Personal tools
Laboratoire d'InfoRmatique en Image et Systèmes d'information

Skip to content. | Skip to navigation

Laboratoire d'InfoRmatique en Image et Systèmes d'information
UMR 5205 CNRS / INSA Lyon / Université Claude Bernard Lyon 1 / Université Lumière Lyon 2 / École Centrale de Lyon
You are here: Home > membres

Romain Mathonat


PhD student

Team Data Mining and Machine Learning
Institution Institut National des Sciences Appliquées de Lyon
Location Blaise Pascal (INSA)
E-mail romain.mathonat at
Contact details Publications Thesis
Subject Methods and tools to exploit data streams and support industrial system supervision : from anomaly detection towards predictive maintenance.
Abstract We consider the context of industrial system supervision. We assume that multiple data streams can be collected from its different sub-systems and that these sub-systems are more or less coupled. Our main goal is to support a more intelligent supervision that can rely on the joint analysis of the collected data from the different sub-systems. Indeed, we know that we have an incomplete knowledge of the relationships between the different variables of the different sub-systems and we suspect that discovering relevant relations between such variables can be extremely valuable. An interesting use case concerns anomaly and weak signal detection. A better detection of such information can help to support maintenance and, hopefully, to design new predictive maintenance services.
Knowledge discovery from multiple data streams is a hot topic and numerous open questions have to be considered. Beyond the use and/or the design of relevant pattern domains to support such complex spatio-temporal data analysis, we want to contribute to the efficient use of expert knowledge (e.g., encoded within expert models) to avoid the discovery of irrelevant or trivial patterns. Various real industrial case studies motivate the research and will be used to assess the impact of results.
Advisor Jean-Francois Boulicaut
Advisor Mehdi Kaytoue

Last update : 2018-02-01 11:05:27