Super-Samples from Kernel Herding
Machine Learning
2012-03-19 v1 Machine Learning
Abstract
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 is much faster than the usual O(1/pT) for iid random samples. We illustrate kernel herding by approximating Bayesian predictive distributions.
Cite
@article{arxiv.1203.3472,
title = {Super-Samples from Kernel Herding},
author = {Yutian Chen and Max Welling and Alex Smola},
journal= {arXiv preprint arXiv:1203.3472},
year = {2012}
}
Comments
Appears in Proceedings of the Twenty-Sixth Conference on Uncertainty in Artificial Intelligence (UAI2010)