Position-based Content Attention for Time Series Forecasting with Sequence-to-sequence RNNs

03/29/2017

∙

We propose here an extended attention model for sequence-to-sequence recurrent neural networks (RNNs) designed to capture (pseudo-)periods in time series. This extended attention model can be deployed on top of any RNN and is shown to yield state-of-the-art performance for time series forecasting on several univariate and multivariate time series.

READ FULL TEXT

Position-based Content Attention for Time Series Forecasting with Sequence-to-sequence RNNs

Foundations of Sequence-to-Sequence Modeling for Time Series

R2N2: Residual Recurrent Neural Networks for Multivariate Time Series Forecasting

A Distance Correlation-Based Approach to Characterize the Effectiveness of Recurrent Neural Networks for Time Series Forecasting

RECOWNs: Probabilistic Circuits for Trustworthy Time Series Forecasting

Topological Attention for Time Series Forecasting

A self-organising eigenspace map for time series clustering

Fast ES-RNN: A GPU Implementation of the ES-RNN Algorithm

Position-based Content Attention for Time Series Forecasting with Sequence-to-sequence RNNs

Related Research

Foundations of Sequence-to-Sequence Modeling for Time Series

R2N2: Residual Recurrent Neural Networks for Multivariate Time Series Forecasting

A Distance Correlation-Based Approach to Characterize the Effectiveness of Recurrent Neural Networks for Time Series Forecasting

RECOWNs: Probabilistic Circuits for Trustworthy Time Series Forecasting

Topological Attention for Time Series Forecasting

A self-organising eigenspace map for time series clustering

Fast ES-RNN: A GPU Implementation of the ES-RNN Algorithm