A hybrid training method for B-spline neural networks

Current and past research has brought up new views related to the optimization of neural networks. For a fixed structure, second order methods are seen as the most promising. From previous works we have shown how second order methods are of easy applicability to a neural network. Namely, we have proved how the Levenberg-Marquard possesses not only better convergence but how it can assure the convergence to a local minima. However, as any gradient-based method, the results obtained depend on the startup point. In this work, a reformulated Evolutionary algorithm - the Bacterial Programming for Levenberg-Marquardt is proposed, as an heuristic which can be used to determine the most suitable starting points, therefore achieving, in most cases, the global optimum.

Environmental impacts on soil and groundwater at airports: origin, contaminants of concern and environmental risks

Environmental impacts of airports are similar to those of many industries, though their operations expand over a very large area. Most international impact assessment studies and environmental management programmes have been giving less focus on the impacts to soil and groundwater than desirable. This may be the result of the large attention given to air and noise pollution, relegating other environmental descriptors to a second role, even when the first are comparatively less relevant. One reason that contributes to such ‘‘biased’’ evaluation is the lack of systematic information about impacts to soil and groundwater from airport activities, something the present study intends to help correct. Results presented here include the review of over seven hundred documents and online databases, with the objective of obtaining the following information to support environmental studies: (i) which operations are responsible for chemical releases?; (ii) where are these releases located?; (iii) which contaminants of concern are released?; (iv) what are the associated environmental risks? Results showed that the main impacts occur as a result of fuel storage, stormwater runoff and drainage systems, fuel hydrant systems, fuel transport and refuelling, atmospheric deposition, rescue and fire fighting training areas, winter operations, electrical substations, storage of chemical products by airport owners or tenants, and maintenance of green areas. A new method for ranking environmental risks of organic substances, based on chemical properties, is proposed and applied. Results show that the contaminants with the highest risks are the perfluorochemicals, benzene, trichloroethylene and CCl4.

Completely supervised training algorithms for B-spline neural networks and neuro-fuzzy systems

Complete supervised training algorithms for B-spline neural networks and fuzzy rule-based systems are discussed. By interducing the relationship between B-spline neural networks and certain types of fuzzy models, training algorithms developed initially for neural networks can be adapted by fuzzy systems.

Training multilayer perceptrons - an optimization perspective

Multilayer perceptrons (MLPs) (1) are the most common artificial neural networks employed in a large field of applications. In control and signal processing applications, MLPs are mainly used as nonlinear mapping approximators. The most common training algorithm used with MLPs is the error back-propagation (BP) alg. (1).

Training multilayer perceptrons for control systems applications - a comparison of different approaches

Multilayer perceptrons (MLPs) (1) are the most common artificial neural networks employed in a large field of applications. In control and signal processing applications, MLPs are mainly used as nonlinear mapping approximators. The most common training algorithm used with MLPs is the error back-propagation (BP) alg. (1).

Supervised training approach using spiking neural networks

One of the basic aspects of some neural networks is their attempt to approximate as much as possible their biological counterparts. The goal is to achieve a simple and robust network, easy to understand and able of simulating the human brain at a computational level. Recently a third generation of neural networks (NN) [1], called Spiking Neural Networks(SNN) was appeared. This new kind of networks use the time of a electrical pulse, or spike, to encode the information. In the first and second generation of NN analog values are used in the communication between neurons.

Training neuro-fuzzy models using evolution based algorithms

The normal design process for neural networks or fuzzy systems involve two different phases: the determination of the best topology, which can be seen as a system identification problem, and the determination of its parameters, which can be envisaged as a parameter estimation problem. This latter issue, the determination of the model parameters (linear weights and interior knots) is the simplest task and is usually solved using gradient or hybrid schemes. The former issue, the topology determination, is an extremely complex task, especially if dealing with real-world problems.

Bio-synthesis of nanosized semiconductors using mine wastes as material sources

This work describes the synthesis of nanosized metal sulfides and respective SiO2 and/or TiO2 composites in high yield via a straightforward process, under ambient conditions (temperature and pressure), by adding to aqueous metals a nutrient solution containing biologically generated sulfide from sulfate-reducing bacteria (SRB). The nanoparticles‘ (NPs) morphological properties were shown not to be markedly altered by the SRB growth media composition neither by the presence of bacterial cells. We further extended the work carried out, using the effluent of a bioremediation system previously established. The process results in the synthesis of added value products obtained from metal rich effluents, such as Acid Mine Drainage (AMD), when associated with the bioremediation process. Precipitation of metals using sulfide allows for the possibility of selective recovery, as different metal sulfides possess different solubilities. We have evaluated the selective precipitation of CuS, ZnS and FeS as nanosized metal sulfides. Again, we have also tested the precipitation of these metal sulfides in the presence of support structures, such as SiO2. Studies were carried out using both artificial and real solutions in a continuous bioremediation system. We found that this method allowed for a highly selective precipitation of copper and a lower selectivity in the precipitation of zinc and iron, though all metals were efficiently removed (>93% removal). This research has also demonstrated the potential of ZnS-TiO2 nanocomposites as catalysts in the photodegradation of organic pollutants using the cationic dye, Safranin-T, as a model contaminant. The influence of the catalyst amount, initial pH and dye concentration were also evaluated. Finally, the efficiency of the precipitates as catalysts in sunlight mediated photodegradation was investigated, using different volumes of dye-contaminated water (150 mL and 10 L). This work demonstrates that all tested composites have the potential to be used as photocatalysts for the degradation of Safranin-T.