In this project we are developing a novel approach for searching in high-dimensional spaces taking into account behaviors drawn from fish schools. The derived algorithm – Fish-School Search (FSS) – is mainly composed of three operators: feeding, swimming and breeding. Together these operators afford the evoked computation: (i) wide-ranging search abilities, (ii) automatic capability to switch between exploration and exploitation, and (iii) self-adaptable global guidance for the search process.
Simulations showed that the FSS algorithm can outperform well-known intelligent algorithms such as Particle Swarm Optimization searches in some cases.
Venn networks are computational intelligent architectures (i.e. a type of artificial neural network), proposed by the Buarque, which have the ability to mimic some brain function at the same time their internal activity resemble what is seen in functional imaging of the brain.
Inspired by the morpho-functional organization of the brain, Venn-networks allow: (i) use of different type of processing units; (ii) definition of distinct regions within the network structure; (iii) use of a wide variety of connection type (among processing units); and, (iv) definition of a non-trivial connectivity based on the selection of fibers available.
This research project aims at implementing (on the first generation of Venn-networks) a layered processing structure that includes competition, not only between processing units (i.e. cortical columns) but also, competition between layers.
The ODM Project aims at developing a new paradigm (and applications) of computational intelligence that is entirely based on Philosophy. The Objective Dialectical Method (ODM) is mainly inspired in the Materialistic Method of philosophical inquiry of the reality and in the dynamics of its contradictions. Initial results of ODM, tackling real world problems, are already very encouraging.
WPA is a scientific research project that aims at developing a novel function approximation algorithm inspired on wolf-pack tactics. The algorithm is intelligent, that is, it is able to learn by its own functioning.
Because of its two-layeredness, including symbolic and distributed processing layers, it is also able to self-control (i) sub-task distributions among agents (i.e. wolfs of the pack) as well as (ii) agreement on egoistic and altruistic behaviors.
Wolf-pack algorithms devised in this project are thought to present an interesting trade-off between agents intelligence and communication abilities, both, highly expensive in a real-world approximation tool.
mEvoSys is a research project that aims at studying and developing innovative systems that incorporate multiple simultaneous evolutionary processes of different nature, e.g. symbolic or subsimbolic. We plan to put together aspects of multiagent systems (i.e. memetic abilities such as communication and coordination) and aspects of conventional evolutionary systems (i.e. genetic abilities such as crossover and mutation). Theoretic aspects to be investigated are: (i) communication; (II) social behaviors; (III) semantics of individual actions; (IV) structural elements of environment and society; (v) how genetics may better influence memetics; (vi) how memetics may influence genetics; (vii) emergency of individual behaviors; and (viii) emergency of social behaviors. Pragmatical aspects to be investigated are how to tackle: (i) surroundings; (II) objectives and goals; (III) actions of the agents; (IV) representation of structural elements of the environment and the society; and, (v) how to represent social/individual behaviors. The expected results aim at applications in Sociology as well as in Economics.