I am a Phd student under the supervision of Michel DE LARA at Cermics lab at École nationale des ponts et chaussées in Champs-sur-Marne, France.
I graduated from the highly recognized engineering school ENSTA Paris in applied mathematics and from the Parisian operational research master MPRO.
My PhD subject is Algorithms in generalized convexity with application to sparse optimization.
My research topics are losses based on subdifferential representations for multiclassification in machine learning and nonconvex dual methods to solve sparse minimization problems.
Contribution
- Contribution to the package FrankWolfe.jl: implementation of a second order conditional gradient sliding method.
Talks
- Abstract Cutting Plane Method applied to Sparse Optimization
PGMODays, Palaiseau, November 28-29th 2023; - Schéma de perturbation-dualité pour la
programmation linéaire en nombres entiers (PLNE)
CNAM, Paris, January 23th 2024; - Perturbation-Duality Scheme in Combinatorial Optimization and Algorithms in Generalized Convexity
ROADEF, Amiens, March 6th 2024; - Abstract Cutting Plane Method applied to
Sparse Optimization
Journée des non permanents du Labex Bézout, Champs-sur-Marne, May 29th 2024; - Abstract Cutting Plane Method applied to
Sparse Optimization
PHD Day of MSTIC, Champs-sur-Marne, June 12th 2024; - Abstract Cutting Plane Method in Sparse Optimization
14th international symposium on generalized convexity and monotonicity, Pisa, September 2-6th 2024; - Poster session: Abstract cutting plane method for ℓ0 minimization
L0-based minimization: from continuous relaxations and non-convex algorithms to global optimization,
Paris, September 16th 2024; - Abstract Cutting Plane Method in Sparse Optimization
PGMODays, Palaiseau, November 19-20th 2024; - Learning with Fitzpatrick Losses
NEURIPS@Paris, Paris, December 4-5th 2024; - Poster session: Learning with Fitzpatrick Losses
2024 Conference on Neural Information Processing Systems, Vancouver, December 10-15th 2024; - The Perturbation-Duality Scheme in Optimization
CNAM, Paris, January 23th 2024; - Abstract Cutting Plane Method in Sparse Optimization
ROADEF, Amiens, February 26-28th 2025.
Publication & Preprint
- S. Rakotomandimby, J.-P. Chancelier, M. De Lara, and M. Blondel. Learning with Fitzpatrick losses.
In The Thirty-eighth Annual Conference on Neural Information Processing Systems, 2024. link; - S. Rakotomandimby, J.-P. Chancelier, M. De Lara, and A. Le Franc.
Subgradient selector in thegeneralized cutting plane method with an application to sparse optimization.
PREPRINT, 2024. link.