Sparse Orthogonal Variational Inference for Gaussian Processes
This work improves scalability for Gaussian process models, which is a key bottleneck in machine learning, though it appears incremental as it builds on existing sparse approximations.
The authors tackled the scalability of Gaussian process models by introducing a sparse variational approximation based on decomposing the process into inducing points and residual variation, resulting in tighter lower bounds and new algorithms that achieved state-of-the-art results on CIFAR-10 among GP-based models.
We introduce a new interpretation of sparse variational approximations for Gaussian processes using inducing points, which can lead to more scalable algorithms than previous methods. It is based on decomposing a Gaussian process as a sum of two independent processes: one spanned by a finite basis of inducing points and the other capturing the remaining variation. We show that this formulation recovers existing approximations and at the same time allows to obtain tighter lower bounds on the marginal likelihood and new stochastic variational inference algorithms. We demonstrate the efficiency of these algorithms in several Gaussian process models ranging from standard regression to multi-class classification using (deep) convolutional Gaussian processes and report state-of-the-art results on CIFAR-10 among purely GP-based models.