Machine Learning in Geomechanics 1

4Classification Techniques in Machine Learning

Noel JAKSE

SIMaP, CNRS, Grenoble INP, Université Grenoble Alpes, France

4.1. Introduction

At the turn of the 21st century, machine learning (ML) techniques aroused a growing interest across various scientific disciplines due to their inherent capability to identify trends, forecast outcomes and draw invaluable knowledge from vast and intricate datasets (Hey and Trefethen 2020; Hey et al. 2009). Tools designed for collecting data online, scientific apparatus and computer-simulated environments are generating data at an exponential rate. This has spurred new scientific methodologies for dissecting and structuring enormous volumes of data, while simultaneously detecting subtle effects that were previously hidden. These techniques have become increasingly prevalent in areas such as materials science (Schmidt et al. 2019) and geophysics (Karapiperis and Andrade 2021).

ML approaches can be primarily grouped into three categories: supervised learning, unsupervised learning and semi-supervised learning. Each category presents unique ways of deriving knowledge from data, each possessing their own benefits and limitations. Among these, supervised learning is widely adopted, where a model is trained using labeled data, consisting of both input attributes and associated output labels. The model is trained to associate inputs with their corresponding outputs, enabling it to make predictions for novel, unobserved data.

This chapter is devoted ...