Design and analysis of generative adversarial...

Design and analysis of generative adversarial networks for time series data augmentation

Author Name : Nikhil Dupally, Hrithik Puppala, Kuppuri Varun Sai Reddy

ABSTRACT
Having a large number of data points is essential for any machine learning activity.  It is fairly unusual for datasets to be tiny or uneven due to the difficulty and expense of collecting information. Researchers in a variety of sectors are increasingly turning to synthetic data production as a solution to the difficulties they have in collecting data. Due to their vast range of applications and effective results, Generative Adversarial Networks (GANs) have been a popular technique for generating data in the last decade. Data generated by a modified TimeGAN architecture was trained and assessed in this research study. It was shown that the model was able to learn about the data's distribution and its underlying information and successfully generate synthetic data.

Keywords: generative adversarial network, synthetic data, time-series data, recurrent neural network, gated recurrent unit