Code coming soon

Modeling Continuous Stochastic Processes with Dynamic Normalizing Flows

Accepted to NeurIPS 2020. We propose a novel type of normalizing flow that transforms Wiener processes and use it to model irregular time series data.