NeurIPS 2020

Deep Rao-Blackwellised Particle Filters for Time Series Forecasting

Meta Review

This paper proposes an extension to state-switching Gaussian linear systems, which extends the ideas of traditional Kalman filtering to the setting of switching states. Strengths: - the use of recurrent switch transitions through Gaussian switch variables is interesting - mathematical elegance is appealing Weaknesses: - It is unclear if the proposed techniques can be scaled to larger models - video experiments could be strengthened