ocitn's Introduction

OCITN

In this repository, we provide source code for "Less complexity one-class classification approach using construction error of convolutional image transformation network". The paper is published from Information Sciences (Elsevier). https://www.sciencedirect.com/science/article/pii/S0020025521001079

Overview

One-class classification is a machine learning problem, where training data has only one class. The objective is to determine if the input data is seen class or unseen class. Traditional deep learning algorithms are not suitable for this task since the algorithm can predict only class in training data. In this paper, the one-class classification algorithm using construction error of image transformation network (OCITN) is proposed. In particular, image transformation network (ITN) is introduced as a subtask, which transforms input image into one image, namely goal image. Moreover, the error of ITN, namely construction error (CE), is computed as a distance metric between the goal image and model output. ITN model is trained using only one-class images and is applied for testing images. Since the model is trained with only one-class images, the CE for one-class is small relative to other classes. Thus, one-class classification is made determining CE is large or small.

Please cite the following paper.

Toshitaka Hayashi, Hamido Fujita, Andres Hernandez-Matamoros, Less complexity one-class classification approach using construction error of convolutional image transformation network, Information Sciences, Volume 560, 2021, Pages 217-234, https://doi.org/10.1016/j.ins.2021.01.069

ocitn's People

Contributors

Stargazers

Watchers

ocitn's Issues

Lenna is banned from IEEE.

Please use other images as a goal image.

Pytorch Implemention

Hello ToshiHayashi san, is there any pytorch implementation so i can try it on my own dataset? thank you~

This code is hard to read.

Please wait for the update...

Question about a specific part of the code

Hello, I'm reorganizing the code to better understand the method.
However, there is a line of the code that I did not understand:

for n in range(len(x_train)):
y_train[n]=y2.rotate(rot)

It seems to me that you are putting the rotated image of Leena.png, n times (which n is the number of images in x_train) inside of a vector that should represent the labels of the training data.
In your code the y_train is also a matrix because you do hotencoder, ok, However...
If it is the case, it seems wrong because it is just adding n times the image of leena.png inside of the train labels, putting RGB values instead of labels.

Maybe you wanted to put this image inside of an X to posteriorly feed the autoencoder, isn't it?
and after it, correctly calculate the construction error regarding this image.

Thanks for sharing the code

Recommend Projects