Stacking as Accelerated Gradient Descent

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Agarwal, Naman, Awasthi, Pranjal, Kale, Satyen, Zhao, Eric
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916619752570880
author Agarwal, Naman
Awasthi, Pranjal
Kale, Satyen
Zhao, Eric
author_facet Agarwal, Naman
Awasthi, Pranjal
Kale, Satyen
Zhao, Eric
contents Stacking, a heuristic technique for training deep residual networks by progressively increasing the number of layers and initializing new layers by copying parameters from older layers, has proven quite successful in improving the efficiency of training deep neural networks. In this paper, we propose a theoretical explanation for the efficacy of stacking: viz., stacking implements a form of Nesterov's accelerated gradient descent. The theory also covers simpler models such as the additive ensembles constructed in boosting methods, and provides an explanation for a similar widely-used practical heuristic for initializing the new classifier in each round of boosting. We also prove that for certain deep linear residual networks, stacking does provide accelerated training, via a new potential function analysis of the Nesterov's accelerated gradient method which allows errors in updates. We conduct proof-of-concept experiments to validate our theory as well.
format Preprint
id arxiv_https___arxiv_org_abs_2403_04978
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Stacking as Accelerated Gradient Descent
Agarwal, Naman
Awasthi, Pranjal
Kale, Satyen
Zhao, Eric
Machine Learning
Stacking, a heuristic technique for training deep residual networks by progressively increasing the number of layers and initializing new layers by copying parameters from older layers, has proven quite successful in improving the efficiency of training deep neural networks. In this paper, we propose a theoretical explanation for the efficacy of stacking: viz., stacking implements a form of Nesterov's accelerated gradient descent. The theory also covers simpler models such as the additive ensembles constructed in boosting methods, and provides an explanation for a similar widely-used practical heuristic for initializing the new classifier in each round of boosting. We also prove that for certain deep linear residual networks, stacking does provide accelerated training, via a new potential function analysis of the Nesterov's accelerated gradient method which allows errors in updates. We conduct proof-of-concept experiments to validate our theory as well.
title Stacking as Accelerated Gradient Descent
topic Machine Learning
url https://arxiv.org/abs/2403.04978