Multilinear Dynamical System for Tensor Time Series

This package provides matlab implementation of MLDS (also called Tensor Kalman Filter). MLDS is an extension of traditional linear dynamical systems, also known as Kalman filters. It replaces the states and observations to a sequence of tensors instead of vectors. Therefore it is able to handle more complex time series data (e.g. a video clip or graph time series). It is based on the dynammo package for learning and mining with linear dynamical systems.

1. to run the example on synthetic data

  make demo

2. to run mlds on SST data set

  matlab -r demo_sst.m

Reference

  Multilinear Dynamical Systems for Tensor Time Series,
  Mark Rogers, Lei Li, and Stuart J. Russell.
  In the 27th Conference on Neural Information Processing Systems(NeurIPS) , 2013.

Code is implemented by Mark Rogers and Lei Li.