Conditional and Adversarial Euler-based Generators for Time Series

It is demonstrated empirically that CEGEN outperforms state-of-the-art and GAN generators on both marginal and temporal dynamics metrics, and it identifies accurate correlation structures in high dimension.