Jiaqian Yu 俞佳茜


Center for Visual Computing,
Université Paris-Saclay,
Grande Voie des Vignes,
92295 Châtenay-Malabry,

I completed my PhD at Center for Visual Computing at CentraleSupélec, Université Paris-Saclay, supervised by Matthew B. Blaschko. I was also a member of INRIA Galen team.

My major research interests are structured output prediction, non-modular loss functions, surrogate functions etc. I am also interested in the application of image classification, image segmentation, and medical images analysis.


I've successfully defended my PhD thesis!
I gave an oral presentation at BMVC2016!
I presented our work at BeneLearn2016!
I was visiting KU Leuven, VISICS group this summer.
I gave an oral presentation at AISTATS2016
An extended version of our ICML2015 paper is online.
I gave an oral presentation at ICML2015 on learning submodular losses with the Lovasz Hinge, at Lille France.
I gave an oral presentation at NIPS 2014 Representation and Learning Methods for Complex Outputs Workshop, at Montreal Canada.


J. Yu: Empirical risk minimization with non-modular loss functions. PhD Thesis. March 2017. [slides]
J. Yu and M. B. Blaschko: Efficient Learning for Discriminative Segmentation with Supermodular Losses. British Machine Vision Conference (BMVC), 2016. [bibtex; slidescode]
J. Yu and M. B. Blaschko: A Convex Surrogate Operator for General Non-Modular Loss Functions. International Conference on Artificial Intelligence and Statistics (AISTATS), 2016. [bibtexslidescode]
J. Yu and M. B. Blaschko: The Lovász Hinge: A Convex Surrogate for Submodular Losses. arXiv:1512.07797, 2015. [bibtex]
M. B. Blaschko and J. Yu: Hardness Results for Structured Learning and Inference with Multiple Correct Outputs. Constructive Machine Learning Workshop at ICML, 2015.[bibtex]
J. Yu and M. B. Blaschko: Learning Submodular Losses with the Lovász Hinge. In Proceedings of the International Conference on Machine Learning (ICML), 2015.[bibtex;code]
J. Yu and M. B. Blaschko: Lovasz Hinge for Learning Submodular Losses. NIPS Workshop on Representation and Learning Methods for Complex Outputs, 2014.


Fall 2016 : Foundations of Machine Learning, by Eva Zacharaki & Chloé-Agathe Azencot, at CentraleSupélec
Fall 2015 : Foundations of Machine Learning, by Chloé-Agathe Azencott, at CentraleSupélec


M.Sc. in Control Science and Engineering, 2014, Beihang University
Engineer's degree, Applied Mathematics and Systems, 2013, Ecole Centrale de Pekin
Exchange Program, Mathematics and Information, 2012, Ecole Centrale de Lyon
B.E. in Applied Mathematics, 2011, Beihang University