
[J]. TRANSPORT IN POROUS MEDIA, 2019, 126 (03) : 713 - 741
[10] Reduced-Order Modeling of Subsurface Multi-phase Flow Models Using Deep Residual Recurrent Neural Networks
Ahmed H. J. Nagoor Kani
[J]. Transport in Porous Media, 2019, 126 : 713 - 741

← 1 2 3 4 5 →