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.

Root ferric chelate reductase is regulated by iron and copper in straberry plants

In the present experiment, we studied the interaction between copper (Cu) and iron (Fe) in strawberry plants grown in nutrient solutions containing different concentrations of Fe. Plants grown in the absence of iron (Fe0) had the characteristic symptoms of Fe deficiency, with smaller chlorotic leaves, less biomass, acidification of the nutrient solution, and roots that were smaller and less ramified, while no symptoms of Fe deficiency were observed in plants grown with Fe. A greater amount of Cu was found in roots of chlorotic plants than in those grown with Fe, while plants grown with 20M of Fe (Fe20) in the nutrient solution had a greater amount of Fe compared with plants from the other treatments. Chlorotic plants (Fe0) and plants grown with the greatest level of Fe (Fe20) had a greater root ferric chelate reductase (FC-R; EC 1.16.1.17) activity compared with the other treatments with 5 or 10M Fe in the nutrient solution. The same pattern was obtained for relative FC-R mRNA concentration and for the sum of Fe and Cu contents in shoots (leaves plus crowns). The DNA obtained from amplification of the FC-R mRNA was cloned and several of the inserts analysed by single strand confirmation polymorphism (SSCP). Although there were different SSCP patterns in the Fe20 treatment, all the inserts that were sequenced were very similar, excluding the hypothesis of more than one FC-R mRNA species being present. The results suggest that Cu as well as Fe is involved in FC-R expression and activity, although the mechanism involved in this regulation is unknown so far. Both small contents of Fe and Cu in plants led to an over-expression of the FC-R gene and enhanced FC-R activity in strawberry roots.