Bayesian Nonparametrics

The study of complex phenomena through the analysis of data often requires us to make assumptions about the underlying dynamics. In modern applications of data science, we are facing the challenge of doing so when very little is known about the mechanistic description of many systems of interest; even when we do, the complexity of simulating such systems is so that we can’t use it effectively. Probabilistic nonparametric models offer a principled way to analyze data so as to interpret the system under study as well as quantify the confidence in predictions. While probabilistic nonparametric models are attractive from a theoretical point of view, they pose some serious computational challenges. Part of the research at EURECOM is devoted to solve or alleviate these problems with the ultimate goal of enabling practical and scalable nonparametric models to tackle important scientific questions in environmental and life sciences.

Here is a list of research topics in this area:

  • Gaussian Processes and Deep Gaussian Processes

  • Dirichlet Process Mixture Models

  • Stochastic and Distributed Linear Algebra

  • Stochastic Variational Inference

  • Markov chain Monte Carlo for Probabilistic Nonparametric Models


  • K. Cutajar, E. V. Bonilla, P. Michiardi, and M. Filippone. Practical learning of deep Gaussian processes via random Fourier features, 2016. arXiv:1610.04386. [ bib | code | pdf | http ]

  • K. Krauth, E. V. Bonilla, K. Cutajar, and M. Filippone. AutoGP: Exploring the capabilities and limitations of Gaussian process models, 2016. arXiv:1610.05392. [ bib | pdf | http ]

  • K. Cutajar, M. A. Osborne, J. P. Cunningham, and M. Filippone. Preconditioning kernel matrices. In Proceedings of the 32nd International Conference on Machine Learning, ICML 2016, New York City, USA, June 19-24, 2016, 2016. [ bib | code | pdf ]

  • J. Hensman, A. G. de G. Matthews, M. Filippone, and Z. Ghahramani. MCMC for variationally sparse Gaussian processes. In Advances in Neural Information Processing Systems 28: Annual Conference on Neural Information Processing Systems 2015, December 7-12 2015, Montreal, Quebec, Canada, 2015. [ bib | code | pdf ]

  • M. Filippone and R. Engler. Enabling scalable stochastic gradient-based inference for Gaussian processes by employing the Unbiased LInear System SolvEr (ULISSE). In Proceedings of the 32nd International Conference on Machine Learning, ICML 2015, Lille, France, July 6-11, 2015, 2015. [ bib | code | pdf ]

  • M. Filippone and M. Girolami. Pseudo-marginal Bayesian inference for Gaussian processes. IEEE Transactions on Pattern Analysis and Machine Intelligence, 36(11):2214-2226, 2014. [ bib | code | pdf | http ]


Syndicate content

Data Science