Jeremy Boes' Personal Page

Jeremy Boes
AI Researcher @Brennus Analytics
(Multi-Agent Systems, Self-Organization, Machine Learning)

Toulouse, France

Timeline

Master's Degree and Internship
University of Toulouse and IRIT - Team VORTEX (Visual Objects: from Reality To EXpression)
The major of the Master's Degree was 3D modeling, rendering, and animation, but I also followed courses on image processing and analysis, human-computer interaction, and game design.
I did my internship at the Toulouse Computer Science Research Institute (IRIT) where I worked on a 3D agent-based pedestrian simulation.

Thesis
IRIT - Team SMAC (Cooperative Multi-Agent Systems)
During my thesis, I studied the question of complex systems control with multi-agent systems. I designed and implemented a cooperative self-organizing MAS able to learn in real-time how to control a given complex system. It was successfully tested on real combustion engines. Four international conferences papers and one workshop paper were puplished about this work. Alongside, I accumulated a total of 284 hours of teaching, and have been for a year a member of the committee for the organization of a monthly doctoral seminar.
Advisors: Pierre Glize and Frédéric Migeon

Post-doc
IRIT - Team SMAC (Cooperative Multi-Agent Systems)
I continue my work on cooperative self-organizing MASs. This time, I tackle the problem of co-simulation initialization for large electrical networks.

Post-doc
IRIT - Team SMAC (Cooperative Multi-Agent Systems)
My work on MASs brought me to the neoCampus project, where I explore the mutual benefits of MASs and Ambient Systems.

Brennus Analytics
I am currently working at Brennus Analytics as an AI researcher.

Master's Degree in Image and Multimedia Applications
Major: 3D modeling, rendering, and animation
University of Toulouse
Rank: 1st out of 20

Doctorate in Artificial Intelligence
Learning to Control Complex Systems thanks to Self-Organizing Multi-Agent Systems
Toulouse Computer Science Research Institute (IRIT) - University of Toulouse

Dissertation (in French)

Journal article about my Master's Degree work on agent-based crowd simulation.

Intuitive Method for Pedestrians in Virtual Environments,
J. Boes, C. Sanza and S. Sanchez, Studies in Computational Intelligence, Intelligent Computer Graphics, Vol. 374, p. 117-138, Springer, 2011.

ICINCO: Conference paper on my thesis main work, a MAS to learn how to control complex systems.
SIMULTECH: Conference paper on another work from my thesis, a method to generate black-box test cases.

Self-Organizing Agents for an Adaptive Control of Heat Engines,
J. Boes, F. Migeon and F. Gatto, in International Conference on Informatics in Control, Automation and Robotics (ICINCO 2013), SciTePress, p. 243-250, Reykjavik, Iceland, 2013.

Mimicking Complexity: Automatic Generation of Models for the Development of Self-Adaptive Systems,
J. Boes, P. Glize and F. Migeon, in International Conference on Simulation and Modeling Methodologies, Technologies and Applications (SIMULTECH 2013), SciTePress, p. 353-360, Reykjavik, Iceland, 2013.

Conference paper providing a good abstraction of my thesis and other works of the team.

The Self-Adaptive Context Learning Pattern: Overview and Proposal,
J. Boes, J. Nigon, N. Verstaevel, M.-P. Gleizes, F. Migeon, in International and Interdisciplinary Conference on Modeling and Using Context (CONTEXT 2015), Springer, LNAI 9405, p. 350-359, Larnaca, Cyprus, 2015.

Conference paper about my work with EDF R&D.

Principles and Experiments of a Multi-Agent Approach for Large Co-Simulation Networks Initialization,
J. Boes, T. Jorquera, G. Camilleri, in International Conference on Agents and Artificial Intelligence (ICAART 2017), p. 58-66, Porto, Portugal, 2017.
Paper about the work of our research group on MASs and Machine Learning.

Lifelong Machine Learning with Adaptive Multi-Agent Systems,
N. Verstaevel, J. Boes, J. Nigon, D. d'Amico, M.-P. Gleizes, in International Conference on Agents and Artificial Intelligence (ICAART 2017), p. 275-286, Porto, Portugal, 2017.

Journal paper on a self-adaptive MAS for the control of complex systems with results on real combustion engines..

Self-Organizing Multi-Agent Systems for the Control of Complex Systems,
J. Boes, F. Migeon, The Journal of Systems & Software, 134C, p. 12-28, Elsevier, 2017.

Diploma

Selected Publications

Research

My research focuses on self-organization in multi-agent systems as a way to self-adapt to a dynamic environment and learn complex behaviors. Dealing with complexity cannot be done with the usual top-down global design methods. Instead, a local standpoint is needed. Each agent has to be simple enough to be quite easily designed and implemented, and able to face unexpected events by locally changing its behavior so the system as a whole behaves accordingly to the designer's expectations. In other words, one has to try to 'steer' emergence towards the requirements of the system by using appropriate simple local rules. This is the challenge that we, Team SMAC, address with the Adaptive Multi-Agent Systems (AMAS) approach. Here is a brief recap of my work with Multi-Agent Systems.

Optimization, Simulation, and Machine Learning with Multi-Agent Systems
R&D @Brennus Analytics, then @Price f(x)
I'm in charge of the design of our AI engine, a self-adaptive multi-agent system that solves multi-dimensional multi-criteria optimization problems and provides explainable results. It is applied to pricing problems in our product.

Ambient Intelligence
Postdoctoral Research @University of Toulouse, IRIT, Team SMAC
The neOCampus initiative brings together ten laboratories, from physics to social sciences, and of course computer science around a common goal: building a connected, intelligent and sustainable campus. Several classrooms in the campus are being instrumented with various connected sensors and controllable devices. In this initiative, I explore how Multi-Agent Systems can be beneficial to ambient systems, from network optimization to decision making in pervasive environments.

Co-Simulation Initialization thanks to Self-Organization
Postdoctoral Research @University of Toulouse, IRIT, Team SMAC
Initializing a network of heterogenous models is a challenging task. Usual top-down global methods fail to provide coherent input/output values for each model. In the context of large-scale electrical networks simulation, a multi-agent system has been developped to tackle this problem with a local standpoint. Following cooperation principles, agents find a fixed point (i.e. coherent values for each model) for the whole network while having limited perceptions.

Complex Systems Control thanks to a Self-Organizing Multi-Agent System
PhD @University of Toulouse, IRIT, Team SMAC
Self-adaptation, as described in the Adaptive Multi-Agent Systems (AMASs) theory, is used to tackle complex problems for which no algorithmic solution is known. The agents cooperative self-organizing behavior gives the system the capacity to adapt itself to a dynamic environment, and to maintain a state of functional adequacy. In my thesis, this approach is applied to the control of complex systems, in particular of combustion engines. A combustion engine is a dynamic, non-linear and noisy system. Its tuning involves optimization and regulation of dynamic outputs, and constraint solving. My thesis focused on using cooperative self-organization principles to design a multi-agent control system that does not recquire previous knowledge on the engine. My work took place during a national project involving academic and industrial partners. I had the opportunity to test my system on a real fuel engine. As it was successful, a technology transfer towards the industry is ongoing. I also worked on the problem of complex system simulation for the design of self-adaptive systems.

3D Urban Crowd Simulation
Master Internship @University of Toulouse, IRIT, Team VORTEX
During my Master’s Degree I did a 6 months internship with Team VORTEX of IRIT. I focused on enhancing the behavioral realism of an existing 3D urban crowd simulation. I designed a simple agent-based method that naturally combines collision avoidance and movement coherence (staying on sidewalks and crosswalks, etc). This internship has been a bridge between the field of 3D modelling and rendering of my Master’s Degree and the field of Multi-Agent Systems of my PhD.

Science Animation

ICAART Program Comittee Member
I am a member of the ICAART program comittee, starting with the 2018 edition.

Program Comittee Member of the IEEE Workshop on Smart and Sustainable City 2016
I was a member of the program comittee for the First International Workshop on Smart and Sustainable City. It was part of the UIC Conference during the Smart World Congress (August 2017, San Francisco, California, USA).

Session Chair at the International Conference on Agents and Artificial Intelligence (ICAART 2017)
I was a session chair at ICAART 2017 in Porto, Portugal.

Horizon 2020 Project Proposal, Smart Cities and Communities SCC01
I have been taking part in a international consortium of more than 40 partners, writing a proposal submission to the H2020 European call for projects. Toulouse, Seville and Olso are partners in this transdisciplinary project on Smart Cities.

Organization Chair and Session Chair at the IEEE Workshop on Smart and Sustainable City 2016
I was the organization chair and a session chair of the First International Workshop on Smart and Sustainable City. It was part of the UIC Conference during the Smart World Congress (July 18th-21st 2016, Toulouse, France).

Doctoral Seminars Organization Comittee Member
During the year 2012-2013, I have been an active member of the DocToMe organization committee. DocToMe is a monthly doctoral seminar for the PhD students of the laboratory. My main role was to design posters, but I was also involved in recruiting speakers and in holding events.

Teaching

During my three years as a teaching PhD Student and one year as a teacher-research assistant, I accumulated a total of 284 hours of teaching, mainly practical work from Bachelor to Master’s Degree (in several majors: Computer Science, Chemistry, Management). I also supervised interns and group projects, attended to jurys, and took part in all the aspects of teaching (preparing and rating exams, maintaining online courses, etc).

Year-by-year Detail

2015–2016 - Temporary Teacher
Introduction to Algorithmics (Ada), 10h - Computer Science Bachelor's Degree

2013–2014 - Assistant Teacher-Researcher
Tools for Decision Support (Excel), 2x12h - Management Bachelor’s Degree
Computer Science and Internet Basics, 2x36h - Law and Management Bachelor’s Degrees

2012–2013 - Third Year Teaching PhD Student
Object-Oriented Modeling and Programming (UML, JAVA), 2x12h - Software Engineering Master’s Degree
Functional Programming (Caml), 8h - Computer Science Bachelor’s Degree
Logical Programming (Prolog), 6h - Computer Science Bachelor’s Degree
Introduction to Computer Science (Ada), 16h - Chemistry Bachelor’s Degree

2011–2012 - Second Year Teaching PhD Student
Object-Oriented Modeling and Programming (UML, JAVA), 2x12h - Software Engineering Master’s Degree
Functional Programming (Caml), 8h - Computer Science Bachelor’s Degree
Logical Programming (Prolog), 16h - Computer Science Bachelor’s Degree
Introduction to Computer Science (Ada), 16h - Chemistry Bachelor’s Degree

2010–2011 - First Year Teaching PhD Student
Object-Oriented Modeling and Programming (UML, JAVA), 2x12h - Software Engineering Master’s Degree
Large Group Project Supervising, 2x12h - Software Engineering Master’s Degree
Logical Programming (Prolog), 12h - Computer Science Bachelor’s Degree

Other Teaching Activities

2011–2013 - In charge of the online courses platform (Moodle)
Updating and maintaining courses, participation in student threads, etc.

2012 Internship Supervision
Runtime Visualization of Self-Adaptive Multi-Agent System, 3rd year Bachelor student, Duration: 7 weeks

2011-2012 Group Project Supervision
Log-based Combustion Engine Simulator, 4 students, Computer Science Master’s Degree.

Publications

Journals
Self-Organizing Multi-Agent Systems for the Control of Complex Systems,
J. Boes, F. Migeon, The Journal of Systems & Software, 134C, p. 12-28, Elsevier, 2017.
Intuitive Method for Pedestrians in Virtual Environments,
J. Boes, C. Sanza and S. Sanchez, Studies in Computational Intelligence, Intelligent Computer Graphics, Vol. 374, p. 117-138, Springer, 2011.

Conferences
Smart Is a Matter of Context,
J. Nigon, N. Verstaevel, J. Boes, F. Migeon, and M.-P. Gleizes, in the International and Interdisciplinary Conference on Modeling and Using Context (CONTEXT 2017), p. 189-202, Paris, France, 2017.
neOCampus: A Demonstrator of Connected, Innovative, Intelligent and Sustainable Campus,
M.-P. Gleizes, J. Boes, B. Lartigue, F. Thiébolt, in the International Conference on Intelligent Interactive Multimedia Systems and Services (KES-IIMSS 2017), p. 482-491, Vilamoura, Portugal, 2017.
Principles and Experiments of a Multi-Agent Approach for Large Co-Simulation Networks Initialization,
J. Boes, T. Jorquera, G. Camilleri, in International Conference on Agents and Artificial Intelligence (ICAART 2017), p. 58-66, Porto, Portugal, 2017.
Lifelong Machine Learning with Adaptive Multi-Agent Systems,
N. Verstaevel, J. Boes, J. Nigon, D. d'Amico, M.-P. Gleizes, in International Conference on Agents and Artificial Intelligence (ICAART 2017), p. 275-286, Porto, Portugal, 2017.
The Self-Adaptive Context Learning Pattern: Overview and Proposal,
J. Boes, J. Nigon, N. Verstaevel, M.-P. Gleizes, F. Migeon, in the International and Interdisciplinary Conference on Modeling and Using Context (CONTEXT 2015), Springer, LNAI 9405, p. 91-104, Larnaca, Cyprus, 2015.
The Modelica language and the FMI standard for modeling and simulation of Smart Grids ,
O. Chilard, J. Boes, A. Perles, G. Camilleri, M.-P. Gleizes, J.-P. Tavella and D. Croteau, in the International Modelica Conference (MODELICA 2015), Linköping University, p. 189-196, Paris, France, 2015.
Model-free Optimization of an Engine Control Unit thanks to Self-Adaptive Multi-Agent Systems,
J. Boes, F. Migeon, P. Glize and E. Salvy, in Embedded Real-Time Software and Systems (ERTS² 2014), SIA/3AF/SEE, p. 350-359, Toulouse, France, 2014.
Mimicking Complexity: Automatic Generation of Models for the Development of Self-Adaptive Systems,
J. Boes, P. Glize and F. Migeon, in International Conference on Simulation and Modeling Methodologies, Technologies and Applications (SIMULTECH 2013), SciTePress, p. 353-360, Reykjavik, Iceland, 2013.
Self-Organizing Agents for an Adaptive Control of Heat Engines,
J. Boes, F. Migeon and F. Gatto, in International Conference on Informatics in Control, Automation and Robotics (ICINCO 2013), SciTePress, p. 243-250, Reykjavik, Iceland, 2013.
Controlling Complex Systems Dynamics without Prior Model,
J. Boes, F. Gatto, P. Glize and F. Migeon, in International Conference on Agents and Artificial Intelligence (ICAART 2013), SciTePress, p. 389-392, Barcelona, Spain, 2013.
Intuitive Method for Pedestrians Simulation,
J. Boes, S. Sanchez and C. Sanza, in International Conference on Computer Graphics and Artificial Intelligence (3IA 2011), University of Athens, p. 73-81, Athens, Greece, 2011.

Workshops
From Smart Campus to Smart Cities: Issues of the Smart Revolution,
N. Verstaevel, J. Boes, M.-P. Gleizes, in IEEE Workshop on Smart and Sustainable City (WSSC 2017), San Francisco, California, USA, 2017.
Use Cases of Pervasive Artificial Intelligence for Smart Cities Challenges,
J. Nigon, E. Glize, D. Dupas, F. Crasnier and J. Boes, in IEEE Workshop on Smart and Sustainable City (WSSC 2016), Toulouse, France, 2016.
Self-Organizing Agents to Learn the Control of Heat Engines,
J. Boes, F. Migeon and F. Gatto, in 11th European Workshop on Multi-Agent Systems (EUMAS 2013), Toulouse, France, 2013.

Thesis Dissertation
Apprentissage du contrôle de systèmes complexes par l’auto-organisation coopérative d’un SMA,
J. Boes, Thesis Dissertation, University of Toulouse - IRIT, Toulouse, 2014.

Master Dissertation
Méthode intuitive de simulation de piétons en milieu urbain,
J. Boes, Master Dissertation, University of Toulouse - IRIT, Toulouse, 2010.

Mirror: here