Design of neuro-fuzzy models by evolutionary and gradient-based algorithms

All systems found in nature exhibit, with different degrees, a nonlinear behavior. To emulate this behavior, classical systems identification techniques use, typically, linear models, for mathematical simplicity. Models inspired by biological principles (artificial neural networks) and linguistically motivated (fuzzy systems), due to their universal approximation property, are becoming alternatives to classical mathematical models. In systems identification, the design of this type of models is an iterative process, requiring, among other steps, the need to identify the model structure, as well as the estimation of the model parameters. This thesis addresses the applicability of gradient-basis algorithms for the parameter estimation phase, and the use of evolutionary algorithms for model structure selection, for the design of neuro-fuzzy systems, i.e., models that offer the transparency property found in fuzzy systems, but use, for their design, algorithms introduced in the context of neural networks. A new methodology, based on the minimization of the integral of the error, and exploiting the parameter separability property typically found in neuro-fuzzy systems, is proposed for parameter estimation. A recent evolutionary technique (bacterial algorithms), based on the natural phenomenon of microbial evolution, is combined with genetic programming, and the resulting algorithm, bacterial programming, advocated for structure determination. Different versions of this evolutionary technique are combined with gradient-based algorithms, solving problems found in fuzzy and neuro-fuzzy design, namely incorporation of a-priori knowledge, gradient algorithms initialization and model complexity reduction.

Soft-computing techniques applied to artificial tissue temperature estimation

Tese dout., Engenharia electrónica e computação - Processamento de sinal, Universidade do Algarve, 2008

Cape Verde coastal ecosystem: a study of community structure, trophic interactions and exploitation pattern

Tese de dout., Ciências e Tecnologia das Pescas, Faculdade de Ciências do Mar e do Ambiente, Universidade do Algarve, 2005

Spatial and temporal variation of commercially important bivalve species in the Algarve coast, Portugal

Dissertação de mestrado, Biologia Marinha, Faculdade de Ciências e Tecnologia, Universidade do Algarve, 2015

Phosphoglucose isomerase variability of Cerastoderma glaucum as a model for testing the influence of environmental conditions and dispersal patterns through quantitative ecology approaches

Extreme conditions of coastal lagoons could directly modify the genetic patterns of species. The aim of this work was to investigate the influence of environmental conditions and small scale dispersal patterns on the phosphoglucose isomerase (PGI*) genetic variability of Cerastoderma glaucum from the Mar Menor coastal lagoon. For this purpose, 284 cockles were collected around the perimeter of the lagoon. Vertical polyacrylamide gel electrophoresis was used to scan for PGI* polymorphisms, giving a total of seven alleles. The spatial genetic distribution of the PGI* variability, which seems to be marked by the main circulation in the lagoon, discriminates four hydrological basins. In the central basin, a gradient of allelic composition reflects the circulation forced by the dominant winds and the main channel communicated to the open sea. This result is well supported by the salinity GAM model that defines this gradient. The other three basins are defined by the distribution of fine sand in a more complex model that tries to explain the isolation of the three sites localized inside these basins. The southern, western and northern basins show the lowest degree of interconnection and are considered the most confined areas of the Mar Menor lagoon. This situation agrees with the confinement theory for benthic assemblages in the lagoon. The greater degree of differentiation seen in the Isla del Ciervo population is probably due to recent human intervention on the nearby Marchamalo channel, which has been drained in recent years thus altering the influence of the Mediterranean Sea on the southern basin.

Environmental influence on commercial fishery landings of small pelagic fish in Portugal

Small pelagic fishes are particularly abundant in areas with high environmental variability (zones of coastal upwelling and areas of tidal mixing and river discharge), and because of this, their abundance suffers large inter-annual and inter-decadal fluctuations. In Portugal, the most important species in terms of landings are European sardine, Atlantic horse mackerel and Atlantic chub mackerel. Small pelagic fish landings account for 62.8 % of the total fish biomass and represent 32.7 % of the economical value of all catches. We have investigated trends in landings of these small pelagic fishes and detected the effects of environmental factors in this fishery. In order to explain the variability of landings of small pelagic fishes, we have used official landings (1965-2012) for trawling and purse seine fisheries and applied generalized linear models, using the North Atlantic Oscillation index (NAO) (annual and winter NAO index), sea surface temperature (SST), wind data (strength and North-South and East-West wind components) and rainfall, as explanatory variables. Regression analysis was used to describe the relationship between landings and SST. The models explained between 50.16 and 51.07 % of the variability of the LPUE, with the most important factors being winter NAO index, SST and wind strength. The LPUE of European sardine and Atlantic horse mackerel was negatively correlated with SST, and LPUE of Atlantic chub mackerel was positively correlated with SST. The use of landings of three important species of small pelagic fishes allowed the detection of variations in landings associated with changes in sea water temperature and NAO index.

Small area estimation using linear mixed models with heterogeneous covariance structures of random effects

The problem of Small Area Estimation is about how to produce reliable estimates of domain characteristics when the sample sizes within the domain is very small ou even zero.