Research

Research interests

I am currently co-head of the Machine Learning & Data Mining Group in the LIRIS laboratory (LIRIS, UMR 5205 CNRS) and I am in charge of the Master of Science in Data Science at University of Lyon 1. My main research interest is in developing principled machine learning and probabilistic graphical methods in a variety of application areas including bioinformatics (gene selection and gene network discovery), information retrieval, medicine (medical diagnosis), fraud and anomaly detection, time series forecasting, process monitoring, sentiment analysis, opinion mining, neural conversational models etc.

My current research activities are oriented towards the development of scalable and robusts algorithms for:

- Graphical model structure learning with high dimensional data sets;
- Supervised, semi-supervised and unsupervised ensemble learning,
- Multi-label, multi-target and multi-class learning,
- Learning with missing data and selection bias (caused by preferential exclusion of samples from the data),
- Investigating the circumstances under which non-experimental data can reveal causal effects. Recent projects include puzzling patterns of old-age diseases (e.g. osteoporosis, Elzheimer).
- Deep auto-encoders and generative adversarial networks.
- Since 2018: Deep/NLP techniques for multi document text summarization (RAG), information extraction from scanned documents and autonomous Web agents

---

Application domains

  Bioinformatics, Fault Detection, Forecasting, Process Monitoring, Medical Diagnosis, Natural Language Processing.

---

Current and former Ph.D. students

Imene Kerboua (Autonomous Web agents)
Jonathan Pattin-Cottet (Deep Learning for information extraction in medical presriptions),
Florian Baud (Conversational agents and text summarization),
Miguel Palencia-Olivar (Nonparametric neural topic modeling),
Clement Sage (Deep learning: Extraction of Structured Information from Business Documents),
Anil Narassiguin (Ensemble Selection with Probabilistic Classifier Chains),
Maxime Gasse (Multilabel learning with graphical models),
Tinh Tran (Learning with missing data and selection bias),
Ouadie Gharroudi (Multilabel learning),
Sergio Rodrigues de Morais (Bayesian network structure discovery),
Antoine Mahul (Optimal routing with end-to-end quality of service constraints),
Zahra Hamou Mamar (Monitoring of a tramway guiding system)