Super-Samples from Kernel Herding
Yutian Chen, Max Welling, Alex Smola
We extend the herding algorithm to continuous spaces by using the kernel trick. The resulting "kernel herding" algorithm is an infinite memory deterministic process that learns to approximate a PDF with a collection of samples. We show that kernel herding decreases the error of expectations of functions in the Hilbert space at a rate O(1/T )which ismuch faster than the usual O(1/pT) for iid random samples. We illustrate kernel herding by approximating Bayesian predictive distributions.
PDF Link: /papers/10/p109-chen.pdf
AUTHOR = "Yutian Chen
and Max Welling and Alex Smola",
TITLE = "Super-Samples from Kernel Herding",
BOOKTITLE = "Proceedings of the Twenty-Sixth Conference Annual Conference on Uncertainty in Artificial Intelligence (UAI-10)",
PUBLISHER = "AUAI Press",
ADDRESS = "Corvallis, Oregon",
YEAR = "2010",
PAGES = "109--116"