Ulicny, Matej and Krylov, Vladimir A and Dahyot, Rozenn (2019) Harmonic Networks for Image Classification. British Machine Vision Conference (BMVC).

Preview

Download (1MB) | Preview

Abstract

Convolutional neural networks (CNNs) learn filters in order to capture local correlation patterns in feature space. In contrast, in this paper we propose harmonic blocks that produce features by learning optimal combinations of responses to preset spectral filters. We rely on the use of the Discrete Cosine Transform filters which have excellent energy compaction properties and are widely used for image compression. The proposed harmonic blocks are intended to replace conventional convolutional layers to produce partially or fully harmonic versions of new or existing CNN architectures. We demonstrate how the harmonic networks can be efficiently compressed by exploiting redundancy in spectral domain and truncating high-frequency information. We extensively validate our approach and show that the introduction of harmonic blocks into state-of-the-art CNN models results in improved classification performance on CIFAR and ImageNet datasets.

Item Type:	Article
Keywords:	Harmonic; Networks; Image; Classification;
Academic Unit:	Faculty of Science and Engineering > Computer Science
Item ID:	15155
Depositing User:	Rozenn Dahyot
Date Deposited:	20 Dec 2021 12:38
Journal or Publication Title:	British Machine Vision Conference (BMVC)
Publisher:	British Machine Vision Association
Refereed:	Yes
URI:	Publisher
Use Licence:	This item is available under a Creative Commons Attribution Non Commercial Share Alike Licence (CC BY-NC-SA). Details of this licence are available here

Repository Staff Only(login required)

Item control page

Downloads

Origin of downloads