In what would be of interest to a large company that plans to establish warehouses or offices across the country, virtual ants run across the computer screen exhaling greater or lesser amounts of pheromone to indicate the best location options, according to a series of economic criteria, including population density. “In demonstrating the potential of this tool, problems of great practical interest are solved. Example: recording the 186 most populous cities on the map of Brazil, you can find the best locations for installing a fixed number of offices or distribution points. Considering only population density, the State of São Paulo is always considered, especially the region of Campinas”, explains professor Fernando José Von Zuben, from the Faculty of Electrical and Computer Engineering (FEEC) at Unicamp.
Von Zuben coordinates the recently opened Bioinformatics and Bio-inspired Computing Laboratory (LBiC), which was created centralizing research by 21 post-doctoral students, doctoral students, master's students and scientific initiation students. LBiC is supported entirely by research projects funded by CNPq and Fapesp. In addition to the new laboratory, bio-inspired computing is the subject of the book Recent Developments in Biologically Inspired Computing (Idea Group Publishing), organized by the FEEC professor and Leandro Nunes de Castro, professor of the Master's Program in Informatics at the Catholic University of Santos and LBiC collaborator, who signed the introductory chapter. In total, there are 16 chapters with authors from ten countries: Japan, France, Italy, England, United States, Canada, Australia, Portugal, Spain and Brazil.
“The book covers all bio-inspired areas, such as artificial immune systems, ant colonies, collective behavior of robots, artificial life and DNA computing. Our objective was to generate a range of options, homogenizing the content that supports various aspects within bio-inspired computing. The book project included submissions of work that were evaluated by a broad body of reviewers, resulting in conceptual chapters and also reports on advanced experiments, such as those on autonomous robotics by the Japanese. They are really diverse groups and it took a lot of work to find the coherence behind this divergence of thoughts”, says Fernando Von Zuben.
The professor attributes to Leandro de Castro not only the idea of the book, but the very feasibility of important lines of research in the laboratory now installed at FEEC. “He had already defended his doctoral thesis at FEEC/Unicamp and went on to do a post-doctorate in England in the same line of research, when he then published, in co-authorship, the first textbook on artificial immune systems. With a Research Project in Bio-Inspired Computing, Leandro de Castro returned to Brazil to participate in a Doctoral Attachment Program (Profix) financed by CNPq and proposed this new work to me. The success of his research project was essential for funding agencies to approve other works in bio-inspired computing, which complement the line of action in bioinformatics that was already underway, giving rise to our laboratory”, explains the researcher.
Leandro de Castro states that bio-inspired computational algorithms can be successfully employed in solving data mining problems, including information clustering on the Web, as well as many other engineering and optimization problems, such as autonomous robot navigation, combinatorial optimization and multimodal search. All these lines of research are in full swing at LBiC.
Environment – Von Zuben remembers that insect colonies have been studied for a long time, inside and outside biology. This study typically aims to explore aspects of hierarchy and self-organization, as well as social relationships and the role of each individual in the colony. He notes, however, that this work by naturalists, with insects and many other species of fauna, has never been properly valued. “It was not believed, for example, that carnivorous mammals used sophisticated hunting strategies, carrying out coordinated attacks to weaken the defense of a herd, separate prey and place it on the trail with positioned hunters. This line of research was only recognized in 1973, with the Nobel Prize for Austrian zoologist Konrad Lorenz, for his comparative study of human and animal behavior. If today we have bio-inspired computing, it is due to the migration of focus also to the observation of nature”, he criticizes.
Another area also bio-inspired is spatio-temporal modeling, which can serve research on the reproduction of cellular organisms and the growth of coral reefs, the dispersion of pollutants or the devastation of forests, simulating future scenarios based on the dynamics presented today. “In the case of the existence of underwater pipelines for transporting oil, for example, it is possible to obtain a much more reliable environmental impact report. Among countless variables, it is necessary to consider that there will be one type of impact if the currents bring the oil to the coast, and another type of impact if the direction is the high seas, and that in these regions the currents change during the year”, he illustrates. Von Zuben.
In a more scientific context, the FEEC professor states that bio-inspired computing can also help in modeling natural evolutionary processes. “We have more and more technology to measure how much one species differs from another, but we ignore ancestry, that is, we do not witness the process of differentiation that led to what we have today. It is necessary to propose hypotheses to explain this differentiation, and the identification of the most credible hypotheses cannot be carried out without the help of sophisticated computational algorithms. Among the proposals that have been considered in the literature, bio-inspired algorithms for phylogenetic inference occupy a prominent position”, points out the researcher.
in art – Bio-inspired computing proposes solutions including in the field of art, aimed at researchers who work with the help of the computer to the artist – and not necessarily in the production of art by computer. In interaction with the Interdisciplinary Center for Sound Communication (NICS) at Unicamp, LBiC welcomes scholarship students who work in sound synthesis, in the production of sound bases for compositions. “Using a co-evolutionary approach inspired by a prey-predator system, the goal of 'prey' sounds is to move away from 'predator' sounds, and the goal of the latter is to get closer to the former. By measuring the attributes of these sounds and the distance between them on the computer, one travels in the sound space, producing very creative sound strands”, explains Von Zuben.
According to the professor, another challenging area for the laboratory is signal processing, given the rapid evolution of telecommunications. As the main researcher of a Thematic Project funded by Fapesp and coordinated by professor João Marcos Travassos Romano (FEEC/Unicamp), he works on the development of supervised and unsupervised approaches for adaptive filtering. “Without mentioning specific applications, the basic objective is to explore emerging technologies to make telecommunications projects cheaper and also extend the functionality of existing devices,” he explains.
Interaction – Fernando Von Zuben clarifies that the Bioinformatics and Bio-Inspired Computing Laboratory essentially deals with problems that are difficult to solve using conventional methodologies, and it is essential to identify peculiarities that make treatment through bio-inspired computing interesting. “In several bioinformatics problems, for example, it is not relevant to look for the 'optimal' (single) answer, but rather to propose multiple solutions, all taking into account relevant particularities of the problem. Afterwards, the alternatives will be presented for analysis, which can be based on various scenarios. In short, the specialties of our research group are in multidisciplinary, interdisciplinary and transdisciplinary action, so that LBiC cannot be sustained without interaction with other research groups, aiming for joint, complementary and cooperative action”, he concludes.
AAnts have the peculiar power of discovering food sources in a vast territory around the nest, communicating the discovery to their companions and showing them the path to follow. Among other forms of chemical communication, the ant marks the trail with the pheromone it releases as it walks and, if there is a fork in the path, its companions tend to follow the trail where the concentration of the pheromone is greater. Bio-inspired computing – a specialty that has been gaining strength alongside bioinformatics – uses this and other analogies with natural situations to solve human problems. An analogy occurs when transporting “pheromone” trails to a road map on the computer, using probability theory to propose the best path among a combinatorial explosion of possibilities.
PRESS ROOM - � 1994-2003 State University of Campinas / Press Office