Name: Thierry Denoeux
Institution: Université de Technologie de Compiègne, France
Role at institution: Full Professor (Exceptional Class)
Journal role: Editor-in-Chief
- What inspired your career in research?
I think that, as long as I can remember, I have always felt attracted by science. Around the age of eight, I received an educational microscope kit, which may have sparked my passion for scientific research! I studied at the Ecole des Ponts ParisTech after which I prepared a PhD in environmental engineering, working on short-term rainfall forecasting by tracking rain cells in weather radar images. During that time, I became interested in statistical pattern recognition (now part of machine learning) and the problem of quantifying prediction uncertainty, two research topics I continue to explore now. After three years in the water industry, developing expert systems and neural networks, I finally joined the computer science department at Université de technologie de Compiègne. I was also lucky to encounter prominent scholars and outstanding individuals through my research career. In particular, I am thinking of the late Philippe Smets who passed on to me his enthusiasm for the theory of belief functions, and also Didier Dubois and Henri Prade who broadened my view on the various theories of uncertainty and their interconnections.
- How would you describe a typical working day?
I start early in the morning and respond to urgent emails, especially those about the two journals under my responsibility as Editor-in Chief. I then log in to EVISE and EES and handle submissions. I consider it my duty as editor to handle these tasks with the highest priority, as authors rightly expect their submissions to be handled expeditiously. I then spend the rest of the day reading and writing papers, preparing lectures, teaching, advising students, and also doing some administrative work, which is not the best part of the job!
- How do you measure success in your work?
Success to me is finding what I believe to be an interesting solution to some research problem, or providing a new perspective on a research question. For instance, I was recently excited to discover that neural networks can be seen as mechanisms for combining evidence expressed as Dempster-Shafer belief functions, which establishes a connection between two fields that have developed independently until now. From a general point of view, I think that we must be very cautious about measuring research “performance” using quantitative indicators such as the numbers of publications or citations, as researchers and decision-makers may focus too much on these indicators and lose sight of what research really is about, i.e., advancing the state of knowledge and addressing problems faced by society.
- Do you have any particular advice for younger researchers?
Being a young researcher is undoubtedly more difficult nowadays than it was twenty or thirty years ago, with harsher competition and increasing pressure to publish and get funding. My advice would perhaps be not to focus exclusively on fashionable and “hot” topics, but to venture into uncharted territories and explore less trodden paths, especially at boundaries between different scientific disciplines.
- What drove you to become an editor?
When Piero Bonissone, the former Editor-in-Chief of the International Journal of Approximate Reasoning told me that he intended to resign and asked if he could recommend me as his successor, I accepted enthusiastically, and I never regretted that decision. I believe that peer-reviewed scientific journals have played and will continue to play a unique role in the progress of science and the dissemination of knowledge. It is also very interesting to see how the world of scientific publishing is evolving with, for instance, the development of open access. This is why I recently agreed to meet a new challenge and become the Editor-in-Chief of Array, a new broad scope open access journal covering the whole field of computer science.
- What is the most rewarding aspect of editorial work for you and what do you find difficult about the role?
The most rewarding aspect of editorial work is helping authors to improve their papers. There is often a drastic improvement between the first and final version of a manuscript, and this reflects the added value of traditional peer-reviewed journals as compared to, e.g., public repositories or journals ensuring fast publication with only a shallow review. The most difficult part is finding the right balance between the legitimate aspiration of authors for fast publication, and the necessity of maintaining a high quality of peer-review, bearing in mind that reviewers accept this time-consuming task as a voluntary service to the research community.
- What is the most important attribute for being an editor?
Being hard-working maybe, as the flow of papers never stops, even on weekends and during holidays! Beyond that, it is important to have a broad view of a research field and its evolution, and also to be rigorous to ensure fairness in the editorial and review process.
- Name one item/tool/resource that you cannot do without in your editorial role?
My laptop of course, a 15” MacBook Pro I carry all over the world!
- What would you be doing now if you were not X?
It is hard for me to imagine myself not being a professor or a researcher, but I could have embraced completely different fields. In high school I was also interested in literature and philosophy, and I could have followed one of those paths. Not sure I would have been successful though!
- What is the most interesting image/photograph you have come across in your journal?
We do not publish many images or photographs actually, as most of our published articles are rather theoretical, but I chose the below interesting picture from a recent paper (Marie Lachaize, Sylvie Le Hégarat-Mascle, Emanuel Aldea, Aude Maitrot, Roger Reynaud, Evidential split-and-merge: Application to object-based image analysis, International Journal of Approximate Reasoning, 103: 303-319, 2018). This picture shows examples of results from a complex algorithm for segmenting objects on a waste conveyor belt and simultaneously recognizing the material they are made of, using information from several sensors. The method relies on the theory of belief functions for information fusion, taking into account uncertainties encountered at all stages of the process. I think this is a good example of applied research making use of state-of-the-art methodologies for solving a practical problem of great importance in today’s society.