I am a research associate working with Arthur Gretton
at the Gatsby Computational Neuroscience Unit.
My main research interests are information theory (consistent estimators, distribution regression; ITE toolbox
), kernel methods and dictionary learning problems (structured sparse tasks, independent subspace analysis
and its extensions). I am also working on applications including remote sensing, blind signal separation, collaborative filtering,
natural language processing and facial expression recognition.
- I will be visiting Prof. Bernhard Schölkopf's lab in Tübingen. [Jan. 14-18, 2015]
- We had an external review with a poster session [poster-1, poster-2; Oct. 29, 2014].
- Two-stage Sampled Learning Theory on Distributions: the new version is on arXiv. [abstract, paper, code; Oct. 28, 2014]
- Bayesian Manifold Learning: Locally Linear Latent Variable Model (LL-LVM): made available on arXiv. [abstract, paper; Oct. 27, 2014]
- Consistent Vector-valued Distribution Regression: accepted at the 1st UCL Workshop on the Theory of Big Data [abstract; Oct. 21, 2015]
- Vector-valued Distribution Regression: A Simple and Consistent Approach. Invited talk at Statistical Science Seminars. [abstract, slides; Oct. 9, 2014]
- Modern Nonparametrics 3: Automating the Learning Pipeline: our NIPS-2014 workshop proposal has been accepted. [Aug. 30, 2014]
- Simple Consistent Distribution Regression on Compact Metric Domains: accepted at UCL-Duke Workshop on Sensing and Analysis of High-Dimensional Data (SAHD-2014). [poster; Sept. 4-5, 2014; abstract, code; July 1, 2014]
- Spatio-temporal event classification using time-series kernel based structured sparsity: accepted at ECCV-2014. [poster; Sept. 6-12, 2014; paper, supplementary material, video demo; June 15, 2014]
- Two-stage Sampled Learning Theory on Distributions: available on arXiv. [abstract, paper, code; June 7, 2014]
- I started to organize the machine learning external seminars at Gatsby. [May, 2014]
- Distribution Regression - the Set Kernel Heuristic is Consistent: invited talk at CSML Lunch Talk Series. [slides, paper, code; May 2, 2014]
- Learning on Distributions: invited talk at Kernel methods for big data workshop, Lille. [slides, paper, code; Apr. 2, 2014]
- Information Theoretical Estimators (ITE) toolbox: appeared in JMLR. [abstract, paper; Mar. 6, 2014]
- Consistent Distribution Regression via Mean Embedding: invited talk at the University of Hertfordshire. [slides, paper, code; Mar. 5, 2014]
- Consistent, Two-Stage Sampled Distribution Regression: available on arXiv. [abstract, paper; Feb. 7, 2014]