Data and AI

Artificial intelligence makes it possible to extract meaning from the big data sets that technological advances have made it possible to collect. The challenge for researchers is to find innovative ways of using computers to aggregate, analyze, and cross-reference data to reach new and reliable conclusions. They are using computational neurobiology, mathematical logic, and computer science to solve logically or algorithmically complex problems. Big data analysis, processing, and mining have applications in many fields, including diagnostic assistance in medicine, decision support and task automation in industry and transportation, and machine translation in communications.

The growing abundance of statistical data and the ever larger databases in fields such as biology, ecology, and genetics are prompting researchers to develop new analysis tools using statistical sampling, capture-recapture models, random effects models, and non-standard data defined using complex mathematical sets.

Infrared is a form of electromagnetic radiation whose waves have a lower frequency than visible light (red); it is directly linked to heat because heat waves are emitted in the infrared spectrum. Multispectral remote sensing systems capture infrared (or thermal) radiation along with reflected infrared. New infrared devices and increased computing power have opened up new possibilities for using infrared technology, including detecting weak points in building insulation, locating hard-to-access victims during rescue operations, studying nocturnal animal species, and diagnosing pathologies.

Machine learning is a subfield of AI that focuses on designing, analyzing, developing, and implementing methods that teach computers to acquire new knowledge and skills by processing data in order to perform complex tasks. The ability of machines to make inferences from large amounts of data using algorithms is used in many areas, including voice, facial, and object recognition and machine translation, among others.

Multiphysical mathematical modelling makes it possible to create mathematical models that take into account all the complex physical phenomena (e.g., elasticity, plasticity, thermal and fluid dynamics, and electromagnetism) that occur during product manufacturing and use. These models are used to solve applied problems with a view to designing better quality products. Researchers use partial differential equations (Fourier, Navier-Stokes, Maxwell, etc.) to study the theoretical resolution of complex situations and develop tools to solve them.

For insurance companies, risk theory involves assessing the overall risk of risk portfolios covered by insurance contracts. The distribution of total claims for a portfolio is used to measure the overall risk and determine the premium, which offsets the compensation paid out to insured persons in the event of a claim. Frequency and severity are the two parameters used to measure risks.

Signal processing involves techniques for processing, analyzing, and interpreting signals. It includes control, filtering, data transmission, noise reduction, and identification. Signal processing draws on electronics and automation, as well as a number of other fields, including mathematics (linear algebra and stochastic processes), information theory, and numerical analysis. Analog signals, which are produced by sensors and amplifiers, are distinguished from digital signals, which are produced by computers and terminals.