About: Using Wavelets and Spectral Methods to Study Patterns in Image-Classification Datasets

Facets (new session)
Description
Metadata
Settings
- owl:sameAs
- Inference Rule:

About: Using Wavelets and Spectral Methods to Study Patterns in Image-Classification Datasets Goto Sponge NotDistinct Permalink

An Entity of Type : schema:ScholarlyArticle, within Data Space : wasabi.inria.fr associated with source document(s)

Attributes	Values
type	Academic Article research paper schema:ScholarlyArticle
isDefinedBy	Covid-on-the-Web dataset
has title	Using Wavelets and Spectral Methods to Study Patterns in Image-Classification Datasets
Creator	Huang, Furong Yousefzadeh, Roozbeh
Source	ArXiv
abstract	Deep learning models extract, before a final classification layer, features or patterns which are key for their unprecedented advantageous performance. However, the process of complex nonlinear feature extraction is not well understood, a major reason why interpretation, adversarial robustness, and generalization of deep neural nets are all open research problems. In this paper, we use wavelet transformation and spectral methods to analyze the contents of image classification datasets, extract specific patterns from the datasets and find the associations between patterns and classes. We show that each image can be written as the summation of a finite number of rank-1 patterns in the wavelet space, providing a low rank approximation that captures the structures and patterns essential for learning. Regarding the studies on memorization vs learning, our results clearly reveal disassociation of patterns from classes, when images are randomly labeled. Our method can be used as a pattern recognition approach to understand and interpret learnability of these datasets. It may also be used for gaining insights about the features and patterns that deep classifiers learn from the datasets.
has issue date	2020-06-17(xsd:dateTime)
has license	arxiv
sha1sum (hex)	3484d9d7058afa54f1b5406166df77e9ea605236
resource representing a document's title	Using Wavelets and Spectral Methods to Study Patterns in Image-Classification Datasets
resource representing a document's body	covid:3484d9d7058afa54f1b5406166df77e9ea605236#body_text
is schema:about of	named entity 'generalization' named entity 'summation' named entity 'captures' named entity 'number' covid:arg/3484d9d7058afa54f1b5406166df77e9ea605236 named entity 'patterns' named entity 'pattern recognition' named entity 'layer' named entity 'image classification' named entity 'disassociation' named entity 'patterns' named entity 'learn' named entity 'patterns' named entity 'features' named entity 'models' named entity 'classifiers' named entity 'datasets' named entity 'deep neural nets' named entity 'image classification' named entity 'classifiers' named entity 'wavelet transformation' named entity 'nonlinear' named entity 'Neyshabur' named entity 'singular values' named entity 'wavelet' named entity 'subset' named entity 'wavelet' named entity 'edge detection' named entity 'training set' named entity 'wavelet' named entity 'deep learning' named entity 'wavelets' named entity 'nonlinear' named entity 'CIFAR-10' named entity 'extract features' named entity 'numerical algorithms' named entity 'spectral decomposition' named entity 'Belkin' named entity 'Deep learning' named entity 'nonlinear' named entity '2-norm' named entity 'wavelet' named entity 'low-rank approximation' named entity 'pattern analysis' named entity 'spectral decomposition' named entity 'basis vectors' named entity 'wavelet' named entity 'spectral methods' named entity 'Fourier transform' named entity 'similarity matrix' named entity 'Fourier transform' named entity 'over-fitting' named entity 't-SNE' named entity 'D2b' named entity 'COVID-19' named entity 'wavelet' named entity 'singular values' named entity 'wavelet' named entity 'image classification' named entity 'wavelet transformation' named entity 'matrices' named entity 'CIFAR-10' named entity 'GSVD' named entity 'SARS-COV-2' named entity 'vectors' named entity 'wavelet' named entity 'wavelet transformation' named entity 'singular values' named entity 'image classification' named entity 'wavelet' named entity 'Gaussian kernel' named entity 'wavelets' named entity 'graph Laplacian' named entity 'GSVD' named entity 'wavelet' named entity 'wavelet' named entity 'tensor decomposition' named entity 'real numbers' named entity 'wavelet' named entity 'wavelet' named entity 'CIFAR-10' named entity 'pixel' named entity 'orthogonality' named entity 'MNIST' named entity 'tensor decomposition' named entity 'orthogonality'

Faceted Search & Find service v1.13.91 as of Mar 24 2020

Alternative Linked Data Documents: Sponger | ODE Content Formats:

RDF

ODATA

Microdata

About

OpenLink Virtuoso version 07.20.3229 as of Jul 10 2020, on Linux (x86_64-pc-linux-gnu), Single-Server Edition (94 GB total memory)
Data on this page belongs to its respective rights holders.
Virtuoso Faceted Browser Copyright © 2009-2024 OpenLink Software