Temporal Latent Auto-Encoder: A Method for Probabilistic Multivariate Time Series Forecasting
This addresses the challenge of probabilistic forecasting for multivariate time series, which is incremental as it builds on scalable matrix factorization methods by adding nonlinearity and end-to-end training.
The paper tackles probabilistic forecasting of high-dimensional multivariate time series by introducing a temporal latent auto-encoder that enables nonlinear factorization and end-to-end learning, achieving state-of-the-art performance with gains up to 50% on standard metrics.
Probabilistic forecasting of high dimensional multivariate time series is a notoriously challenging task, both in terms of computational burden and distribution modeling. Most previous work either makes simple distribution assumptions or abandons modeling cross-series correlations. A promising line of work exploits scalable matrix factorization for latent-space forecasting, but is limited to linear embeddings, unable to model distributions, and not trainable end-to-end when using deep learning forecasting. We introduce a novel temporal latent auto-encoder method which enables nonlinear factorization of multivariate time series, learned end-to-end with a temporal deep learning latent space forecast model. By imposing a probabilistic latent space model, complex distributions of the input series are modeled via the decoder. Extensive experiments demonstrate that our model achieves state-of-the-art performance on many popular multivariate datasets, with gains sometimes as high as $50\%$ for several standard metrics.