Machine Learning and AI with Simple Python and Matlab Scripts

14Genetic Algorithms

14.1 Introduction

Optimization is different from finding an absolute maximum or minimum for a given constraint in that it finds the most suitable solution for the problem under a set of constraints. For non‐linear, discrete or discontinuous space problems, which are the characteristics of many real‐life applications in the twenty‐first century, finding an optimal solution is often a challenging task. Most optimization problems associated with current engineering and managerial tasks tend to involve a large number of requirements typically conflicting with each other. A passenger plane on its route to Hawaii aims to optimize fuel consumption by selecting the shortest path, while avoiding thick clouds on its way to reduce turbulence which may cause extra stress to its wings, but without sharp movements that may make passengers uncomfortable. A manager in charge of manufacturing a brain‐monitoring chip to reduce the effects of Parkinson's disease on elderly patients has to construct a team with doctors, engineers, scientists, chip manufacturers, legal experts, and sales and marketing experts from a pool of candidates with different experience levels and salary requirements, while staying within a given budget. They will be forced to make decisions with adverse effects on their project since they probably cannot afford to choose only the top candidate from each field due to budgetary restrictions or their availability. Every movie producer, football team manager, ...