I am a postdoctoral fellow in the Department of Statistics at the University of Washington working with the NSF-TRIPODS ADSI members Zaid Harchaoui, Dmitriy Drusvyatskiy, Maryam Fazel, Sham Kakade, Yin Tat Lee.

I got my Ph.D. from École Normale Supérieure Ulm under the supervision of Alexandre d'Aspremont within the Sierra Team led by Francis Bach.

During my thesis, I worked on mathematical optimization approaches for data science problems with an underlying combinatorial structure. I am now working on dynamic optimization problems for data science.

More details can be found in my resume.

Phd Thesis

On the geometry of optimization problems and their structure
Vincent Roulet, supervised by Alexandre d'Aspremont.
École normale supérieure de Paris
[manuscript]

Papers

A smoother way to train structured prediction
Krishna Pillutla, Vincent Roulet, Sham Kakade and Zaid Harchaoui.
To appear in advances in Neural Information Processing Systems 31 (NIPS2018)
Sharpness, Restart and Acceleration
Vincent Roulet and Alexandre d'Aspremont.
Advances in Neural Information Processing Systems 30 (NIPS2017)
[paper]
Integration methods and Accelerated Optimization Algortihms
Damien Scieur, Vincent Roulet, Francis Bach and Alexandre d'Aspremont.
Advances in Neural Information Processing Systems 30 (NIPS2017)
[paper]
Iterative Hard Clustering of Features
Vincent Roulet, Fajwel Fogel, Alexandre d'Aspremont and Francis Bach.
[arxiv]
Complexity verus Statistical Performance on Sparse Recovery Problems
Vincent Roulet, Nicolas Boumal and Alexandre d'Aspremont.
To appear in Information and Inference: a Journal of the IMA
[arxiv]

Talks

February 2017, Sharpness, Restart, Acceleration, Optimization and Satistical Learning Workshop, Les Houches.
Decmeber 2015, Supervised Clustering in the Data Cube, Transfer and Multi-Task Learning Workshop NIPS2015.

Teaching

2014-2017 : Teaching assistant, Convex Optimization , Master Mathematics, Machine Learning and Computer Vision (MVA).
2013-2014 : Oral interrogations in "Classes préparatoires" (MP*).