I am PhD Student in the UCL Department of Computer Science, and the The Gatsby Computational Neuroscience Unit, part of the Centre for Computational Statistics and Machine Learning. My PhD is founded by Centre for Doctoral Training in Financial Computing & Analytics and partially The Gatsby Computational Neuroscience Unit. My research supervisor is Arthur Gretton. My research interests include machine learning and statistics, in particular

June 2015: article on analytic mean embeddings, used to construct almost surely consistent and lineartime twosample test. Available here.
November 2014: I give a presentation on the article"A Wild Bootstrap for Degenerate Kernel Tests." during the NIPS 2014 conference. The details are available here.
September 2014: the video from my presentation of the "A Kernel Independence Test for Random Processes." is available here.
Chwialkowski, Kacper, Dino Sejdinovic, and Arthur Gretton. "Fast TwoSample Testing with Analytic Representations of Probability Measures." Under review. Article, code.
Chwialkowski, Kacper, Dino Sejdinovic, and Arthur Gretton. "A Wild Bootstrap for Degenerate Kernel Tests." In NIPS 2014 (oral presentation). Article, code.
Chwialkowski, Kacper, and Arthur Gretton. "A Kernel Independence Test for Random Processes." In ICML 2014: Proceedings of 31st International Conference on Machine Learning. Article, code.
Reviewer for journals
Journal of Machine Learning Research
ACM Transactions on Intelligent Systems and Technologie special issue on Causality
BMC Medical Research Methodology
Statistics and Computing
Reviewer for Conference
NIPS 2015
COLT 2015
Teaching assistant in the course Advanced Topics in Machine Learning.
A test of independence for random processes . In this talk I explain the asymptotic behavior of Hilbert Schmidt Independence Criterion when computed from samples drawn from random processes. I explain why an earlier bootstrap procedures which workes in the i.i.d. case fails for random processes, and why by shifting one random process with respect to another the correct estimate of null hypothesis can be generated. Talk
email: kacper.chwialkowski@gmail.com