Mixture Density Conditional Generative Adversarial Network Models (MD-CGAN)
This addresses the limited application of GANs to time series forecasting, offering a method for non-Gaussian predictions, though it appears incremental as it builds on existing GAN and mixture model techniques.
The paper tackles time series forecasting by proposing the Mixture Density Conditional Generative Adversarial Model (MD-CGAN), which estimates a probabilistic posterior distribution over forecasts and performs well compared to benchmarks, especially in noisy scenarios.
Generative Adversarial Networks (GANs) have gained significant attention in recent years, with impressive applications highlighted in computer vision in particular. Compared to such examples, however, there have been more limited applications of GANs to time series modelling, including forecasting. In this work, we present the Mixture Density Conditional Generative Adversarial Model (MD-CGAN), with a focus on time series forecasting. We show that our model is capable of estimating a probabilistic posterior distribution over forecasts and that, in comparison to a set of benchmark methods, the MD-CGAN model performs well, particularly in situations where noise is a significant component of the observed time series. Further, by using a Gaussian mixture model as the output distribution, MD-CGAN offers posterior predictions that are non-Gaussian.