Journal Publications
Robust and Reliable Data-Driven Framework for Quality-Related Knowledge Discovery in Large-Scale Rubber Mixing Batch Processes
Submitted to Journal of Cheminformatics
This work introduces an explainable and trustworthy machine learning framework for quality prediction in rubber mixing, combining XGBoost, explainability, and uncertainty quantification. The approach uncovers key process insights, reduces input complexity by 82%, and achieves high predictive accuracy — validated on real-world Michelin data.
Detecting fast-ripples on both micro- and macro-electrodes in epilepsy: A wavelet-based CNN detector
Journal of Neuroscience Methods
We present a novel approach for detecting fast-ripples in epilepsy patients using Fourier Analysis-based Scalogram techniques, potentially improving the diagnosis and treatment of epilepsy.
Conference Proceedings
Knowledge Discovery in Large-Scale Batch Processes through Explainable Boosted Models and Uncertainty Quantification: Application to Rubber Mixing
Ghent, Belgium
ESCAPE 35 - European Symposium on Computer Aided Process Engineering
We developed a novel 2D Semi Bi-Gaussian (2DSBG) filter for detecting adjacent and multi-scale line features in images, with applications in computer vision and image processing.
2DSBG: A 2d Semi Bi-Gaussian Filter Adapted for Adjacent and Multi-Scale Line Feature Detection
Rhodes, Greece
ICASSP 2023 - IEEE International Conference on Acoustics, Speech and Signal Processing
We developed a novel 2D Semi Bi-Gaussian (2DSBG) filter for detecting adjacent and multi-scale line features in images, with applications in computer vision and image processing.
Science Communication & Talks
I present my research at academic conferences, seminars, and science outreach events. Below is a selection of my talks, organized by type.
Ma Thèse en 180 secondes
"Ma Thèse en 180 secondes" challenges PhD students to present their research in just 180 seconds to a non-specialist audience. This competition showcases the ability to explain complex scientific concepts in an engaging and accessible way.
Winner of the Public's Choice Award in the local competition, earning selection to represent IP Paris at the national finals with Amelie Kies.
I presented my research on Robust Real-time Quality Monitoring and Reliable Adaptation of Manufacturing Processes in the Tire Industry.
Conference Presentations
ESCAPE 35
Selected to deliver an oral presentation at the European Symposium on Computer Aided Process Engineering (ESCAPE), one of the premier European conferences in process engineering.
Seminars & Paper Presentations
Research Seminar Founder & Organizer
Founded and currently lead a weekly research seminar bringing together researchers and data scientists from various teams within Michelin. This initiative fosters knowledge sharing and collaboration across departments, creating a vibrant research community.
The seminar focuses on cutting-edge papers in machine learning, computer vision, natural language processing, and uncertainty quantification, with direct applications to industrial challenges in the tire manufacturing domain.
Selected Paper Presentations
Posters
Knowledge Discovery in Large-Scale Batch Processes
Paris, France
Welcome Day IP Paris
Presented research on explainability for knowledge discovery in large-scale batch processes with a focus on uncertainty quantification.
Awarded Best Poster in the "Mathematics" category.
A Framework for Knowledge Discovery in Rubber Mixing Processes
Clermont-Ferrand, France
Michelin Doctoral Day
Presented research dealing with explainability and uncertainty quantification using machine learning models considering offline batch processes.
2DSBG: A 2d Semi Bi-Gaussian Filter for Line Feature Detection
Rhodes, Greece
ICASSP 2023 - IEEE International Conference on Acoustics, Speech and Signal Processing
Presented research on a novel approach to line feature detection in images using 2D Semi Bi-Gaussian filters.