Contact information



University of Cambridge Home

Department of Engineering
Computational and Biological Learning Lab

University of Cambridge >  Department of Engineering >  Information Engineering  >  Computational and Biological Learning Lab  > Sébastien Bratières

Sébastien Bratières

I am a member of the Machine Learning group of the Cambridge University Engineering Department, led by Zoubin Ghahramani.

My research interests lie in probabilistic machine learning, an approach to machine learning in which correct modelling of uncertainty is important. This plays a role in obtaining an estimate of "error bars" in predictions, i.e. evaluating predictive uncertainty, or in active learning, where the exploration/ exploitation trade-off is influenced by how much uncertainty can be removed by exploring a new portion of the search space. More specifically, I work with non-parametric Bayesian models, such as the infinite HMM (a Dirichlet process based model) or the Gaussian process. My thesis is on "Non-parametric Bayesian models for structured output prediction": I apply these models to structured data such as chains or grids to carry out structured classification, for which an example task is part-of-speech tagging on a sequence of words: each tag depends on neighbour tags and words.

I am interested in applications to natural language processing, speech recognition, dialogue systems.

I received my “Ingénieur” degree from Ecole Centrale Paris, and an MPhil in Computer Speech and Language Processing from the University of Cambridge in 2001. I have been working in the speech industry ever since, and started my PhD on a part-time basis in 2009. More information on my career, past and present, is available on my Linkedin profile.

News

July 2017: I have just joined Translated on Pi Campus (venture fund and startup district) as Director of AI. I'm opening a School of AI inside Pi School this Fall.

April 2017: At long last, I have submitted my thesis !

October 2014: Our work (Bratières, Quadrianto, Ghahramani) on GPstruct has been accepted for publication in the IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI). This is a reworked version of our arXiv preprint. TPAMI Final draft.

August 2014: The source code for GPstruct, pygpstruct, is online..

June 2014: Our work (jointly with Novi Quadrianto, Sebastian Nowozin and Zoubin Ghahramani) on GPstruct applied to images will be presented at ICML 2014.

July 2013: Our work (jointly with Novi Quadrianto and Zoubin Ghahramani) on GPstruct is presented in an arXiv preprint.

September 2010: I received another AWS in Education Research Grant for my project A large-scale infinite HMM trained on raw text. This grant comes as a credit towards Amazon Web Services, such as Amazon Elastic Compute Cloud.

August 2010: I have just been awarded the Yahoo! Key Scientfic Challenge Award in the machine learning area. Here is the press release.

August 2009: Jurgen van Gael, Andreas Vlachos and I received an AWS in Education Research Grant to carry out work with Elastic MapReduce, Elastic Compute Cloud, and Simple Storage Service, with the infinite HMM.

Code

The source code for GPstruct, pygpstruct, is now online.

Publications

Presentations

Errata lists

Errata for Sebastian Nowozin and Christoph H. Lampert (2011), "Structured Learning and Prediction in Computer Vision".

Errata for Kevin Murphy (2012), "Machine Learning, a Probabilistic Perspective".

Contact details

E-mail: sb358 usual-symbol cam dot ac and then dot uk


© University of Cambridge, Department of Engineering
Information provided by Sébastien Bratières