The combination of neural estimates in prediction and decision problems

In this dissertation, different ways of combining neural predictive models or neural-based forecasts are discussed. The proposed approaches consider mostly Gaussian radial basis function networks, which can be efficiently identified and estimated through recursive/adaptive methods. Two different ways of combining are explored to get a final estimate – model mixing and model synthesis –, with the aim of obtaining improvements both in terms of efficiency and effectiveness. In the context of model mixing, the usual framework for linearly combining estimates from different models is extended, to deal with the case where the forecast errors from those models are correlated. In the context of model synthesis, and to address the problems raised by heavily nonstationary time series, we propose hybrid dynamic models for more advanced time series forecasting, composed of a dynamic trend regressive model (or, even, a dynamic harmonic regressive model), and a Gaussian radial basis function network. Additionally, using the model mixing procedure, two approaches for decision-making from forecasting models are discussed and compared: either inferring decisions from combined predictive estimates, or combining prescriptive solutions derived from different forecasting models. Finally, the application of some of the models and methods proposed previously is illustrated with two case studies, based on time series from finance and from tourism.

Desenvolvimento de bioblendas poliméricas a partir de gelatina bovina e de tilápia com amido de milho

The natural raw materials acquired special importance beside the mineral raw materials with the need for using alternative sources to oil, because they can be used to produce biopolymers. Gelatin, produced from the denaturation of collagen, and starch, an abundant polysaccharide in various plants, are examples of biopolymers which have several technological applications, especially in films. The objective of this work is to produce polymeric bioblends with gelatin and corn starch using two types of gelatin: commercial bovine gelatin and gelatin produced from mechanically separated flesh of tilapia (Oreochromis niloticus). For the extraction of tilapia gelatin 3 distinct pretreatments, followed by extraction in distilled water under heating were performed. The properties of gelatin extracted were similar to bovine gelatin, and the differences can be explained by the difference in extraction processes and sources. Blends of commercial gelatin and starch were produced in an internal mixer from a Haake torque rheometer, to study the behavior of the gelatin mixture with starch, thus, the same compositions were processed by twin screw extrusion, to define the mixing parameters. Subsequently, the extrusion of blends of tilapia gelatin and corn starch was carried out in the same twin screw extruder. The physico-chemical, rheological and morphological properties of the blends with thermoplastic starch and gelatin were studied. It was found that various properties vary linearly with increasing concentration of the components. The blends produced are immiscible, and among the two gelatins, tilapia gelatin showed a better interfacial adhesion with the corn starch. Regarding the morphology, gelatins formed the dispersed phase in all compositions studied, even in compositions rich in starch. Can be concluded that the procedure for tilapia gelatin extraction is feasible and advantageous, and the increasing in its scale to a reactor of 30 liters is possible, with a satisfactory yield. The bioblends of bovine gelatin/corn starch and tilapia gelatin/corn starch were successfully produced, and the processing conditions were appropriate

Desempenho de juvenis de tilápia-do-nilo alimentados com rações contendo complexo enzimático

Avaliou-se o efeito da inclusão de um complexo enzimático em dietas para tilápias-do-nilo (Oreochromis niloticus) sobre o desempenho, a composição química da carcaça e a qualidade da água. Foram utilizados 200 alevinos revertidos (4,57 ± 1,24 g), distribuídos em delineamento inteiramente casualizado em 20 tanques de 500 litros, com quatro tratamentos e cinco repetições, considerando a unidade experimental uma caixa com dez peixes. Os peixes foram alimentados com dietas contendo 0; 0,033; 0,066 ou 0,099% de complexo enzimático. As dietas foram processadas na forma peletizada e fornecidas quatro vezes ao dia, às 8, 11, 14 e 17 h. Os valores médios de pH, condutividade elétrica, oxigênio dissolvido, temperatura, fósforo total, amônia e nitrato da água de cultivo não foram influenciados pela dieta. A inclusão do complexo enzimático na dieta não afetou o ganho de peso, as taxas de sobrevivência e de crescimento específico, mas influenciou o consumo de ração e a conversão alimentar, cujos valores foram maiores nos peixes alimentados com a dieta com 0,066% de complexo enzimático. Não foram observadas diferenças nos teores de matéria seca, umidade, proteína bruta, matéria mineral, cálcio e fósforo na carcaça dos peixes, no entanto, o teor de extrato etéreo reduziu de forma linear com o aumento do nível de complexo enzimático. A utilização de complexo enzimático (amilase, protease, celulase, lipase, pectinase, xilanase, β-glucanase e fitase) no nível de 0,066% em dietas para juvenis de tilápia-do-nilo piora a conversão alimentar, mas não influencia o desempenho e a composição corporal dos peixes.

Effect of nitrite on larval development of giant river prawn Macrobrachium rosenbergii

The effects of ambient nitrite concentrations on larval development of giant river prawn Macrobrachium rosenbergii were evaluated. The trials were conducted in two phases: phase 1, larvae from stages I through VIII and phase 2, larvae from stage VIII until post-larvae. In both phases larvae were kept in water with nitrite (NO2-N) concentrations of 0, 2, 4, 8 and 16 mg/L. Oxygen consumption was analyzed for larvae in stage II at nitrite concentrations of 0, 4, and 8 mg/L. Survival, weight gain, larval stage index and metamorphosis rate decreased linearly with increasing ambient nitrite concentration. However, there was no significant difference between larvae subjected to 0 and 2 mg/L NO2-N. In phase 1, there was total mortality at 16 mg/L NO2-N, while in phase 2 larval development stopped at stage X in this treatment. The oxygen consumption in stage II increased significantly at NO2-N concentration from 0 to 4 mg/L, but there was no difference between 4 and 8 mg/L NO2-N. In conclusion, increasing ambient nitrite up to 16 mg/L NO2-N delays larval development, reduces larval growth rate and causes mortality, whereas no significant effect occurs for levels below 2 mg/L NO2-N. However, the establishment of a general safe level of nitrite to M rosenbergii hatchery may be difficult due to the great variability in larvae individual sensitivity. (c) 2006 Elsevier B.V. All rights reserved.