Artificial Neural Networks (ANNs) are powerful predictors, however, they essentially function like 'black boxes' because they lack explanatory power. Various algorithms have been developed to examine input influences and interactions thus enhancing understanding of the function being modeled. The study of facial attractiveness is one domain that could potentially benefit from ANN models. The literature shows that the relationship between attractiveness and facial attributes is complex and not yet fully understood. In this project, a feed-forward ANN was trained with backpropagation to 0.86 classification using 8-fold cross validation. The dataset consisted of 88 female facial images, each containing 17 geofacial measurements, a random noise variable, and a rating. Input 'clamping' and the Connection Weight Approach (Olden & Jackson, 2002), were implemented and the results were examined in terms of the facial attractiveness domain. In general, the results suggest that more feminized and asymmetrical features enhance facial attractiveness.

L’utilisation de colorants synthétiques se développe depuis de nombreuses années dans des différentes industries. Les procèdes les plus couramment utilises pour le traitement des eaux usées colorées sont les traitements biologiques mais ils ont leurs limites. Les techniques d’oxydation chimique traditionnelle quant a elles conduisent à la coupure de la molécule au niveau du chromophore ne peuvent pas minéraliser totalement des colorants alors que la dépollution complète de ces effluents l’exigerait. Le procède photo-fenton homogène est une technique de traitement qui peut dégrader les effluents colores efficacement. Mais il y a quelques inconvénients majeurs qui limitent l’application industrielle de cette technologie. L’objectif de ce travail de recherche était d’étudier la décoloration et la dégradation des solutions des colorants: acide orange 7 acide orange 8 acide rouge i4 acide rouge 73 et acide bleu 74 par le procède d’oxydation avancée: photo-Fenton heterogene. Dans ce procède la zeolite fe-zsm5 a été utilisée comme un catalyseur heterogene. L’application de ce système nous a permis d une part de diminuer la quantité de boue formée au cours du traitement ainsi que la consommation d’énergie électrique engendrée par l’utilisation d’UV estimée à l aide de la méthode figures-of-merit et d’autre par d’étendre l’application du procède type photo-fenton aux ph plus élevés. La modélisation des procèdes étudies a été faite par deux méthodes à la méthodologie de surface de réponse (RSM) et les réseaux neuronaux artificiels (ANNS) - afin d optimiser la performance de système et également d'évaluer les effets simples et combines des différentes variables sur l’efficacité du traitement. / Large amounts of dyes are annually produced and applied in different industries. The biological methods are widely used for treatment of coloured ef fluents, but they have some limitations. Traditional chemical oxidations that destroy the chromophore of the molecule could not also result the complete mineralisation of the dyes. Homogeneous photo-fenton is a promising technique for treatment of the effluents but there are still some drawbacks that limit the industrial applications of this method. The aim of this work was to study the decolourisation and the degradation of coloured solution containing acid orange 7 acid orange 8 acid red I4 acid red 73 or acid blue 74 by an advanced oxidation process: heterogeneous Photo-fenton. In this study zeolite fe-zsm5 was used as a heterogeneous catalyst. Application of this system not only allowed us to diminish the quantity of sludge formed during the process but also reduced the consummation of electrical energy process keeps its high efficiency even at neutral phs. The modelling of the process was done by two methods - response surface methodology (RSM) and artificial Neural networks (ANNS) in order to optimise the performance of the system and to evaluate the simple and the combined effects of different variables on the process efficiency.

O biodiesel vem se consolidando como combustível alternativo devido às suas vantagens e importância econômica. Conforme seu percentual no diesel comercial cresce, o monitoramento da qualidade se torna cada vez mais importante para uma comercialização segura. No entanto, a estabilidade oxidativa do biodiesel é inferior à do diesel fóssil, assim antioxidantes sintéticos, tais como BHA, BHT, PG e TBHQ, são adicionados para prevenir a degradação do mesmo, evitando danos ao sistema de combustão automotivo. Os métodos para avaliar a estabilidade oxidativa e a quantidade de antioxidantes são em geral demorados, requerem preparação amostral ou equipamentos de alto custo. Com o objetivo de contornar tais problemas, neste trabalho foi aplicada uma metodologia para análise direta de antioxidantes em amostras de biodiesel por meio de voltametria de pulso diferencial. Conjuntamente, foi estudada a viabilidade da associação dessa metodologia com técnicas quimiométricas para a determinação simultânea desses antioxidantes em misturas de antioxidantes em biodiesel. Para permitir uma análise direta, sem extração e pré-concentração dos analitos, o biodiesel foi diluído em meio etanólico. Medidas de voltametria de pulso diferencial para cada antioxidante individualmente mostraram relação linear entre as concentrações dos antioxidantes e a corrente de oxidação. Os limites de detecção individuais obtidos foram de 20,5 mg L-1 para BHA, 32,4 mg L-1 para BHT, 35,5 mg L-1 para PG e 26,5 mg L-1 para TBHQ. A modelagem quimiométrica foi aplicada por meio das ferramentas Mínimos Quadrados Clássico (CLS), Mínimos Quadrados Parciais (PLS), Redes Neuronais Artificiais (ANN), Componentes Principais-Redes Neuronais Artificiais (PC-ANN) e Árvore de Decisão-Redes Neuronais Artificiais (DT-ANN). O modelo construído por PLS se mostrou melhor quando comparado à modelagem por CLS. O modelo construído por ANN’s sem seleção de dados de entrada apresentou erros semelhantes ao PLS. Quando a redução da quantidade de dados de entrada foi aplicada em conjunto com ANN’s através de Análise por Componentes Principais (PCA) e DT, a aplicação de PCA levou a aumento de 10,2% no erro de predição, enquanto que na seleção por DT os erros de predição foram reduzidos em 8,5%. A determinação simultânea dos quatro compostos pelo modelo DT-ANN apresentou precisão satisfatória, com recuperação de 98% para BHA, 97% para BHT, 103% para PG e 100% para TBHQ, o que indica que a técnica analítica e a modelagem quimiométrica são viáveis e promissoras para aplicação no controle de qualidade do biodiesel, bem como em análises de monitoramento nas plantas industriais. / Biodiesel is becoming established as an alternative fuel because its advantages and economic importance. As the levels of biodiesel in commercial diesel grows, quality monitoring becomes increasingly important for safe marketing. However, the oxidative stability of biodiesel is smaller than that of fossil diesel and synthetic antioxidants such as BHA, BHT, PG and TBHQ are added to it in order to prevent its degradation, avoiding damage to the automotive combustion system. Methods for evaluating oxidative stability and the amount of antioxidants are usually time-consuming, require sample preparation or expensive equipment. In order to overcome such problems, in this work a methodology was applied for direct analysis of antioxidants in biodiesel samples by means of differential pulse voltammetry. The viability of associating this methodology with chemometric techniques was studied for the simultaneous determination of these antioxidants in biodiesel. To allow a direct analysis, without extraction and preconcentration of analytes, biodiesel was diluted in ethanolic medium. Differential pulse voltammetric measurements for each antioxidant individually showed a linear relationship between antioxidant concentrations and oxidation current. The individual detection limits were 20,5 mg L-1 for BHA, 32,4 mg L-1 for BHT, 35,5 mg L-1 for PG and 26,5 mg L-1 for TBHQ. The chemometric modeling was applied using the Classical Least Squares (CLS), Partial Least Squares (PLS), Artificial Neural Networks (ANN), Principal Component-Artificial Neural Networks (PC-ANN) and Decision Tree-Artificial Neural Networks (DT-ANN) techniques. The model constructed by PLS was better than that obtained with CLS. The model constructed by ANN’s without input selection presented similar deviations in comparison to PLS. When amount of input data reduction was applied together with ANNs through Principal Component Analysis (PCA) and DT, the PCA application led to a 10.2% increase in prediction error, whereas in the selection by DT prediction errors were reduced by 8,5%. The simultaneous determination of the four compounds by the DT-ANN model presented satisfactory accuracy with 98% recovery for BHA, 97% for BHT, 103% for PG and 100% for TBHQ, indicating that the analytical technique and the chemometric modeling are feasible and promising for application in biodiesel quality control, as well as in monitoring analyzes in the industrial plants.

This research provides a robust solution based on artificial intelligence (AI) techniques to overcome the challenges in Software Defined Networks (SDNs) that can jeopardise the overall performance of the network. The proposed approach, presented in the form of an intelligent agent appended to the SDN network, comprises of a new hybrid intelligent mechanism that optimises the performance of SDN based on heuristic optimisation methods under an Artificial Neural Network (ANN) paradigm. Evolutionary optimisation techniques, including Particle Swarm Optimisation (PSO) and Genetic Algorithms (GAs) are deployed to find the best set of inputs that give the maximum performance of an SDN-based network. The ANN model is trained and applied as a predictor of SDN behaviour according to effective traffic parameters. The parameters that were used in this study include round-trip time and throughput, which were obtained from the flow table rules of each switch. A POX controller and OpenFlow switches, which characterise the behaviour of an SDN, have been modelled with three different topologies. Generalisation of the prediction model has been tested with new raw data that were unseen in the training stage. The simulation results show a reasonably good performance of the network in terms of obtaining a Mean Square Error (MSE) that is less than 10−6 [superscript]. Following the attainment of the predicted ANN model, utilisation with PSO and GA optimisers was conducted to achieve the best performance of the SDN-based network. The PSO approach combined with the predicted SDN model was identified as being comparatively better than the GA approach in terms of their performance indices and computational efficiency. Overall, this research demonstrates that building an intelligent agent will enhance the overall performance of the SDN network. Three different SDN topologies have been implemented to study the impact of the proposed approach with the findings demonstrating a reduction in the packets dropped ratio (PDR) by 28-31%. Moreover, the packets sent to the SDN controller were also reduced by 35-36%, depending on the generated traffic. The developed approach minimised the round-trip time (RTT) by 23% and enhanced the throughput by 10%. Finally, in the event where SDN controller fails, the optimised intelligent agent can immediately take over and control of the entire network.

My thesis work includes 10 large drawings that explore the narrative qualities of line movement. I use line in these drawings as a recording device to document both the physical activity of the process and the personal experience that the lines represent. Through this work I attempt to develop a language of visual imagery that is autobiographical in nature. A system of personal symbols and recurring forms make up the structure of this visual vocabulary. Line movement and rhythm provides a sense of the dialog.

Last few decades have witnessed the use of artificial neural networks (ANN) in many real-world applications and have offered an attractive paradigm for a broad range of adaptive complex systems. In recent years ANN have enjoyed a great deal of success and have proven useful in wide variety pattern recognition or feature extraction tasks. Examples include optical character recognition, speech recognition and adaptive control to name a few. To keep the pace with its huge demand in diversified application areas, many different kinds of ANN architecture and learning types have been proposed by the researchers to meet varying needs. A novel hybrid learning approach for the training of a feed-forward ANN has been proposed in this thesis. The approach combines evolutionary algorithms with matrix solution methods such as singular value decomposition, Gram-Schmidt etc., to achieve optimum weights for hidden and output layers. The proposed hybrid method is to apply evolutionary algorithm in the first layer and least square method (LS) in the second layer of the ANN. The methodology also finds optimum number of hidden neurons using a hierarchical combination methodology structure for weights and architecture. A learning algorithm has many facets that can make a learning algorithm good for a particular application area. Often there are trade offs between classification accuracy and time complexity, nevertheless, the problem of memory complexity remains. This research explores all the different facets of the proposed new algorithm in terms of classification accuracy, convergence property, generalization ability, time and memory complexity.

<p>This thesis is concerned with the prediction of protein mutations using artificial neural networks. From the biological perspective it is of interest to investigate weather it is possible to find rules of mutation between evolutionary adjacent (or closely related) proteins. Techniques from computer science are used in order to see if it is possible to predict protein mutations i.e. using artificial neural networks. The computer science perspective of this work would be to try optimizing the results from the neural networks. However, the focus of this thesis is primarily on the biological perspective and the performance of the computer science methods are secondary objective i.e. the primary interest is to show the existence of rules for protein mutations.</p><p>The method used in this thesis consists two neural networks. One network is used to predict the actual protein mutations and the other network is used to make a compressed representation of each amino acid. By using a compression network it is possible to make the prediction network much smaller (each amino acid is represented by 3 nodes instead of 22 nodes). The compression network is an auto associative network and the prediction network is a standard feed-forward network. The prediction network predicts a block of amino acids at a time and for comparison a sliding window technique has also been tested.</p><p>It is my belief that the results in this thesis indicate that there exists rules for protein mutations. However, the tests done in this thesis is only performed on a small portion of all proteins. Some protein families tested show really good results while other families are not as good. I believe that extended work using optimized neural networks would improve the predictions further.</p>

<p>Forskning angående styrsystem inom evolutionär robotik fokuserar ofta på vad som går att uträtta med självlärda styrsystem, men inte vad beteendebaserade styrsystem klarar av. I detta projekt utförs systematiska tester för att jämföra beteendebaserade och självlärda styrsystem inom evolutionär robotik. Benämningen beteendebaserade styrsystem används för styrsystem som består av flera underliggande moduler, där robotens övergrippande beteende är uppdelat i flera moduler och där respektive modul har ansvar för ett specifikt beteende eller funktion. Självlärt styrsystem referar i detta sammanhang till styrsystem som inte explicit består av moduler som har åstakommits eller bestämts i förväg.</p><p>Från resultaten framgår det att både självlärda och beteendebaserade styrsystem klarar av att lösa problemen de ställs inför. För beteendebaserade styrsystem krävs dock större ansvar från experimentatorn, vars inflytande ökar på grund av att det är flera moduler som skall skapas och koordineras. De beteendebaserade styrsystemen i projektet använder en beslutsenhet för att hantera när och vilken modul som ska aktiveras och tar bort detta ansvar från experimentatorn. Beslutsenheten ger styrsystemet en mer smidig övergång mellan de olika modulerna och hittar en lämplig användning av modulerna än utan en beslutsenhet. Från resultatet av projektet framgår det att fler systematiska tester angånde beteendebaserade och självlärda styrsystem behövs för att få en bättre förståelse över när och hur de olika styrsystemen bör användas.</p>

Dissertation--Uppsala, 2004. / Bibliogr. p. 389-403.

Thesis (M.H.S.A.)--University of Michigan, 1982.