When reading in the book title the words “Artificial Intelligence”, the image of a computer which possesses that amazing capacity characterizing human beings immediately springs to mind. This capacity is nothing but the result of several biological processes which faithfully follow the laws of physics and chemistry but which at present remain mostly unknown. A computer is, perhaps, the best device to represent technology and, besides, its use may contribute to the development of science.
Science and technology may be considered as two sides of the same coin. Human beings have always used devices which have helped them to expand their knowledge and they have used that newly-acquired knowledge to build new mechanisms. It is impossible to tell which of both aspects prevails in this necessarily symbiotic process. One of the greatest scientific challenges nowadays lies in understanding closely the functioning of biological systems, knowing the complex relations among their elements and reasons for the behavior of live beings, as well as their mutual relations. In this sense, technology assists human beings in their consecution of smaller or greater achievements. It also contributes to a gradual reduction of the time required to overcome this challenge.
In parallel, if a greater understanding of biological systems is achieved, numerous scientific fields may benefit from it, since their work models or systems are based on the behavior of the said biological systems. This is true because a great number of scientific fields, and not just Artificial Intelligence, are searching for solutions to complex problems in biological processes, due to the lack of success in reaching good solutions with other strategies. This happens in areas such as engineering, computing, bioinformatics, etc. Anyhow, this is no easy task, given that any biological process is characterized by having a notorious complexity and being conditioned by multiple parameters which may result in different behaviors. With this borne in mind, it is certainly a hard task to get computers to simulate or even imitate biological processes.
The types of problems that people try to solve by considering biological processes are usually characterized by the absence of a complete mathematical model of the relevant phenomena, the existence of a large number of variables to be adjusted and conditions to be simultaneously satisfied, the presence of high degrees of non-linearity, and the formulation of multi-purpose tasks with a combinatorial explosion of candidate solutions, among other challenging scenarios which become even more complicated due to the need to act in real time in most occasions. Another motivation to be highlighted is the realization that most complex problems have similar versions in the natural world. Adaptation, self-organization, communication, and optimization have to be performed by biological organisms, in parallel and in different scales and structural levels, in order to survive and maintain life. Nature is constantly developing efficient problem-solving techniques and these are now beginning to be explored by engineers, computer scientists and biologists for the design of artificial systems and algorithms to be applied to the greatest variety of problems.
There are many systems based upon biological processes. The initial section of the present book focuses on the description of biological processes whose behavior has very recently been discovered. Numerous models and techniques which imitate nature can be developed within the scope of Artificial Intelligence. These are portrayed in the second section of the book. Among all of them, we could highlight, e. g., Connectionist Systems as models of the human brain (McCulloch & Pitts, 1943), Evolutionary Computing which imitates the processes of species evolution (Holland, 1975; Holland, 1992; Koza, 1992; Whitley, 1999), Biomolecular Computation, which is a scientific discipline concerned with processing information encoded in biological macromolecules such as DNA, RNA or proteins, although the most important advances have been made using DNA (Adleman, 1994; Lipton, 1995; Rodríguez-Patón, 1999), Swarm-intelligence systems based heavily on the emergence of global behavior through local interactions of components (Bonabeau et al., 1999), Robotics, etc. Moreover, these systems once built can be applied to multiple scientific fields. The third section of this book presents some examples of these applications in a series of very interesting works developed by researchers from different centres across the five continents.
ARTIFICIAL INTELLIGENCE AND ITS RELATION TO OTHER SCIENCES
It may be generally stated that Artificial Intelligence is a branch of science whose aim is to build systems and machines showing a type of behavior which, if developed by a person, would be termed as "intelligent". Learning, the capacity to adapt to changing environments, creativity, etc., are skills usually related to intelligent behavior.
It is assumed that the origins of Artificial Intelligence lie in the attempts by human beings to increase their physical and intellectual capabilities by creating artifacts with automatisms, simulating their shape and skills.
The first written reference to an intelligent artificial system appears in Greek mythology, specifically in The Iliad, and it is attributed to Hephaestus, god of fire and forges, who made "two servants of solid gold with intelligent minds and the ability to speak”.
In the Middle Ages, Saint Albert the Great built a "butler" who would open the door and greet visitors.
In the Modern Age (s. XVII) the Drozs, famous Central European watch-makers, built three androids: a boy who wrote another one who drew and a girl who played the organ and pretended to breathe. This achievement was based on clock-work devices and it was the reason for their arrest and imprisonment by the Inquisition.
As regards innovations in the XIX and XX centuries, we should note the works by Pascal, Leibnitz, Babbage and Boole. Also Ada Lovelace, Babbage's collaborator and Lord Byron's wife, who wrote in the well-known Lovelace's regime, that: "machines can only do everything that we can tell them how to do. Their mission is to help facilitate the already known". This is still in force for "conventional computing" but it was left behind by the progress made in Artificial Intelligence, thanks to automatic learning techniques, among other things.
The immediate origin of the concept and development criteria of Artificial Intelligence goes back to the insight by the English mathematician Alan Turing (1943), inventor of the “COLOSSUS” machine created to decipher the messages encrypted by Nazi troops through the “ENIGMA” machine, while the "Artificial Intelligence" term was coined by John McCarthy, one of the members of the "Dartmouth Group”, which gathered in 1956 with funding from the Rockefeller Foundation in order to discuss the possibility to build machines which would not be limited to making pre-fixed calculations, but "intelligent" operations. Among the group members were: Samuel (1959), who had written a chequers program capable of learning from its own experience, therefore overcoming the already mentioned Lovelace's regime; McCarthy (1956), who studied systems capable of making common-sense reasoning; Minsky (1997), who worked with analogical geometry reasoning; Selfridge (1986), who studied computer-aided visual recognition; as well as Newell, Shaw and Simon, who had written a program for the automatic demonstration of theorems.
Two main Artificial Intelligence "schools" were born from this initial group: Newell and Simon (1963) were the leaders of the Carnegie-Mellon University Team, with the intention to develop human behavior models with devices whose structure resembled the brain as close as possible, which later on led to the "connectionist" branch and to the works on "Artificial Neural Networks" also known as "Connectionist Systems".
McCarthy and Minsky integrated another team in the mythical Massachusetts Institute of Technology (MIT), focusing on the processing results having an intelligent nature, without being concerned whether the components functioning or structure are similar to those of human beings or not.
Nevertheless, both approaches correspond to the same priority goals of Artificial Intelligence: understanding natural human intelligence and using intelligent machines in order to acquire knowledge and to solve problems which are considered to be intellectually hard. But in traditional Artificial Intelligence (MIT school) which was more successful in the first twenty years, researchers found out that their systems collapsed before the increasing length and complexity of their programming. Stewart Wilson, a researcher from the Roland Institute (Massachusetts), realised that something was wrong in the field of “traditional” Artificial Intelligence. He wondered about the roots of intelligence and was convinced that the best way to understand how something works in a human being is to understand it first in a simpler being. Given that, essentially, Artificial Intelligence tried to replicate human intelligence, Wilson decided that first he should replicate animal intelligence (Meyer & Wilson, 1991). This idea had never been popular among Artificial Intelligence researchers but he, together with others, soon turned it into a first informal principle of a new approach to Artificial Intelligence based on nature. From that point on, various Artificial Intelligence techniques based on biological principles were created and continue to surface.
Therefore, numerous researchers coming from different scientific fields have taken part in the creation and setting of the Artificial Intelligence foundations. Following what has been mentioned at the beginning of the present book, about what Artificial Intelligence suggests at first sight, we may find a clear interrelation between the various branches of science: Computer Science (computer) – Neuroscience (intelligence). Other relations appear between branches of science which are, in principle, very different, such as: Computer Science – Psychology, Computer Science – Sociology, Computer Science – Genetics, etc. All of these interrelations which gradually surfaced through time have led to innovations which benefit at least one of the two related scientific areas.
Consequently, although it cannot be determined which is the best way to proceed, it is clear that both in Artificial Intelligence and in its foundations or application areas, interdisciplinarity and dissemination of results do facilitate reaching relevant conclusions. If we tried to reach those conclusions only from Computer Science, they would have never been achieved. Therefore, it is obvious that science should not be based only on proofs obtained from a single perspective, since this would limit its scope of action. Other viewpoints should also be considered, viewpoints based upon different theories which bring about ideas facilitating the generation of hypothesis about the phenomenon to be studied on each occasion.
ORGANIZATION OF THE BOOK
This book contains twenty contributed chapters written by internationally renowned researchers and it is organized into three main sections. A brief description of each of the sections follows:
SECTION I presents recent advances in the biological processes related to information processing. Such advances refer to the area of Neuroscience. The authors not only show the results of the study of biological phenomena, but they also propose several options for the later ones to be the basis of the elaboration of new Artificial Intelligence models and techniques.
Chapter I Recent electrophysiological studies indicate the existence of an important somatosensory processing in the trigeminal nucleus which is modulated by the corticofugal projection from the somatosensory cortex. This chapter studies a new mathematical analysis of the temporal structure of neuronal responses during tactile stimulation of the spinal trigeminal nucleus.
Chapter II Knowledge in invertebrate neuroethology has demonstrated unique advantages for engineering biologically-based autonomous systems. This chapter aims at presenting some basic neuronal mechanisms involved in crayfish walking and postural control involving a single key joint of the leg. Due to its relative simplicity, the neuronal network responsible for these motor functions is a suitable model for understanding how sensory and motor components interact in the elaboration of appropriate movement and, therefore, for providing basic principles essential to the design of autonomous embodied systems.
Chapter III reviews the underlying mechanisms and theoretical implications of the role of voltage-dependent dendritic currents on the forward transmission of synaptic inputs. The notion analysed brakes with the classic view of neurons as the elementary units of the brain and attributes them computational/storage capabilities earlier billed to complex brain circuits.
SECTION II - Illustrate some of the more recent biologically inspired Artificial Intelligence models involved in information processing. The biological processes that serve as inspiration for the models are related to the fields of Neuroscience, Molecular Biology, Social Behaviour, etc. Not only the different approaches for the design and construction of these models are reviewed, but also their different areas of application are introduced.
Chapter IV is a quick survey of spiking neural P systems, a branch of membrane computing which was recently introduced with motivation from neural computing based on spiking.
Chapter V presents an evolution of the Recurrent ANN (RANN) to enforce the persistence of activations within the neurons to create activation contexts that generate correct outputs through time. The aim of this work is to develop a process element model with activation output much more similar to the biological neurons one.
Chapter VI presents the functioning methodology of the Artificial NeuroGlial Networks; these artificial networks are not only made of neurons, like the artificial neural networks, but also they are made of elements which try to imitate glial cells. The application of these artificial nets to classification problems is also presented here.
Chapter VII describes the experience gained when developing the path generation modules of autonomous robots, starting with traditional artificial intelligence approaches and ending with the most recent techniques of Evolutionary Robotics. Discussions around the features and suitability of each technique, with special interest on immune-based behavior coordination are proposed to meet the corresponding theoretical arguments supported by empirical experiences.
Chapter VIII. In this chapter, two important issues concerning associative memory by neural networks are studied: a new model of Hebbian learning, as well as the effect of the network capacity when retrieving patterns and performing clustering tasks.
Chapter IX contents a computational model which is inspired in the biologically morphogenesis ideas. This chapter contains the theoretical development of the model and some simple tests executed over an implementation of the theoretical model.
Chapter X shows the interrelations between computing and genetics, which both are based on information and, particularly, self-reproducing artificial systems.
Chapter XI discusses guidelines and models of Mind from Cognitive Sciences in order to generate an integrated architecture for an artificial mind that allows various behavior aspects to be simulated in a coherent and harmonious way, showing believability and computational processing viability.
Chapter XII presents a general central pattern generator (CPG) architecture for legged locomotion. Based on a simple discrete distributed synchronizer, the use of oscillatory building blocks (OBB) is proposed for the production of complicated rhythmic patterns. An OBB network can be easily built to generate a full range of locomotion patterns of a legged animal. The modular CPG structure is amenable to very large scale circuit integration.
SECTION III - Presents real-life applications of recent biologically inspired Artificial Intelligence models and techniques. Some works use hybrid systems which combine different bioinspired techniques. The application of these models to such different areas as Mathematics, Civil Engineering, Computer Science, Biology, etc. is described here.
Chapter XIII tries to establish, the characterization of the multimodal problems and offers a global view of some of the several approaches proposed for adapting the classic functioning of the Genetic Algorithms to the search of multiple solutions. The contributions of the authors and a brief description of several practical cases of their performance at the real world will be also showed.
Chapter XIV focuses on the description of several biomolecular information-processing devices from both the synthetic biology and biomolecular computation fields. Synthetic biology and biomolecular computation are disciplines that fuse when it comes to designing and building information processing devices.
Chapter XV Swarm-intelligence systems involve highly parallel computations across space, based heavily on the emergence of global behavior through local interactions of components. This chapter describes how to provide greater control over swarm intelligence systems, and potentially more useful goal-oriented behavior, by introducing hierarchical controllers in the components.
Chapter XVI proposes a bio-inspired approach for the construction of a self-organizing Grid information system and also describes the SO-Grid Portal, a simulation portal through which registered users can simulate and analyze ant-based protocols. This chapter can foster the understanding and use of swarm intelligence, multi-agent and bio-inspired paradigms in the field of distributed computing.
Chapter XVII presents Graph Based Evolutionary Algorithms (GBEA). GBEA are a generic enhancement and diversity management technique for evolutionary algorithms. GBEA impose restrictions on mating and placement within the evolving population (new structures are placed in their parents’ graph neighborhood). This simulates natural obstacles like geography or social obstacles like the mating dances of some bird species to mating.
Chapter XVIII aims to explain and analyze the connection between Artificial Intelligence domain requirements and the Theory of Systems with Several Equilibria. This approach allows a better understanding of those dynamical behaviors of the artificial recurrent neural networks which are desirable for their proper work i.e. achievement of the tasks they have been designed for.
Chapter XIX involves an application of Artificial Intelligence in the field of Civil Engineering, specifically the prediction of longitudinal dispersion coefficient in rivers. Based on the concept of Genetic Algorithms and Artificial Neural Networks, a novel data-driven method called GNMM (Genetic Neural Mathematical Method) is presented and applied to a very well-studied classic and representative set of data.
Chapter XX concludes this book. The authors employ the fuzzy decision tree classification technique in a series of biological based application problems. A small hypothetical example allows the reader to clearly follow the included analytical rudiments, and the larger applications demonstrate the interpretability allowed through the use of this approach.
CONCLUSIONS
As time elapses, researchers are discovering the way living beings behave, the way their internal organs function, how intercellular substance exchange takes place and which is the composition of the smallest molecule. Most of these biological processes have been imitated by many scientific disciplines with the purpose of trying to solve different complex problems. One of these disciplines is Artificial Intelligence, where diverse research teams have analysed and studied from the late XIX century some biological systems implementing computational models to solve real life problems. Ramon y Cajal (1904), the Spanish Nobel prize winner, used the "daring" and currently used term Neural Engineering even before the XX century started.
This book intends to be an important reference book. It shows recent research works of great quality about information processing. It presents biological processes, some of them recently discovered, which show the treatment of different life processes at distinct levels (molecular, neural, social, etc.) that can be applied to Artificial Intelligence. It also presents the benefits of the new techniques created from these processes, not only to Artificial Intelligence, but also to multiple scientific areas: computer-science, biology, civil engineering, etc.