Timezone: »

 
Conditional Densities and Efficient Models in Infinite Exponential Families
Arthur Gretton

Fri Dec 08 04:00 PM -- 04:35 PM (PST) @
in Advances in Modeling and Learning Interactions from Complex Data »

The exponential family is one of the most powerful and widely used classes of models in statistics. A method was recently developed to fit this model when the natural parameter and sufficient statistic are infinite dimensional, using a score matching approach. The infinite exponential family is a natural generalisation of the finite case, much like the Gaussian and Dirichlet processes generalise their respective finite modfels. In this talk, I'll describe two recent results which make this model more applicable in practice, by reducing the computational burden and improving performance for high-dimensional data. The firsrt is a Nytsrom-like approximation to the full solution. We prove that this approximate solution has the same consistency and convergence rates as the full-rank solution (exactly in Fisher distance, and nearly in other distances), with guarantees on the degree of cost and storage reduction. The second result is a generalisation of the model family to the conditional case, again with consistency guarantees. In experiments, the conditional model generally outperforms a competing approach with consistency guarantees, and is competitive with a deep conditional density model on datasets that exhibit abrupt transitions and heteroscedasticity.

Author Information

Arthur Gretton (Gatsby Unit, UCL)

Arthur Gretton is a Professor with the Gatsby Computational Neuroscience Unit at UCL. He received degrees in Physics and Systems Engineering from the Australian National University, and a PhD with Microsoft Research and the Signal Processing and Communications Laboratory at the University of Cambridge. He previously worked at the MPI for Biological Cybernetics, and at the Machine Learning Department, Carnegie Mellon University. Arthur's recent research interests in machine learning include the design and training of generative models, both implicit (e.g. GANs) and explicit (high/infinite dimensional exponential family models), nonparametric hypothesis testing, and kernel methods. He has been an associate editor at IEEE Transactions on Pattern Analysis and Machine Intelligence from 2009 to 2013, an Action Editor for JMLR since April 2013, an Area Chair for NeurIPS in 2008 and 2009, a Senior Area Chair for NeurIPS in 2018, an Area Chair for ICML in 2011 and 2012, and a member of the COLT Program Committee in 2013. Arthur was program chair for AISTATS in 2016 (with Christian Robert), tutorials chair for ICML 2018 (with Ruslan Salakhutdinov), workshops chair for ICML 2019 (with Honglak Lee), program chair for the Dali workshop in 2019 (with Krikamol Muandet and Shakir Mohammed), and co-organsier of the Machine Learning Summer School 2019 in London (with Marc Deisenroth).

More from the Same Authors