第25回統計的機械学習セミナー / The 25th Statistical Machine Learning Seminar

Place: 統計数理研究所　セミナー室5
/ Seminar Room 5 @ The Institute of Statistical Mathematics

Title: The Generalized Kernel Kalman Filter - Learning Forward Models from High Dimensional Observations

Abstract: Learning forward models from high-dimensional partial observations of the real state is a challeng- ing machine learning problem. Recently, non- parametric inference methods have been pro- posed to tackle such problems. However, such methods either do not provide an uncertainty esti- mate, are computationally expensive, or can only be applied to a limited set of problems. We gen- eralize the formulation of Kalman Filters (KF) embeddings into a reproducing kernel Hilbert space (RKHS) to be applicable to systems with high-dimensional, partial observations. Our for- mulation provides probabilistic state estimations and predictions for non-linear dynamical systems that can also be directly learned from the obser- vations. Additionally, we propose an alternative formulation of the RKHS embedding of a con- ditional density that allows to learn from large data sets, while maintaining computational effi- ciency. We show on a nonlinear state estimation task with high dimensional observations that our approach provides an improved estimation accu- racy.