ML Anthology

A Kernel Approach to Comparing Distributions

Arthur Gretton, Karsten M. Borgwardt, Malte J. Rasch, Bernhard Schölkopf, Alexander J. Smola
AAAI 2007 pp. 1637-1641
/aaai/2007/gretton2007aaai-kernel/

Abstract

We describe a technique for comparing distributions without\nthe need for density estimation as an intermediate step.\nOur approach relies on mapping the distributions into a Reproducing\nKernel Hilbert Space. We apply this technique to\nconstruct a two-sample test, which is used for determining\nwhether two sets of observations arise from the same distribution.\nWe use this test in attribute matching for databases using\nthe Hungarian marriage method, where it performs strongly.\nWe also demonstrate excellent performance when comparing\ndistributions over graphs, for which no alternative tests currently\nexist.

PDF AAAI Semantic Scholar

Cite

Text

Gretton et al. "A Kernel Approach to Comparing Distributions." AAAI Conference on Artificial Intelligence, 2007.

Markdown

[Gretton et al. "A Kernel Approach to Comparing Distributions." AAAI Conference on Artificial Intelligence, 2007.](https://mlanthology.org/aaai/2007/gretton2007aaai-kernel/)

BibTeX

@inproceedings{gretton2007aaai-kernel,
  title     = {{A Kernel Approach to Comparing Distributions}},
  author    = {Gretton, Arthur and Borgwardt, Karsten M. and Rasch, Malte J. and Schölkopf, Bernhard and Smola, Alexander J.},
  booktitle = {AAAI Conference on Artificial Intelligence},
  year      = {2007},
  pages     = {1637-1641},
  url       = {https://mlanthology.org/aaai/2007/gretton2007aaai-kernel/}
}