Deep Gaussian Processes with Decoupled Inducing Inputs
This work addresses computational efficiency for researchers and practitioners using DGPs in supervised regression tasks, representing an incremental improvement.
The paper tackles the computational cost of Deep Gaussian Processes (DGP) by proposing a method that uses separate sets of pseudo points for calculating layerwise mean and variance, reducing cost without performance loss and enabling larger models with better predictive performance.
Deep Gaussian Processes (DGP) are hierarchical generalizations of Gaussian Processes (GP) that have proven to work effectively on a multiple supervised regression tasks. They combine the well calibrated uncertainty estimates of GPs with the great flexibility of multilayer models. In DGPs, given the inputs, the outputs of the layers are Gaussian distributions parameterized by their means and covariances. These layers are realized as Sparse GPs where the training data is approximated using a small set of pseudo points. In this work, we show that the computational cost of DGPs can be reduced with no loss in performance by using a separate, smaller set of pseudo points when calculating the layerwise variance while using a larger set of pseudo points when calculating the layerwise mean. This enabled us to train larger models that have lower cost and better predictive performance.