Discovering Latent Covariance Structures for Multiple Time Series
This work addresses the need for better multivariate time series analysis in domains like finance and manufacturing, though it appears incremental as it builds on existing compositional covariance methods.
The paper tackles the problem of analyzing multivariate time series by developing a new Gaussian Process model that uses an Indian Buffet Process prior to discover shared covariance structures, and it demonstrates improved structure discovery and predictive performance on five real-world datasets.
Analyzing multivariate time series data is important to predict future events and changes of complex systems in finance, manufacturing, and administrative decisions. The expressiveness power of Gaussian Process (GP) regression methods has been significantly improved by compositional covariance structures. In this paper, we present a new GP model which naturally handles multiple time series by placing an Indian Buffet Process (IBP) prior on the presence of shared kernels. Our selective covariance structure decomposition allows exploiting shared parameters over a set of multiple, selected time series. We also investigate the well-definedness of the models when infinite latent components are introduced. We present a pragmatic search algorithm which explores a larger structure space efficiently. Experiments conducted on five real-world data sets demonstrate that our new model outperforms existing methods in term of structure discoveries and predictive performances.