Exchange Rate Forecasting via Generative Adversarial Network (GAN)

Forecasting is an exciting topic in every field of study. Currency exchange rate Forecasting is a typical time series prediction problem which has been solved by different time-series models. Deep learning has been widely used in time series forecasting. Generative Adversarial Networks (GAN) has proved to be a powerful machine learning tool in image data analysis and generation. In this paper, we propose to use Generative Adversarial Networks (GAN) to learn and simulate Currency exchange rate Forecasting. More specifically, in this study, we attempt to forecast the dynamic evolutions of the exchange rates of Japanese Yen against the US dollar (JPYUSD), by using Generative Adversarial Networks (GAN) & Wasserstein Generative Adversarial Networks (WGAN)

I. Introduction

The foreign currency exchange market plays an important role in the global financial economy. Forecasting the Exchange Rate is a challenging topic as a time series prediction. Long short-term memory (LSTM), Gated Recurrent units (GRU) and Autoregressive Integrated Moving Average (ARIMA) are traditional models which are widely used in time series data forecasting. GANs were expanded to learn and simulate financial time series generation data [2]. Basically, Generative AI refers to deep learning models that have been a success to generate real like images and image processing. Recently, there are few studies on GAN for time series data prediction.