TS-6: Deep learning for time series - sequences

Abstract: Time series are, first and foremost, sequences - so it's only natural to apply sequence modeling approach from deep learning to such problems. In this episode we present the vintage DL methods (RNN, GRU and LSTM) and show their applications for single- and multistep forecasting. We also take time to explore the connection RNN have to curve fitting - via the ES-RNN model.

deep learning series time series