Multi-Class Gaussian Process Classification Made Conjugate: Efficient Inference via Data Augmentation
arXiv:1905.09670
Abstract
We propose a new scalable multi-class Gaussian process classification approach building on a novel modified softmax likelihood function. The new likelihood has two benefits: it leads to well-calibrated uncertainty estimates and allows for an efficient latent variable augmentation. The augmented model has the advantage that it is conditionally conjugate leading to a fast variational inference method via block coordinate ascent updates. Previous approaches suffered from a trade-off between uncertainty calibration and speed. Our experiments show that our method leads to well-calibrated uncertainty estimates and competitive predictive performance while being up to two orders faster than the state of the art.
Accepted at UAI 2019
References in corpus (5)
- Scalable Variational Gaussian Process Classification
- Dependent Multinomial Models Made Easy: Stick Breaking with the Pólya-Gamma Augmentation
- MCMC for Variationally Sparse Gaussian Processes
- One-vs-Each Approximation to Softmax for Scalable Estimation of Probabilities
- Inverse Ising problem in continuous time: A latent variable approach