Yann DAUXAIS
Computer scientist, Ph.D.
ABOUT ME
My research field is centered on pattern mining and supervised machine learning. I have mainly worked on the usage of standard machine learning (decision trees, SVM, etc.) for mining discriminant temporal patterns. Pattern mining allows to mine patterns that are interpretable for the final user and then directly usable as new knowledge. In pattern mining framework, machine learning allow to represent the numerical dimension of data, for example the temporal dimension, more expressive. My research contributes to propose new temporal patterns, expressive and interpretable.
ACADEMIC CURRICULUM
Ph.D. thesis: Discriminant chronicle mining
Rennes University | Rennes, France | February 2015 - April 2018
IRISA Laboratory, LACODAM team
Supervisors: David Gross-Amblard, Thomas Guyet and André Happe.
Funding: ANSM (French Agency for the Safety of Health Products), PEPS project (Pharmaco-Epidemiology of Healthcare Products) supported by the University Hospital of Rennes.
Ph.D. defended and written in english [Slides (PDF)]
Jury: Sandra Bringay, Florent Masseglia and Panagiotis Papapetrou.
Computer Science Master’s degree
Caen University | Caen, France | 2012 - 2014
DECIM option (decision support and optimization)
Courses: Graphs, flows and deterministic optimization, Introduction to Data Mining, Stochastic optimization, Metaheuristics, Data mining and business areas, Planning and decision support, Agent and reasoning, Constraints and Advanced Data Mining.
Third year computer science engineering student
Polytech’Nantes | Nantes, France | 2011 - 2012
Second year in University
Caen University | Caen, France | 2009 - 2011
INTERNSHIP AND PROJECTS
Master’s degree internship: Design of a global data link constraint for itemsets mining
GREYC laboratory | Caen, France | 2014 - 6 months
Supervisors: Patrice Boizumault, Arnaud Lallouet and Samir Loudni.
Establishment of adapted data structures making it possible to apply filtering rules relating to the constraints specific to data mining on a set of Boolean variables expressed in constraint programming.
Annual project of master second year: Solution enumeration algorithm for Horn and 2-SAT formulas in propositional logic
Caen University | Caen, France | 2013 - 2014
Supervisor: Etienne Grandjean.
Understanding and implementing the linear or quasi-linear algorithms proposed in Johann Brault-Baron’s Ph.D. thesis enumerating the solutions of Horn and Horn-renominable formulas.
Annual project of master first year: Reconstruction of a polyomino
Caen University | Caen, France | 2012 - 2013
Supervisor: Jean-jacques Hébrard.
Understanding and implementing an algorithm for reducing the reconstruction of hv-convex polyominos to a 2-SAT problem and then adapting it to enumerate all the polyominos of given perimeters in order to study their properties.
INVITATIONS AND MENTORING
Pint of Science: Artificial intelligence, real stakes!
Warpzone bar | Rennes, France | May 4, 2018
Popularization of data mining and associated application. Slides in french. [Slides (PDF)]
Marine Antigny: Healthcare pathway analysis through topic models
IRISA | Rennes, France | 2018 - 3 months
First year master’s degree (ENSAI) internship mentoring.
LAAS
Toulouse, France | March 30, 2018
Yannick Pencolé’s invitation (DISCO team) to present my Ph.D. thesis work during vacances dedicated to the chronicle model.
CMAP
Ecole Polytechnique | Paris-Saclay, France | November 12, 2017
Stéphane Gaiffas’ invitation to present my Ph.D. thesis work.
François Mentec: Chronicle selection for temporal sequence classification
IRISA | Rennes, France | 2017 - 3 months
First year master’s degree internship mentoring.
TEACHING
Imperative Programming 2
Rennes University | 2015 - 2017
Teaching of imperative programming notions through the JAVA language.
Students at second year of computer science bachelor.
Lead professor: Alexandre Termier.
Lab project drafting. 60 hours as lab assistant (3 student groups) and 4 hours as tutorials assistant.
Database and Decision-Making 1
Rennes University | 2015 - 2017
Lessons of Business Intelligence notions through Business Objects tools.
Students at first year of MIAGE master.
Lead professor: David Gross-Amblard.
30 hours as lab assistant (5 student groups).
Database and Decision-making 2
Rennes University | 2016 - 2017
Teaching of MapReduce notions and practice through the uses of Hadoop and PigLatin.
Students at first year of MIAGE master.
Lead professor: David Gross-Amblard.
4 hours as lab assistant and 2 hours as tutorials assistant.
Operating systems organization 1
Rennes University | 2015 - 2016
Teachings of Internet protocols concepts.
Students at third year of computer science bachelor.
Lead professor: Adlen Ksentini.
8 hours as lab assistant and 4 hours as tutorials assistant.
PUBLICATIONS
- C. Gautrais, Y. Dauxais, and M. Guillemé, “Multi-Plant Photovoltaic Energy Forecasting Challenge: Second Place Solution,” in ECML PKDD Discovery Challenges, 2017. [PDF]
- Y. Dauxais, D. Gross-Amblard, T. Guyet, and A. Happe, “Discriminant chronicles mining. Application to care pathways analytics,” in AIME, 2017. [PDF]
- T. Guyet, A. Happe, and Y. Dauxais, “Declarative sequential pattern mining of care pathways,” in AIME, 2017. [PDF]
- Y. Dauxais, D. Gross-Amblard, T. Guyet, and A. Happe, “Extraction de chroniques discriminantes,” in EGC, 2017. [PDF]
- Y. Dauxais, D. Gross-Amblard, T. Guyet, and A. Happe, “Chronicles mining in a database of drugs exposures,” in ECML Doctoral consortium, 2015. [PDF]
- V. Lemière, Y. Dauxais, P. Boizumault, and A. Lallouet, “Mining frequent patterns using CP: a comparative study,” in CP Doctoral program, 2016.
- F. Balusson et al., “PEPS: a platform for supporting studies in pharmaco-epidemiology using medico-administrative databases,” in International Congress on e-Health Research, 2016.
SOFTWARE
DCM is a frequent discriminant chronicles extraction algorithm implemented in C++.
JumpLeo is an Android arcade game developed using Unity3D in collaboration with Mélusine Vilars.