Variational Gaussian Process Dynamical Systems
About
High dimensional time series are endemic in applications of machine learning such as robotics (sensor data), computational biology (gene expression data), vision (video sequences) and graphics (motion capture data). Practical nonlinear probabilistic approaches to this data are required. In this paper we introduce the variational Gaussian process dynamical system. Our work builds on recent variational approximations for Gaussian process latent variable models to allow for nonlinear dimensionality reduction simultaneously with learning a dynamical prior in the latent space. The approach also allows for the appropriate dimensionality of the latent space to be automatically determined. We demonstrate the model on a human motion capture data set and a series of high resolution video sequences.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Factor Analysis | LOOPR (in-sample) | Accuracy27.2 | 35 | |
| Response Prediction | Simulated Longitudinal Data n=10, T=30 (train) | ACC68.7 | 11 | |
| Latent Taxonomy Learning | Simulated Longitudinal Data n=10, T=30 (full) | CMD26.2 | 11 | |
| Response Prediction | Simulated Longitudinal Data n=10, T=30 (test) | ACC66.7 | 9 | |
| In-sample prediction | Longitudinal data (In-sample) | Accuracy38.2 | 8 | |
| Future Frame Prediction | CMU mocap Dataset 1 (test) | Mean MSE142.2 | 7 |