Scalable Multi-Output Gaussian Processes with Stochastic Variational Inference
About
The Multi-Output Gaussian Process is is a popular tool for modelling data from multiple sources. A typical choice to build a covariance function for a MOGP is the Linear Model of Coregionalization (LMC) which parametrically models the covariance between outputs. The Latent Variable MOGP (LV-MOGP) generalises this idea by modelling the covariance between outputs using a kernel applied to latent variables, one per output, leading to a flexible MOGP model that allows efficient generalization to new outputs with few data points. Computational complexity in LV-MOGP grows linearly with the number of outputs, which makes it unsuitable for problems with a large number of outputs. In this paper, we propose a stochastic variational inference approach for the LV-MOGP that allows mini-batches for both inputs and outputs, making computational complexity per training iteration independent of the number of outputs.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Multi-Output Gaussian Process Regression | ERA5 block-wise splitting (test) | MSE0.019 | 7 | |
| EEG Prediction | EEG held-out (test) | MSE0.366 | 7 | |
| Spatiotemporal Prediction | ERA5 (random splitting) | MSE0.014 | 7 | |
| Inverse Dynamics Prediction | SARCOS (test) | MSE0.037 | 5 | |
| gene expression prediction | Spatial Transcriptomics dataset | MSE11.024 | 4 | |
| Output extrapolation | Copernicus Marine | MSE0.035 | 4 |