Latent Gaussian Process Regression
This work addresses the limitation of standard GPs in handling complex, non-stationary data for researchers in fields like robotics and spatial statistics, though it appears incremental as an extension of existing GP methods.
The authors tackled the problem of modeling non-stationary multi-modal processes with Gaussian Processes by introducing a latent variable extension that modulates the covariance function, and they demonstrated its application on synthetic and real-world datasets like motion capture and geostatistics.
We introduce Latent Gaussian Process Regression which is a latent variable extension allowing modelling of non-stationary multi-modal processes using GPs. The approach is built on extending the input space of a regression problem with a latent variable that is used to modulate the covariance function over the training data. We show how our approach can be used to model multi-modal and non-stationary processes. We exemplify the approach on a set of synthetic data and provide results on real data from motion capture and geostatistics.