I am a 3-rd year PhD student working at Université de Rennes and IRISA lab in the DiverSE working group in Rennes (France).
I am working on testing Computer vision algorithms with the use of Software Engineering techniques. (more in the PhD topic section)
Furthermore, I am a tennis player. I began to play tennis at 6 until my 18. I stopped to focus on my studies and started again during my PhD.
After graduating in Computer Science with a fous on Data Image Processing (a 2-years diploma teachning fundamentals of a panel of facets of computer science, including OS, Programming, Mathematics, Web technologies, network managing, Systems design, etc.)from the Institute of Technology of Lannion (France); I went to ESIR (Ecole Supérieure d’Ingénieurs de Rennes, Rennes, France), an engeeniering school where I chose to continue focusing on Data Image Processing. During the last year of ESIR, I also graduated from the University of Rennes 1 with a Master in Computer Science with a focus on introducing the world of research to student. Thus, I have a strong background in Data Image Processing (as well as Machine Learning) on one hand and in Softwarer Engineering thanks to my PhD on the other hand.
In the context of this PhD, we try to better understand how variability affects configurable systems in their performances. In particular, we focus on Computer Vision (CV) systems which are based on mathematical fundamentals. They are also configurable systems with options that can influence their performances. There are plenty of CV techniques developped to address different yet similar issues (such as tracking objects in videos or tracking pedestrians). Choosing a CV configuration that will give the best performances w.r.t. an input (i.e., videos or inputs) is a non-trivial process. Inputs can have features that affect drastically the performances of the CV technique. On the other hand, the configuration of the CV technique can also affect its performances.
Part of my work is to leverage Machine Learning techniques in order to understand how the CV configuration, on one hand, and given inputs, on the other, affect the performance of the program.
My internship took place in the TEXMEX (now LinkMedia )working group under the supervision of Ewa Kijak and Laurent Amsaleg. The topic of this internship was the security of Machine Learning processes applied to multimedia content which followed the work of Thanh-Toan Do and his PhD. The goal of this internship was to have a better understanding of how Machine Learning techniques can be influenced in the establishment of their separating functions depending on data that they have seen. An other concern was how sensitive the establishment of separating functions is w.r.t. the distribution of data points. This topic is very close to the work conducted by Battista Biggio and the PRALab in Cagliari.
Paul Temple, José Angel Galindo Duarte, Mathieu Acher, Jean-Marc Jézéquel. Using Machine Learning to Infer Constraints for Product Lines. Software Product Line Conference (SPLC), Sep 2016, Beijing, China
Paul Temple, Mathieu Acher, Jean-Marc Jézéquel, Léo Noel-Baron, José Galindo. Learning-Based Performance Specialization of Configurable Systems
Paul Temple, Mathieu Acher, Jean-Marc Jézéquel, Olivier Barias. Learning Contextual-Variability Models. IEEE Software Special Issue on Context Aware and Smart Healthcare, Novembre-Décembre 2017
Mathieu Acher, Paul Temple, Jean-Marc Jézéquel, José Angel Galindo Duarte, Jabier Martinez, Tewfik Ziadi. VaryLaTeX: Learning Paper Variants That Meet Constraints. 12th International Workshop on Variability Modelling of Software-Intensive Systems (VaMoS), Feb 2018, Madrid, Spain
during the universitary year 2017-2108:
during the universitary year 2016-2017:
during the universitary year 2015-2016:
during the universitary year 2015-2014 (at ENSICaen):