Igor Colin
Researcher
I am a machine learning researcher in S2A team at Télécom Paris. Previously, I was a research in Noah's Ark research team at Huawei. During 2017, I was a postdoctoral researcher at INRIA, SIERRA under the supervision of Alexandre d'Aspremont. From 2013 to 2016, I was a Ph.D. student, co-advised by Stephan Clémençon and Joseph Salmon. In 2013, I graduated the Master MVA (machine learning and computer vision) at ENS Cachan and obtained the engineering degree of École des Ponts ParisTech.
Research Interests
- Large-scale machine learning
- Distributed/decentralized methods
- Graphs
- Statistical learning theory
- Non-convex optimization
- Bandits
Research Papers
Parallel Contextual Bandits in Wireless Handover Optimization
In ICDM 2018, DaMNet.
[PDF]
Stable Bounds on the Duality Gap of Finite Sum Minimization Problems
[PDF]
Decentralized Topic Modelling with Latent Dirichlet Allocation
In NIPS 2016, Workshop on Private Multi‑Party Machine Learning.
[Hal]
Adapting Machine Learning Techniques to U-statistics
Ph.D. Thesis.
[PDF]
Scaling-up Empirical Risk Minimization: Optimization of Incomplete U-statistics
In JMLR.
Gossip Dual Averaging for Decentralized Optimization of Pairwise Functions
In ICML 2016.
Extending Gossip Algorithms to Distributed Estimation of U-Statistics
In NIPS 2015, selected for spotlight.
Learning information cascades in social networks
Master's Thesis, 2013.
[PDF]
Teaching
- Master MASH: projets informatiques, at Université Paris Dauphine.
- Data science : introduction au machine learning, at Télécom Évolution.
- Machine Learning master's: practical work, at Télécom ParisTech.
Ph.D. Students
- Xiaoling Zhu, co-directed with Kevin Scaman and Richard Combes.
- Geovani Rizk, co-directed with Albert Thomas, Yann Chevaleyre and Rida Laraki.
Miscellaneous
- Organizer of Distributed Machine Learning Workshop, at Télécom ParisTech