Machine Learning in Geomechanics 1

Preface

Ioannis STEFANOU¹ and Félix DARVE²

¹GeM, UMR 6183, CNRS, Ecole Centrale Nantes, Nantes Université, France

²3SR Laboratory, CNRS, Grenoble INP – ENSE3, Université Grenoble Alpes, France

When discussing artificial intelligence (AI), some basic questions immediately emerge: What is AI? How does it work?

Behind this well-known term, a collection of methods of applied mathematics allow the computer to learn and identify patterns in data. This collection of methods is called machine learning (ML) and it is the target of these two volumes, which were authored for the 2023 ALERT Geomaterials Doctoral School.

In combination with the tremendous increase in the computational power, ML has led to incredible achievements in many disciplines of science and technology. These achievements were that striking that some researchers believe that ML could become a turning point for humanity, as the discovery of fire was for our far ancestors!

Until the 1960s, the scientific development has been characterized by the so-called “linear physics” and by modeling represented by analytical equations solved explicitly by the available mathematical tools giving rise to analytical solutions. The field of problems that can be solved in this way is, of course, precious, but very limited.

Then the numerical revolution, based on powerful numerical methods and computers, allowed to solve numerically a large variety of problems, which can be described by a known system of equations. Many limits of this ...