Machine learning is revolutionizing our world. It is difficult to conceive of any other information technology that has developed so rapidly in recent years, in terms of real impact.

The fields of machine learning and optimization are highly interwoven. Optimization problems form the core of machine learning methods and modern optimization algorithms are using machine learning more and more to improve their efficiency.

Machine learning has applications in all areas of science. There are many learning methods, each of which uses a different algorithmic structure to optimize predictions, based on the data received. Hence, the first objective of this book is to shed light on key principles and methods that are common within both fields.

Machine learning and optimization share three components: representation, evaluation and iterative search. Yet while optimization solvers are generally designed to be fast and accurate on implicit models, machine learning methods need to be generic and trained offline on datasets. Machine learning problems present new challenges for optimization researchers, and machine learning practitioners seek simpler, generic optimization algorithms.

Quite recently, modern approaches to machine learning have also been applied to the design of optimization algorithms themselves, taking advantage of their ability to capture valuable information from complex structures in large spaces. Those capacities appear to be useful, especially for the representation and evaluation components. As large, complex structures are ubiquitous in optimization problems, and can be used as huge implicit datasets, the use of machine learning enabled the efficiency and genericity of optimization methods to be improved.

This book presents modern advances in the selection, configuration and engineering of algorithms that rely on machine learning and optimization. It is structured into two parts. Part 1 is dedicated to the most common optimization applications. Part 2 describes and implements several applications of machine learning.

Part 1 comprises four chapters which focus on real-world application of optimization algorithms.

Chapter 1 addresses the problem of vehicle routing with loading constraints and combines two combinatorial optimization problems: the capacity vehicle routing problem (CVRP) and the two-/three-dimensional bin packing problem (2/3D-BPP). The authors have studied real transport problems such as the transport of furniture or industrial machinery.

The main objective of Chapter 2 is to create the most appropriate scheduling solution that optimizes several QoS metrics simultaneously; thus, the authors adapt the widely used metaheuristic, “Genetic Algorithm” as an optimization method. The proposed scheduling approach is tested by simulating a healthcare IoT application, modeled as a workflow and several scientific workflow benchmarks. The results show the effectiveness of the proposed approach; it generates a scheduling plan that better optimizes the various QoS metrics considered.

Chapter 3 focuses on the grey wolf optimization (GWO) and its adaptation to a continuous search space. It begins by addressing the mathematical modeling of optimization in a binary discrete search space. Binarization modules are then provided, allowing continuous metaheuristics for the solution of feature selection problems in a binary search space. These binarization modules are then used to create the binary metaheuristic bGWO. Finally, an experimental demonstration shows the performance of bGWO in solving feature selection problems on 18 datasets from the UCI Machine Learning Repository

Chapter 4 addresses the type-2 mixed-model assembly line balancing problem with deterministic task times. To solve this problem, an enhancement of the greedy randomized adaptive search procedure – known as the reactive greedy randomized adaptive search procedure – is proposed. This reactive version is based on variation of the restricted candidate list parameter value, alpha. The proposed reactive GRASP is hybridized with the ranked positional weight heuristic to construct initial solutions. Results obtained by the proposed hybrid reactive GRASP are compared with those obtained by the basic GRASP, demonstrating the effect of the learning mechanism.

Part 2 comprises four chapters devoted to artificial intelligence and machine learning and their applications.

The main challenge of recommender systems comes from modeling the dependence between the various entities, incorporating multifaceted information such as user preferences, item attributes and users’ mutual influence, which results in more complex features. To deal with this issue, the authors of Chapter 5 design stacked ensemble machine learning models for recommendations. Their recommender system incorporates a collaborative filtering (CF) module and a stacking recommender module. An interactive attention mechanism is then introduced to model the mutual influence relationship between aspect users and items. Experiments on real-world datasets demonstrate that the proposed algorithm can achieve more accurate predictions and higher recommendation efficiency.

In internal auditing, the ability to process all of the available information related to the audit universe or subject could improve the quality of results. Classifying the audit text documents (unstructured data) could enable the use of additional information to improve the existing structured data, creating better knowledge support for the audit process. A comparison of results of classical machine learning and deep learning algorithms, combined with advanced word embeddings to classify the findings of internal audit reports, is presented in Chapter 6.

The design of a control architecture is a central problem in a project to realize an autonomous mobile robot. In the absence of a generic solution, it is essential to come up with a new approach detailing the design process of an intelligent system that is capable of adapting to all changes in the navigation environment. Chapter 7 proposes to use the multiagent paradigm and fuzzy logic in the design of the control architecture for the autonomous navigation of the mobile robot in a constrained environment. The control architecture is designed to solve various problems created during navigation. It is made up of four agents: the perception of the agent, the feasibility of the agent, the locomotion agent and the fuzzy control agent.

Intrusion detection is a key concept in modern computer network security. It is aimed at analyzing the current state of a network in real time and identifying potential anomalies in the system, reporting them as soon as they are identified. This allows for the detection of previously unknown malware. Artificial neural networks are supervised machine learning algorithms inspired by the human brain. This kind of network is a popular choice among data mining techniques today and has already been proven to be a valuable choice for intrusion detection. In Chapter 8, the author builds a feed-forward neural network trained on the NSL-KDD dataset, in order to classify network connections as belonging to one of two possible categories: normal or anomalous. Its goal is to maximize the level of accuracy in recognizing new data samples.