Data Generation Using Gene Expression Generator

EasyChair Preprint 3948

Abstract

Generative adversarial networks (GANs) could be used efficiently for image and video generation, where labeled training data are available in bulk. In general, building a good machine learning model requires a reasonable amount of labeled training data. However, there are areas such that the biomedical field where the creation of such a data set is time-consuming and requires expert knowledge. Our goal is to use data augmentation techniques as an alternative to data collection to improve data classification. We propose the use of a modified version of GAN named Gene Expression Generator (GEG) to augment data samples at hand. The proposed approach was used to generate synthetic data for binary biomedical data sets to trains existing supervised machine learning approaches. Experimental results showed that using GEG for data augmentation with a modified version of leave one out cross-validation increased the performance of classification accuracy.

Keyphrases: Cancer Classification, Generative Adversarial Networks, adversarial network, breast cancer dataset, classification accuracy, colon cancer dataset, cross-validation, data augmentation, data augmentation technique, data generation, gene expression data, machine learning, synthetic data

@booklet{EasyChair:3948,
  author    = {Zakarya Farou and Noureddine Mouhoub and Tomáš Horváth},
  title     = {Data Generation Using Gene Expression Generator},
  howpublished = {EasyChair Preprint 3948},
  year      = {EasyChair, 2020}}