PrefaceIntelligence is one of the qualities that distinguish the human being from all the other animals. In spite of having inferior physical abilities to those of many other species, human intelligence has allowed him to overcome them and to become the dominant species. The ability to manipulate objects, and more fundamentally, his superior intellect, has allowed him to create tools and to develop complex reasoning and plans impossible for other species. This has allowed him, for example, to adapt to practically all conditions of planet Earth.
However, human inventiveness has been used not only to overcome the various problems in order to adapt and survive. Intellectual development entails different concerns, and perhaps one of the most interesting has been the determination to create a human being artificially, trying to simulate him both physically and intellectually. In general, this fact is applicable to the artificial development of a living organism.
From a physical point of view, the attempts to create a living organism artificially are well-known, and they have been carried out for a number of years. Sometimes, due to the fact they are made highly complex by means of clock mechanisms, these hominids and robots have reached such a point that they are able to reproduce everyday movements, usually succeeding to deeply impress those people who had the chance to see them. One early example is given in the Roman Empire, in which Heron of Alexandria built artificial actors that represented the Trojan War. Other significant examples are "The Pigeon" of Archytas of Tarentum, "The Mechanical Lion" of Leonardo da Vinci or "The Flute Player"-life-size figure of a shepherd that played the tabor and the pipe and had a repertoire of twelve songs, "The Tambourine Player" and "The Digesting Duck", with over 400 moving parts, able to flap its wings, drink water, digest grain and defecate, the latter three automatons belonging to Jacques Vaucanson. Other examples are the robots invented by Wolfgang von Kempelen, "The Turk", a chess-playing machine that played against the best opponents of its era, but later it was revealed to be a farce, and "The Mechanism of human speech”, this time not an automaton but a talking machine.
From an intellectual point of view, there are also different approaches on creating machines that either exhibit a behavior that reproduces a certain level of intellect when it comes to solving a problem, or represent a support to humans in achieving intellectual tasks. The latter is the case, for example, of the abacus, designed to improve calculating abilities. Regarding the development of machines or systems that solve problems, they are a very old dream. The Arab mathematician al-Jwarizmi had already laid the foundations of algebra and especially of algorithmics around the 9th century. The algorithm is known as an established, ordered and finite step-by-step procedure for solving a problem. Given an initial state and an input, through well-established recursive steps, a final state is reached, which leads to the solution of the problem. Therefore, this set of steps could be programmed into a machine, so that it is able to solve a particular problem.
However, it was only with the emergence of the first computers and software when systems with features of living beings, such as memory or calculation, were developed and when complex algorithms that allowed the solution of problems began to be programmed. These algorithms, although they gave good results, lack some features, essential for the living beings, such as adaptation or learning abilities. Overall, the first algorithms that led unequivocally to the solution of a problem, had not the proper functioning when facing somehow uncertain data, noise or inaccuracy, as it usually occurs with real-world data. In an informal way, the techniques based on such algorithms are known as "hard computing" techniques.
In comparison to these techniques, a number of techniques aimed at solving real problems by imitating the way humans do it, have also arisen. They are known as "Soft Computing". This term was coined by Lotfi A. Zadeh in 1991, and ever since it has undergone a quick development both regarding the theoretical aspects, and above all, its applications. Soft Computing techniques approach problems of great diversity, both in their type (modeling, optimization, planning, monitoring, forecasting, data mining, ...) and in the field of their implementation (industrial production, telecommunications, energy, logistics, banking, food processing, …)
Therefore, Soft Computing techniques could be considered as a branch of Artificial Intelligence (AI) focused on the design of intelligent systems able to operate properly with inaccurate, unclear and/or incomplete information. This property enables to approach real problems in order to find more reliable, manageable and less expensive solutions compared to those obtained by means of conventional techniques. The main techniques that make up Soft Computing are fuzzy logic, neural networks, evolutionary computation and probabilistic reasoning.
AI is a multi-disciplinary science field that deals with the study in depth of the possibility of creating artificial beings. Its starting point was Babbage's determination that his machine could "think, learn and create," so that the ability to perform these actions might be increased and applied to the problems that human beings face. AI, whose name is attributed to John McCarthy from the College Dormouth group in the summer of 1956, is divided into two main branches, known as symbolic and connectionist, depending on whether it is attempted to simulate or emulate the human brain in intelligent artificial beings, respectively. According to McCorduck and McCarthy, artificial beings are considered intelligent if they show a behavior that, when performed by a biological being, could be considered intelligent.
Nevertheless, and despite the high rate at which developments occur today, we are still far from reproducing artificially something that is inherent in all living beings, such as creativity, critical capacity (including self-critical capacity), consciousness or common sense, among others.
Although we are still far from reproducing the daily behavior of biological systems, all these studies and researches have achieved spectacular results. In recent decades, the attention of the scientific community has increased because of recent advances in computing, the latter due to the fact that the miniaturization of computers has evolved along with their increase in the information computing and storage capacities. Thus, more complex systems are being developed progressively in order to carry out more complex functions.
The efforts made so far approach two different situations. On the one hand, they are the basis for all the advances achieved so far in order to reproduce the defining features of the living beings. On the other hand, they also reflect the poor-but-spectacular advances regarding the creation of real intelligent beings. In spite of the fact that the connectionist systems are the most advanced in the field of emulation of intelligent biological systems, the latter show certain restrictions. These limitations are related mainly to the need to reduce training time and to optimize the architecture and also to the lack of explanation about their behavior. It is required to look back to Nature again, as it was done when great strides were taken to this end, to seek new information to inspire the search for these solutions. In general, Nature has provided guidance for the creation of a large number of techniques which are encompassed within the AI and Soft Computing.
Technology also provides solutions. In this respect, different technologies are meant to be integrated under a common label: MNBIC (Micro and Nanotechnologies, Biotechnology, Information Technologies, and Cognitive Technologies) Convergent Technologies. MNBIC is expected to be a revolution in scientific, technological and socio-economic fields since it helps making possible the construction of hybrid systems: biological and artificial.
Some possibilities consist of the use of micro- or nano-elements that could be introduced into biological systems with the aim of replacing damaged or non-functional parts, whereas biological particles could be embedded in artificial systems in order to perform certain actions. According to a recent report of the U.S. National Science Foundation, “The convergence of micro and nanoscience, biotechnology, information technology, and cognitive science (MNBIC) offers immense opportunities for the improvement of human abilities, social outcomes, the nation’s productivity, and its quality of life. It also represents a major new frontier in research and development. MNBIC convergence is a broad, cross-cutting, emerging, and timely opportunity of interest to individuals, society, and humanity in the long term.”
There is a significant scientific agreement on the fact that the most complex part to be integrated with the rest of converging technologies is the one representing cognitive science. The issues related to knowledge technologies have the highest level of integration by means of knowledge engineering models.
In general, considering the multidisciplinary nature of the AI and Soft Computing techniques, the scientific community benefits from their employment, since they can be applied in a wide variety of different fields. This publication presents a series of practical applications of different Soft Computing techniques to real-world problems. The aim is to show the enormous potential of these techniques in solving all kinds of problems. Thus, with the latest advances in these techniques, an extensive state-of-the-art and a vast theoretical study on them are provided.
Therefore, Soft Computing techniques are presented from a theoretical point of view, but above all from a practical point of view. This can be of great use to a student who wants to lean towards this branch of AI, as well as for a researcher that needs to become familiar with these techniques. Likewise, a professional or experienced researcher will also find this reading material very useful as it offers novel and recent applications, which gives a boost to expanding and exploring new research areas.
This publication is divided into four sections, which are meant to cover different areas of knowledge in which Soft Computing has been applied in order to solve various problems.
The first section is devoted to various tasks related to the employment of Soft Computing techniques in the field of information processing. This area is of great importance, since, due to its multidisciplinary nature, these techniques may be applied in several different fields.
Chapter I provides an introduction to two of the most commonly used techniques in the Soft Computing: Artificial Neural Networks, Genetic Algorithms and Genetic Programming, the latter two categorized as being part of the Evolutionary Computation techniques
Chapter II presents a new technique within Evolutionary Computation, inspired by the adaptation of biological cells. This technique is used in this work to solve information processing problems.
Chapter III uses Soft Computing techniques to carry out successfully two completely different applications in the field of human-computer interaction.
Chapter IV shows an application in the field of image processing: how it is possible to segment color images using a special type of Artificial Neural Networks: LVQ.
Chapter V deals with 3D modelling. In this chapter, AI techniques are used in order to implement 3D environments, and, on the other hand, 3D models are employed in order to make the most of the AI techniques.
Chapter VI describes the use of Soft Computing techniques with the aim of developing computer user models. If this goal is achieved, a series of systems improving the interaction with the users could be created.
The second section, Industrial Applications, describes different ways of applying these techniques to solve various problems related to distinct industrial and, on the whole, engineering processes. Chapter VII describes how a Genetic Algorithm can be used in order to optimize the design of single- and dual-band antennas.
The basis of Chapter VIII is the effect exerted by the chemical emissions of motor vehicles on the environment. This chapter provides an analysis of these chemical emissions using various techniques.
Chapter IX shows a study in which a modelling of laser milling manufacture of steel components has been performed. As a result, this manufacturing process is optimized.
Chapter X is related to an issue of Civil Engineering. This chapter describes how to use a technique of Evolutionary Computation, Genetic Programming, in order to predict the behavior of structural concrete.
Chapter XI presents the use of Evolutionary Computation techniques in the field of hardware development. More specifically, intrinsic evolvable hardware gives the possibility of being integrated into applications with high degree of autonomy and which require real-time response.
Chapter XII exposes a study in which Artificial Neural Networks were used in order to extract different atmospheric parameters (temperature, surface gravity, etc.) characterizing a particular star.
The third section contains a series of works related to the use of Soft Computing in the field of medicine and bioinformatics. Chapter XIII shows an application in which different techniques are used in order to give rise to a system capable of an automatic arrhythmia detection in ECG signals.
Chapter XIV describes a new implementation of the Iterative Rule Learning algorithm for extracting the genotype-phenotype association rules in complex diseases.
Chapter XV describes an application in the field of Ontology. More specifically, it is shown how to perform the automatic combination of several measures into a single metrics using an Evolutionary Computation technique.
Finally, the fourth section, on Natural Environment Applications, is focused on the Soft Computing use to solve issues directly related to the environment, such as natural resource management or prediction or modelling of environmental effects. Chapter XVI shows the employment of an Artificial Neural Network to model net radiation starting from different meteorological parameters like temperature, precipitation, wind speed, etc.
Chapter XVII describes a method to perform a prediction of tropospheric ozone in the atmosphere, of great importance to public health, using Machine Learning techniques.
Chapter XVIII presents a method to model PMx pollutant dispersion, with the aim of determining the spatio-temporal behavior of the concentration of these particles, using Evolutionary Computation techniques.
Chapter XIX shows how Soft Computing techniques can be used to solve different challenges related to natural resources management: data analysis, optimization and control.
Chapter XX is related to the use of self-organizing maps employed for the study of soil pollution caused by road traffic, and sea pollution as a consequence of oil spillage.
Chapter XXI focuses on the prediction of a weather phenomenon: rain. Starting from observations of vegetation activity and several climatic data, a model capable of predicting precipitation has been developed by means of different techniques.
As it can be seen, this publication aims at providing an outlook of the most recent works in the field of Soft Computing, including not only theoretical aspects but also describing implementations of these systems in fields with very different characteristics, thus proving the multi-disciplinary nature of these techniques.
That being said, we consider this book a proposal which, on the one hand, will contribute to solving a lot of problems but that, on the other hand, will also open new questions that will undoubtedly have a decisive role in the research progress within this field.
This book does not provide definitive solutions but it contributes to the creation of new and imaginative viewpoints instead.