Comparing linear structure-based and data-driven latent spatial representations for sequence prediction
This work addresses the challenge of joint time and spatial dependency modeling for GTS prediction, which is incremental as it compares existing linear methods rather than introducing new ones.
The paper compared linear structure-based and data-driven latent spatial representations for predicting Graph-supported Time Series (GTS) in domains like brain activity and videos, finding that data-driven methods generally outperformed structure-based ones in terms of prediction accuracy, with improvements of up to 15% on certain datasets.
Predicting the future of Graph-supported Time Series (GTS) is a key challenge in many domains, such as climate monitoring, finance or neuroimaging. Yet it is a highly difficult problem as it requires to account jointly for time and graph (spatial) dependencies. To simplify this process, it is common to use a two-step procedure in which spatial and time dependencies are dealt with separately. In this paper, we are interested in comparing various linear spatial representations, namely structure-based ones and data-driven ones, in terms of how they help predict the future of GTS. To that end, we perform experiments with various datasets including spontaneous brain activity and raw videos.