GT-GAN: General Purpose Time Series Synthesis with Generative Adversarial Networks

Time series synthesis is an important research topic in the field of deep learning, which can be used for data augmentation. Time series data types can be broadly classified into regular or irregular. However, there are no existing generative models that show good performance for both types without any model changes. Therefore, we present a general purpose model capable of synthesizing regular and irregular time series data. To our knowledge, we are the first designing a general purpose time series synthesis model, which is one of the most challenging settings for time series synthesis. To this end, we design a generative adversarial network-based method, where many related techniques are carefully integrated into a single framework, ranging from neural ordinary/controlled differential equations to continuous time-flow processes. Our method outperforms all existing methods. Related work and preliminaries GANs are one of the most representative generative technology. Ever since the first introduction in its seminal research paper, GANs have been adopted to main different domains. Recently, researchers focused on their synthesis for time series data. Therefore, there have been proposed several GANs for time series synthesis. C-RNN-GAN [Mogren, 2016] has a regular GAN framework that can be applied to sequential data by using LSTM in its generator and discriminator. Recurrent Conditional GAN (RCGAN [Esteban et al., 2017]) took a similar approach except that its generator and discriminator take conditional input for better synthesis. WaveNet [van den Oord et al., 2016 ] also generates time series data from the conditional probability of previous data by using the dilated casual convolution.