Modular Gaussian Processes for Transfer Learning
This work addresses transfer learning challenges in machine learning, particularly for Gaussian process models, by offering an incremental improvement that reduces computational overhead and enhances scalability.
The authors tackled the problem of transfer learning with Gaussian processes by introducing a modular variational framework that enables building ensemble models from a dictionary of pre-trained GPs without revisiting data, resulting in reduced computational costs and preserved uncertainty metrics, with extensive results demonstrating usability in large-scale and multi-task experiments.
We present a framework for transfer learning based on modular variational Gaussian processes (GP). We develop a module-based method that having a dictionary of well fitted GPs, one could build ensemble GP models without revisiting any data. Each model is characterised by its hyperparameters, pseudo-inputs and their corresponding posterior densities. Our method avoids undesired data centralisation, reduces rising computational costs and allows the transfer of learned uncertainty metrics after training. We exploit the augmentation of high-dimensional integral operators based on the Kullback-Leibler divergence between stochastic processes to introduce an efficient lower bound under all the sparse variational GPs, with different complexity and even likelihood distribution. The method is also valid for multi-output GPs, learning correlations a posteriori between independent modules. Extensive results illustrate the usability of our framework in large-scale and multi-task experiments, also compared with the exact inference methods in the literature.