Fourier Neural Operator for Image Classification

The present work seeks to analyze the performance of the Fourier Neural Operator (symbolized by FNO) as a convolution method for an image classification and how is its performance when compared to ResNet20 (benchmarking). The possible advantage of this technique is the success in pattern recognition...

Full description

Saved in:
Bibliographic Details
Main Authors: Johnny, Williamson, Brigido, Hatzinakis, Ladeira, Marcelo, Souza, Joao Carlos Felix
Format: Conference Proceeding
Language:English
Subjects:
Convolution
Convolutional Neural Networks
Fourier Neural Network
Fourier Neural Operator
Fourier transforms
Image classification
Image recognition
Mathematical models
Measurement
Partial differential equations
ResNet
Training
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The present work seeks to analyze the performance of the Fourier Neural Operator (symbolized by FNO) as a convolution method for an image classification and how is its performance when compared to ResNet20 (benchmarking). The possible advantage of this technique is the success in pattern recognition through the wave-forms properties of Fourier analysis and due the power of synthesis that the Fourier Transform has. The FNO was very efficient to solve parametric partial differential equations like Navier-Stokes Equation. ResNet20 took 21 minutes and 51 seconds for training, while the FNO took 4:11:14 hours to complete a hundred of epochs. The convolution occurs in a competitive way in the FNO, being perfectly possible to be used in the image recognition processes, with accuracy, recall, precision and F-Score slightly better than ResNet20 and quite similar to other neural networks available in the literature. However, based on time consuming, the FNO is not indicated to image classification. It should be used for other purpose like solving partial derivative equations.
ISSN:2166-0727
DOI:10.23919/CISTI54924.2022.9820128