Visualization as you've never seen it before

First historical steps in visualization

To put in image the invisible, it is possible thanks to visualization! This field of research facilitates the understanding of data by representing them in visual form. From monitoring the evolution of the Covid-19 to real-time projections of the American elections, visualization has democratized its tools in record time, to the point of fully integrating our daily life.

Indeed, this discipline has a short but intense history. It was born in 1990, at the time of the creation of the annual IEEE* Visualization conference (Institute of Electrical and Electronics Engineers). Originally, it was dedicated solely to the 3D representation of real objects and their environment, such as an airplane and the flow of fluids around it. But it has since been greatly enriched and is now at the interface between computer graphics, statistics, cartography and human-computer interaction.

Pioneering research

Until 2002, France was invisible on an international level, having at that time only a few teams specialized in 3D graphics. Everything was still to be done when Jean-Daniel Fekete, recruited at Inria Futurs, joined the InSitu project-team focused on human-computer interactions. "The research I conducted at the time was original, not at all in the spirit of the times. Inria allowed me to follow my scientific intuitions and impressions, which are paying off in a central discipline, and even standard today," says the researcher.

In the 2000s, he created a toolkit (Infovis Toolkit) for the design and development of visualization tools. He also expanded the field of visual representation of networks, previously limited to "nodes and links", by highlighting new ways of displaying them (adjacency matrices, hybrids, etc.). This aspect of his research now applies to all types of networks, both to the study of social networks and neural networks.

Rise of the Aviz team and international recognition

In 2006, Jean-Daniel Fekete created the Aviz (Analysis and VisualiZation) project-team, a joint venture between the Inria Saclay - Île-de-France research center and the LRI (INS2I's computer science research laboratory). She collaborates with Microsoft and Google, and publishes extensively in prestigious conferences in the field. Its former students settle around the world, sometimes in the service of research, sometimes in the service of industry. As a result, the team is growing rapidly. "In 2014, we organized the annual IEEE VIS conference in Paris with a record attendance. It was the first time it was held outside the United States in almost 25 years, and it marked the crowning achievement of our international visibility," recalls Jean-Daniel Fekete.

At the time, his team was one of the most cited in the world in connection with the conference, placing Inria and France at the forefront of the scene. This recognition by the community reached its peak in 2020 for Jean-Daniel Fekete, when he received the honorary title of Technical Achievement Award from the IEEE Visualization and Graphics Technical Committee (VGTC) in recognition of his career.  He was then inducted into the Visualization Academy of the IEEE VGCT.

De nouvelles visions pour l’aide à la décision

« Désormais, nous devons rendre les systèmes d’analyse de données plus faciles à manier pour des raisons scientifiques, démocratiques et sociales, souligne le chercheur. La visualisation sert à l’appréhension simplifiée d’importantes quantités de données en vue de mieux comprendre le monde et prendre de meilleures décisions. » En ce sens, la recherche se concentre notamment sur les utilisateurs. Qui va utiliser et interpréter ces données ? À qui se destinent-elles ? Comprendre cet acteur est essentiel à la visualisation interactive, car il sera amené à agir sur le contenu afin d’en extraire des enseignements.

Charles Perin, Pierre Dragicevic and Jean-Daniel Fekete

Par ailleurs, l’un des enjeux de la visualisation n’est plus d’observer des données telles qu’elles sont, mais d’y appliquer de nouvelles analyses. Une démarche au cœur de grandes ambitions : au-delà de ce qui se passe maintenant, il faut pouvoir prédire le déroulement à venir. Pourquoi ? Principalement afin d’aider la prise de décision d’ingénieurs ou encore de décideurs politiques. Comment ? Grâce à des projections visuelles du comportement d’un système industriel, du climat futur, de l’impact d’une pandémie, etc.

Enfin, la visualisation n’a jamais été aussi transverse. Qu’elle s’associe à l’astronomie, à la médecine ou encore à l’histoire, ses outils contribuent à la compréhension de nombreuses disciplines. Et Jean-Daniel Fekete de préciser : « Un même outil de visualisation des réseaux peut servir à représenter des migrations pour un historien, ainsi que des connexions cérébrales aux yeux d’un neurobiologiste. C’est alors l’application spécialisée qui donne le sens aux images que l’on obtient.