Guardado en:
Detalles Bibliográficos
Autores principales: Sun, Youran, Liu, Yihua, Niu, Yi-Shuai
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:https://arxiv.org/abs/2412.09853
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866912154352877568
author Sun, Youran
Liu, Yihua
Niu, Yi-Shuai
author_facet Sun, Youran
Liu, Yihua
Niu, Yi-Shuai
contents Difference-of-Convex Algorithm (DCA) is a well-known nonconvex optimization algorithm for minimizing a nonconvex function that can be expressed as the difference of two convex ones. Many famous existing optimization algorithms, such as SGD and proximal point methods, can be viewed as special DCAs with specific DC decompositions, making it a powerful framework for optimization. On the other hand, shortcuts are a key architectural feature in modern deep neural networks, facilitating both training and optimization. We showed that the shortcut neural network gradient can be obtained by applying DCA to vanilla neural networks, networks without shortcut connections. Therefore, from the perspective of DCA, we can better understand the effectiveness of networks with shortcuts. Moreover, we proposed a new architecture called NegNet that does not fit the previous interpretation but performs on par with ResNet and can be included in the DCA framework.
format Preprint
id arxiv_https___arxiv_org_abs_2412_09853
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Understand the Effectiveness of Shortcuts through the Lens of DCA
Sun, Youran
Liu, Yihua
Niu, Yi-Shuai
Machine Learning
Neural and Evolutionary Computing
Optimization and Control
Difference-of-Convex Algorithm (DCA) is a well-known nonconvex optimization algorithm for minimizing a nonconvex function that can be expressed as the difference of two convex ones. Many famous existing optimization algorithms, such as SGD and proximal point methods, can be viewed as special DCAs with specific DC decompositions, making it a powerful framework for optimization. On the other hand, shortcuts are a key architectural feature in modern deep neural networks, facilitating both training and optimization. We showed that the shortcut neural network gradient can be obtained by applying DCA to vanilla neural networks, networks without shortcut connections. Therefore, from the perspective of DCA, we can better understand the effectiveness of networks with shortcuts. Moreover, we proposed a new architecture called NegNet that does not fit the previous interpretation but performs on par with ResNet and can be included in the DCA framework.
title Understand the Effectiveness of Shortcuts through the Lens of DCA
topic Machine Learning
Neural and Evolutionary Computing
Optimization and Control
url https://arxiv.org/abs/2412.09853