Miguel Lazaro-Gredilla Homepage
Code for "Doubly Stochastic Variational Bayes for non-Conjugate Inference"
Code for "Overlapping Mixtures of Gaussian Processes for the data association problem"
Code for "A Bayesian Approach to Tracking with Kernel Recursive Least-Squares"
- Code and usage demo for the KRLS Tracker, including several pruning methods:
-
KL divergence using marginals only
-
Full KL divergence (expensive!)
-
MSE at the dictionary bases (fast, default option, described in the paper)
-
Integrated MSE at the whole input space (analytic for Gaussian kernel, expensive)
-
RKHS norm of the error
Code for "Variational Heteroscedastic Gaussian Process Regression"
Code for "Inter-Domain Gaussian Processes"
- Coming soon, in both FITC and VAR flavours.
Code for "Marginalized Neural Networks Mixtures for Large-Scale Regression"
Code for "Sparse Spectrum Gaussian Process Regression"
Note: Code is provided for research purposes only. Contact me for more information, comments, bugs, etc.