Johann Dreo1

Expert Research Engineer in Artificial Intelligence Algorithmics for Decision Support Systems, at Institut Pasteur.

Research

Scientific interests: optimization, search heuristics, explainable artificial intelligence, machine learning, algorithm design and engineering, graph mining, semantic knowledge graphs, reproducibility, systems biomedicine.

I'm currently2 working on:

  1. analysis of omics data against cancer,
  2. integration of explainable AI with decision support systems,
  3. software tooling around semantic knowledge graphs to help with the two previous problems,
  4. automated design/configuration (of bioinformatics tools) with search heuristics.

Wants to read more about my team? See my professional page at Institut Pasteur.

A robotic hand holding a red snake (DNA), which is facing another darker red snake (RNA), both of which are interlaced in a double helix recalling the DNA one. The snakes get out of a geometric figure figuring a virus capside. The title reads 'bioinformatics'.

Science

Prominent achievements:

  • Science: seminal papers and breakthrough in (automated) algorithm design.
  • Innovation: award-winning open-source stochastic optimization solvers.
  • Transfer: new planning algorithms used in actual products.

I also like:

  • software development (I master C++, Python, Bash, Linux, and modular architectures),
  • popular science communication (especially diagrams & illustrations),
  • free-culture movement and open-science,
  • art/science projects (do not hesitate to contact me if you're an artist looking for a collaboration),
  • philosophy and empirical science epistemology.

Softwares

Scientific Softwares

  • OntoWeaver: a tool to easily extract tabular data into semantic knowledge graphs (using BioCypher). See the publications: about OntoWeaver and about Biocypher.
  • Oncodash: a decision support system for tumor boards.
  • Paradiseo: an open-source full-featured evolutionary computation framework which main purpose is to help you write your own stochastic optimization algorithms. See the summary publication: 22 Years of Paradiseo.
  • Frictionlesser: a tool to search for cancer signatures in transcriptomics data.
  • Descarwin: The DaE and YAHSP temporal planning solvers, which won IPC 2011. Implemented with Paradiseo.
  • Elpida: The simplest possible message protocol able to connect black-box optimization problems and solvers.

See more on my and .

Free Softwares

Some Free softwares I'm proud of:

  • Liquidprompt: a full-featured & carefully designed adaptive prompt for Bash & Zsh. And its Jolly theme (see my presentation at FOSDEM 2024).
  • Colout: color text streams with a polished command line interface.
  • Clutchlog: C++ logging system which targets versatile debugging instead of service event storage.
  • Algopattern: an educational demo of design patterns which are useful for algorithmics.
  • Übergeekism: an attempt at using as many as possible cool computer science stuff to produce a single image.

See more on my personal Github.

Thumbnail of a poster showing mathematical formulas and diagrams

Publications

Ten noteworthy scientific articles

As seen by me2.

Fundamental Research

Innovation/Methodology Research

  • (). One of the very first Pareto-optimal approach to automated AI planning. This work extends the DAE solver and proposes a simple multi-objective benchmark with proven optimal solutions. It also outperforms the only known metric-sensitive solver, competing on an —albeit simpler— aggregated objective problems.
  • (). The method that won the 2017 Black Box Optimization Competition in the single objective track. The seminal work on learning parameters-features mapping that can be embedded within solvers. Classical CMA-ES versions were leading the competition since years and this work enabled a performances breakthrough.
  • (). This solver won the 2011 AI planning competition in the temporal track. It was the first time a stochastic metaheuristic won the IPC and it was on its hardest problems. It's currently used in command and control prototypes. This work with ONERA and INRIA also led to the solver that won the following IPC.
  • (). Optimal threatening trajectories computation using a recent fast marching algorithm (developed in a framework of differential geometry), which takes into account curvature constraints. A surveillance system optimization leverages a reverse-mode semi-automatic differentiation, estimating the gradient of a value function related to a sensor location.

Applied Research

Find out more

See also all my publications (in a single table, with links to PDFs), and the corresponding BiBTeX database (open it with JabRef to enjoy additional metadata information).

Cover of the book 'Metaheuristics for Hard Optimization', red background with a blue section, title written in white.

About me

I am a research engineer with a widespread culture and experience in several corners of computer science and applied research, with a background in biology and free-culture movement.

I hold master's degrees in "Ecology, Ethology & Evolution", in "Biomathematics", and a Ph.D. in "Biomedical Engineering & Algorithmics".

I started my career working in academics, conducting research on biomedical applications of operation research and teaching computer science & network security. I then worked for 13 years in an industrial research center, conducting research on various subjects around AI; generally applied to automated decision & design; mainly on planning of advanced robotic systems and automated design of optimization solvers. I am now working as a research engineer at Institut Pasteur.

I have teached several courses in computer science and AI, most prominently at Institut Polytechnique de Paris and Paris-Est University.

I have been involved in various open-source software projects (some of them very successful, like Liquidprompt), and free-culture movements; most notably in Wikipedia (since 2003) and Wikimédia France (since… the beginning, and for which I have been an early board member).

Curriculum

Background

  1. Institut Pasteur (Paris) Computational Systems Biomedecine laboratory / Bioinformatics & Biostatistics hub

    Expert Research Engineer

    Explainable Artificial Intelligence Algorithmics for Decision Support Systems against Cancer.

  2. Institut Pasteur (Paris) Computational Biology department > Systems Biology group

    Researcher

    Statistical analysis of transcriptomics data with randomized local search algorithms.

  3. Thales Research (Palaiseau) Decision & Optimization laboratory

    Senior Research Engineer

    Algorithmics: Mission Planning, Motion Planning, Path Planning, static Hamilton-Jacobi equations, Robotics AI, Design of Experiments, Machine Learning for Stochastic Optimization, Statistics.

  4. Thales Research (Palaiseau) Mathematics & Technology of Decision laboratory

    Research Engineer

    Computer science: planning, metaheuristics (stochastic global optimization algorithms), operations research, artificial intelligence.

  5. École Nationale Supérieure des Mines (Saint-Étienne) Health Science department

    Post-doctoral Researcher

    Metaheuristics and column generation hybridation for operating rooms stochastic planning. Swarm intelligence for multi-robots systems in hospitals.

  6. Université Paris Est (Créteil) Research in Instruments, Signals & Systems laboratory / University Technical Institute > Network & Security

    Researcher & Teacher

    Metaheuristics: tests, validation and programming, hard optimization, data analysis. Lectures on network, security and administration.

Education

  1. Université Paris Est (Créteil) Research in Instruments, Signals & Systems laboratory

    Ph.D. “Biomedical Engineering & Bio-inspired Algorithmics”

    Adaptation of the ant colony optimization metaheuristic to continuous problems, applications in biomedical engineering. Metaheuristics, ant colony optimization, estimation of distribution algorithms, imaging.

  2. Sorbonne University (Paris) & Free University of Brussels (Brussels)

    Postgraduate certificate “Biomathematics”

    Establishment of a model of an ant colony behaviour. Multi-agent modeling, complex systems, self-organization.

  3. University of Rennes 1 (Rennes)

    Master's degree “Ecology, Ethology & Evolution”

    Establishment of a differential equations model of a social monkey behaviour.

A Penrose aperiodic tiling, with its weighted graph superimposed.

Skills

Since I left my position as a researcher, my job is to help people conduct their own research. I can most notably help you on the following topics.

Decision-Support Systems

A diagram showing colored boxes with arrows.
Design and implementation of decision-aid software
Workflow modelling, automated decision integration, data representation and integration, high-level UX design..
Visualization and diagrams
How to summarize complex data/concepts in a visual way.
Integration of third-party modules
How to design API to use external services, how to integrate software that does not really want to be integrated.

Automated Decision

A black control board showing round buttons, a progress bar and a formula 'y=argmin g(y)'. The same panel is embedded inside the panel.
Score function modelling
How to design a metric defining a quality for a solution to a decision problem, while maintaining good mathematical properties.
Optimization problem modelling
How to design a formal model of a decision problem to be automatically solved by a computer.
Solving automated configuration problems
How to set parameters of a complex system so as to maximize its performances.

Scientific Computing

A set of lego bricks being plugged in together, inside a circular cycle.
Efficient algorithmics
How to cope with combinatorial explosion or curse of dimension when implementing complex algorithms.
Highly modular software architectures
How to structure your code to allow efficient —and automated— exploration of your ideas.
Parallel Programming
Multi-core and multi-process programming, Message Passing Interface.
Modern C++ / Python bindings
How to program with C++ using —almost— the same concepts than in Python, or within Python itself.
Shell scripting
How to use the existing Unix tools to —very— efficiently automatize any task.

Artifical Intelligence

A diagram showing columns of gray and blue squares, with red boxes and arrozs inbetween them.
Semantic graph mining
How to build, integrate and exploit semantic knowledge graphs.
Search heuristics, metaheuristics or evolutionary computation
How to solve hard optimization problems.
Automated planning
How to compute shortest paths, and more generally optimize sequences of actions.

Specific Tools

Programming languages
C++, Python, Bash, Cypher/GQL, Bash, PDDL.
Databases
Neo4j, PostgreSQL.
Frameworks
ParadisEO, IOH experimenter, CGAL
Development tools
Git, CMake, Liquidprompt, Vim, Kakoune.
Vector drawing
Inkscape, Dia.
Text processing
LaTeX, LibreOffice, SPIP.
Sysadmin
GNU/Linux, i3, many shell tools.
HPC
Slurm, SnakeMake, Apptainer.
Popular science
Wikipedia.

Awards

  1. Best poster prize, awarded in the "AI" track, for the poster "Finding personalized drugs from the integration of clinical and molecular data into a semantic knowledge graph".

    European Society for Medical Oncology's conference on Molecular Analysis for Precision Oncology.

  2. Best poster prize, awarded (independently) by the two patients representatives, for the poster "Make sense of heterogeneous data and agree on a treatment, using a Web application tailored for Tumor Boards".

    DECIDER project general meeting

  3. Supervisor of masters students having won the regional "André Blanc-Lapierre" prize

    French society of electricity, electronics and information & communication technologies

  4. Among the winners of the Open Optimization Competition, for code on "Extensible Logging and Empirical Attainment Function for IOHexperimenter".

    Facebook AI Research, Sorbonne Université, Leiden University

  5. Among the winners of the Open Optimization Competition, for code on "on-the-fly computation of the area under the EDCF curve, for use within automated algorithm configuration frameworks".

    Facebook AI Research, Sorbonne Université, Leiden University

  6. Best poster award, for the poster "Netted multi-function radars positioning and modes selection by non-holonomic fast marching computation of highest threatening trajectories & by CMA-ES optimization".

    RADAR conference

  7. "Best Expertise" prize, for a software library and a chatbot for polls with multiple voting systems.

    Thales Inner-source Hackathon

  8. Winner team of the Black-Box Optimization Competition

    Association for Computing Machinery, our student also won in 2019

  9. Winner team of the International Planning Competition (temporal satisficing track)

    International Conference on Automated Planning and Scheduling

  10. Winner of the French Competition on Abstract Combinatorial Board Games

    Dourdan Board Game Festival

Ph.D. Genealogy

Below is the history of my Ph.D. "descendants" and "ancestors". I.E. the Ph.D. students I supervised (not always as an official director), and the list of the directors of my directors (while historically fun, probably not very useful).

Ph.D. students

  1. Quentin Renau

    Algorithmics

  2. Nacim Belkhir

    Algorithmics

  3. Thomas Sousselier

    Robotics

Ph.D. Ancestors

  1. Johann Dreo (France)

    Algorithmics

  2. Patrick Siarry (France)

    Algorithmics

  3. Gérard Dreyfus (France)

    Machine Learning

  4. Jacques Lewiner (France)

    Physics

  5. Paul Herman Ernst Meijer (Netherlands)

    Physics

  6. Hendrik Anthony Kramers (Netherlands)

    Quantum Physics

  7. Paul Ehrenfest (Austria)

    Particle Physics

  8. Ludwig Boltzmann (Austria)

    Mechanical Statistics

  9. Jožef Štefan (Austria)

    Thermodynamics

  10. Andreas von Ettingshausen (Austria)

    Mathematics & Physics

  11. Ignaz Lindner (Austria)

    Mathematics

  12. Georg Jurij Bartolomej Veha von Vega (Slovenia)

    Mathematics

  13. Gabriel Gruber (Austria)

    Mathematics

  14. Joseph Giuseppe Jakob von Maffei (Austria)

    Unknown (to me)

  15. Nikolaus Boda Poda von Neuhaus (Austria)

    Mathematics, Entomology, Physics

Notes

  • 1: My family name may sometime be written with an accent: Johann Dréo. It is in any case actually pronounced with an accent (sounds like "jo-ann dré-o", IPA: [ʒoan dʁeo]).
  • 2: Updated on .