Abstract
From childhood through to research, the back-and-forth between concrete problems and abstraction enables the discovery and understanding of new mathematical concepts. In practice, it's difficult to extend the manipulative approaches deployed in primary school to high school, because the mathematics on the syllabus is more complex and there are severe time constraints.
To get students excited about mathematics again, and improve their understanding of it, the MathAData program offers secondary school teachers concrete, fun problems derived from AI challenges, on subjects as diverse as image recognition, medical diagnosis, whale song analysis, text author recognition... These challenges are quickly translated into mathematical problems. Co-developed with teachers, the teaching materials lead students to understand and manipulate the mathematical concepts of each chapter of the program. This begins with digital experiments in class, where students develop creative solutions to the challenge, while at the same time bringing out the mathematics, and then deepening their knowledge with exercises. They also discover the mathematical principles of artificial intelligence, which becomes important for their education.
The conference will introduce the pedagogical principles of this lecture in relation to the mathematics of machine learning, which lies at the heart of artificial intelligence. There will be a first look back at the tests underway, with around a hundred teachers trained in the Lille, Créteil and Paris academies, discussing the challenges of scaling up.
