We explore the intersection of deep learning and probabilistic machine learning, addressing limitations of Gaussian processes in comparison to neural networks and proposing …
We introduce spherical inter-domain inducing features that yield more flexible, data-dependent basis functions for orthogonally-decoupled GP approximations, narrowing the …
We give a short and practical guide to efficiently computing the Cholesky decomposition of matrices perturbed by low-rank updates.
We extend BORE to the batch setting and establish theoretical convergence guarantees for parallel Bayesian optimization.
The 38th International Conference on Machine Learning (ICML 2021), virtual.
ELLIS AutoML Seminars (virtual).
We swap Gibbs sampling for mean-field variational inference in the Pólya-Gamma augmented model and watch the classical Jaakkola-Jordan bound on the logistic sigmoid fall out, EM …
We reformulate the computation of the acquisition function in Bayesian optimization (BO) as a probabilistic classification problem, providing advantages in scalability, …
Our paper "BORE — Bayesian Optimization by Density-Ratio Estimation" was accepted to ICML 2021 as a Long Talk (top 3% of submissions).