Hello, my name is
Lucas
Gnecco Heredia
And I'm a
Contact

About me

I am a third year Ph.D student at LAMSADE, Université Paris Dauphine - PSL, where I work under the supervision of of Yann Chevaleyre and Benjamin Negrevergne

I am interested in achieving robustness to adversarial attacks by randomization, particularly mixtures of models, both in theory and practice.

Mixtures of models are like ensembles, but instead of aggregating the decision of every model in a deterministic way, the final decision is taken by first sampling one of the models and using it to predict.

These models are stochastic in nature and therefore harder to attack in the context of adversarial attacks. My goal is to better understand mixtures, how to attack them and ultimately how to build robust mixtures to adversarial attacks.

Publications

Teaching

2023 - 2024
Algorithmique et programmation 3 - L2
TD / TP. Bases of algorithms and complexity.
Fondamentaux du Machine Learning - L3
TD / TP. Mathematical bases for machine learning. Linear algebra and estimation statistics.
2022 - 2023
Fondamentaux du Machine Learning - L3
TD / TP. Mathematical bases for machine learning. Linear algebra and estimation statistics.
Introduction au Machine Learning - M1 MIAGE App
CM / TD. Introductory notions of machine learning. Classification algorithms, clustering and model evaluation.
2021 - 2022
Semi-Structured Data Bases - L3
TP. XML, DOM, RDF and JSON. Practical exercises in Python Notebooks
Algorithmique et Programmation 2 - L1
TD / TP. Basic notions of algorithms and coplexity. Sorting and data structures like binary trees and heaps
Outils en Informatique - L1 (MS Access, MS Excel)
TP. Practical exercises for first year students in MS Access and Excel.

Misc

Grade calculator
Metro Trivia Game - Single player

Contact me

Get in Touch

Name
Lucas Gnecco Heredia
Address
Paris, France
Email
lucas.gnecco-heredia[at]dauphine.psl.eu
Github
lucasgneccoh