Enhancing Neural Training via a Correlated Dynamics Model

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Brokman, Jonathan, Betser, Roy, Turjeman, Rotem, Berkov, Tom, Cohen, Ido, Gilboa, Guy
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929431047569408
author Brokman, Jonathan
Betser, Roy
Turjeman, Rotem
Berkov, Tom
Cohen, Ido
Gilboa, Guy
author_facet Brokman, Jonathan
Betser, Roy
Turjeman, Rotem
Berkov, Tom
Cohen, Ido
Gilboa, Guy
contents As neural networks grow in scale, their training becomes both computationally demanding and rich in dynamics. Amidst the flourishing interest in these training dynamics, we present a novel observation: Parameters during training exhibit intrinsic correlations over time. Capitalizing on this, we introduce Correlation Mode Decomposition (CMD). This algorithm clusters the parameter space into groups, termed modes, that display synchronized behavior across epochs. This enables CMD to efficiently represent the training dynamics of complex networks, like ResNets and Transformers, using only a few modes. Moreover, test set generalization is enhanced. We introduce an efficient CMD variant, designed to run concurrently with training. Our experiments indicate that CMD surpasses the state-of-the-art method for compactly modeled dynamics on image classification. Our modeling can improve training efficiency and lower communication overhead, as shown by our preliminary experiments in the context of federated learning.
format Preprint
id arxiv_https___arxiv_org_abs_2312_13247
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Enhancing Neural Training via a Correlated Dynamics Model
Brokman, Jonathan
Betser, Roy
Turjeman, Rotem
Berkov, Tom
Cohen, Ido
Gilboa, Guy
Machine Learning
Dynamical Systems
As neural networks grow in scale, their training becomes both computationally demanding and rich in dynamics. Amidst the flourishing interest in these training dynamics, we present a novel observation: Parameters during training exhibit intrinsic correlations over time. Capitalizing on this, we introduce Correlation Mode Decomposition (CMD). This algorithm clusters the parameter space into groups, termed modes, that display synchronized behavior across epochs. This enables CMD to efficiently represent the training dynamics of complex networks, like ResNets and Transformers, using only a few modes. Moreover, test set generalization is enhanced. We introduce an efficient CMD variant, designed to run concurrently with training. Our experiments indicate that CMD surpasses the state-of-the-art method for compactly modeled dynamics on image classification. Our modeling can improve training efficiency and lower communication overhead, as shown by our preliminary experiments in the context of federated learning.
title Enhancing Neural Training via a Correlated Dynamics Model
topic Machine Learning
Dynamical Systems
url https://arxiv.org/abs/2312.13247