The project deals with artificial neural networks. After designing and debugging the test data set and the training sample set, we created a multilayer perceptron network in the Neural NetworkToolbox (NNT) of Matlab. When creating networks, we used different training algorithms and algorithms improving the generalization of the network. When creating a radial basis network, we did not use the NNT, but a specific source code in Matlab was written. Functionality of neural networks was tested on simple training and testing patterns. Realistic training data were obtained by the simulation of twelve monoconic antennas operating in the frequency range from 2 to 6 GHz. Antennas were located inside a mathematical model of Octavia II. Using CST simulations, electromagnetic fields in a car were obtained. Trained networks are described by regressive characteristics andthe mean square error of training. Algorithms improving generalization are applied on the created and trained networks. The performance of individual networks is mutually compared.

This study proposes novel optimization methodologies for the optimization of problems that reveal non-physical step discontinuities. More specifically, it is proposed to use gradient-only techniques that do not use any zeroth order information at all for step discontinuous problems. A step discontinuous problem of note is the shape optimization problem in the presence of remeshing strategies, since changes in mesh topologies may - and normally do - introduce non-physical step discontinuities. These discontinuities may in turn manifest themselves as non-physical local minima in which optimization algorithms may become trapped. Conventional optimization approaches for step discontinuous problems include evolutionary strategies, and design of experiment (DoE) techniques. These conventional approaches typically rely on the exclusive use of zeroth order information to overcome the discontinuities, but are characterized by two important shortcomings: Firstly, the computational demands of zero order methods may be very high, since many function values are in general required. Secondly, the use of zero order information only does not necessarily guarantee that the algorithms will not terminate in highly unfit local minima. In contrast, the methodologies proposed herein use only first order information, rather than only zeroth order information. The motivation for this approach is that associated gradient information in the presence of remeshing remains accurately and uniquely computable, notwithstanding the presence of discontinuities. From a computational effort point of view, a gradient-only approach is of course comparable to conventional gradient based techniques. In addition, the step discontinuities do not manifest themselves as local minima. / Thesis (PhD)--University of Pretoria, 2010. / Mechanical and Aeronautical Engineering / unrestricted

In the 21st century, information is the new currency. With the omnipresence of devices connected to the internet, humanity can instantly avail any information. However, there are certain are cybercrime groups which steal the information. An Intrusion Detection System (IDS) monitors a network for suspicious activities and alerts its owner about an undesired intrusion. These commercial IDS’es react after detecting intrusion attempts. With the cyber attacks becoming increasingly complex, it is expensive to wait for the attacks to happen and respond later. It is crucial for network owners to employ IDS’es that preemptively differentiate a harmless data request from a malicious one. Machine Learning (ML) can solve this problem by recognizing patterns in internet traffic to predict the behaviour of network users. This project studies how effectively Radial Basis Function Neural Network (RBFN) with Deep Learning Architecture can impact intrusion detection. On the basis of the existing framework, it asks how well can an RBFN predict malicious intrusive attempts, especially when compared to contemporary detection practices.Here, an RBFN is a multi-layered neural network model that uses a radial basis function to transform input traffic data. Once transformed, it is possible to separate the various traffic data points using a single straight line in extradimensional space. The outcome of the project indicates that the proposed method is severely affected by limitations. E.g. the model needs to be fine tuned over several trials to achieve a desired accuracy. The results of the implementation show that RBFN is accurate at predicting various cyber attacks such as web attacks, infiltrations, brute force, SSH etc, and normal internet behaviour on an average 80% of the time. Other algorithms in identical testbed are more than 90% accurate. Despite the lower accuracy, RBFN model is more than 94% accurate at recording specific kinds of attacks such as Port Scans and BotNet malware. One possible solution is to restrict this model to predict only malware attacks and use different machine learning algorithm for other attacks. / I det 21: a århundradet är information den nya valutan. Med allnärvaro av enheter anslutna till internet har mänskligheten tillgång till information inom ett ögonblick. Det finns dock vissa grupper som använder metoder för att stjäla information för personlig vinst via internet. Ett intrångsdetekteringssystem (IDS) övervakar ett nätverk för misstänkta aktiviteter och varnar dess ägare om ett oönskat intrång skett. Kommersiella IDS reagerar efter detekteringen av ett intrångsförsök. Angreppen blir alltmer komplexa och det kan vara dyrt att vänta på att attackerna ska ske för att reagera senare. Det är avgörande för nätverksägare att använda IDS:er som på ett förebyggande sätt kan skilja på oskadlig dataanvändning från skadlig. Maskininlärning kan lösa detta problem. Den kan analysera all befintliga data om internettrafik, känna igen mönster och förutse användarnas beteende. Detta projekt syftar till att studera hur effektivt Radial Basis Function Neural Networks (RBFN) med Djupinlärnings arkitektur kan påverka intrångsdetektering. Från detta perspektiv ställs frågan hur väl en RBFN kan förutsäga skadliga intrångsförsök, särskilt i jämförelse med befintliga detektionsmetoder.Här är RBFN definierad som en flera-lagers neuralt nätverksmodell som använder en radiell grundfunktion för att omvandla data till linjärt separerbar. Efter en undersökning av modern litteratur och lokalisering av ett namngivet dataset användes kvantitativ forskningsmetodik med prestanda indikatorer för att utvärdera RBFN: s prestanda. En Random Forest Classifier algorithm användes också för jämförelse. Resultaten erhölls efter en serie finjusteringar av parametrar på modellerna. Resultaten visar att RBFN är korrekt när den förutsäger avvikande internetbeteende i genomsnitt 80% av tiden. Andra algoritmer i litteraturen beskrivs som mer än 90% korrekta. Den föreslagna RBFN-modellen är emellertid mycket exakt när man registrerar specifika typer av attacker som Port Scans och BotNet malware. Resultatet av projektet visar att den föreslagna metoden är allvarligt påverkad av begränsningar. T.ex. så behöver modellen finjusteras över flera försök för att uppnå önskad noggrannhet. En möjlig lösning är att begränsa denna modell till att endast förutsäga malware-attacker och använda andra maskininlärnings-algoritmer för andra attacker.

[pt] Esquemas semi-Lagrangeanos usados para a aproximação do princípio da programação dinâmica são baseados em uma discretização temporal reconstruída no espaço de estado. O uso de uma malha estruturada torna essa abordagem inviável para problemas de alta dimensão devido à maldição da dimensionalidade. Nesta tese, apresentamos uma nova abordagem para problemas de controle ótimo de horizonte infinito onde a função valor é calculada usando Funções de Base Radial (RBFs) pelo método de aproximação de mínimos quadrados móveis de Shepard em malhas irregulares. Propomos um novo método para gerar uma malha irregular guiada pela dinâmica e uma rotina de otimizada para selecionar o parâmetro responsável pelo formato nas RBFs. Esta malha ajudará a localizar o problema e aproximar o princípio da programação dinâmica em alta dimensão. As estimativas de erro para a função valor também são fornecidas. Testes numéricos para problemas de alta dimensão mostrarão a eficácia do método proposto. Além do controle ótimo de EDPs clássicas mostramos como o método também pode ser aplicado ao controle de equações não-locais. Também fornecemos um exemplo analisando a convergência numérica de uma equação não-local controlada para o modelo contínuo. / [en] Semi-Lagrangian schemes for the approximation of the dynamic programming principle are based on a time discretization projected on a state-space grid. The use of a structured grid makes this approach not feasible for highdimensional problems due to the curse of dimensionality. In this thesis, we present a new approach for infinite horizon optimal control problems where the value function is computed using Radial Basis Functions (RBF) by the Shepard s moving least squares approximation method on scattered grids. We propose a new method to generate a scattered mesh driven by the dynamics and an optimal routine to select the shape parameter in the RBF. This mesh will help to localize the problem and approximate the dynamic programming principle in high dimension. Error estimates for the value function are also provided. Numerical tests for high dimensional problems will show the effectiveness of the proposed method. In addition to the optimal control of classical PDEs, we show how the method can also be applied to the control of nonlocal equations. We also provide an example analyzing the numerical convergence of a nonlocal controlled equation towards the continuous model.

<p>Математички модели за естимацију тешко мерљивих величина називају<br />се софт сензорима. Процес формирања софт сензора није тривијалан и<br />квалитет естимације тешко мерљиве величине директно зависи од<br />начина формирања. Недостаци постојећих алгоритама за формирање<br />спречавају аутоматску конфигурацију софт сензора. У овом раду су<br />реализовани нови алгоритми који имају за сврху аутоматизацију<br />конфигурације софт сензора. Реализовани алгоритми решавају<br />проблеме проналаска оптималног сета улаза у софт сензор и кашњења<br />сваког од њих као и одабира структуре и начина обуке софт сензора<br />заснованих на вештачким неуронским мрежама са радијално базираним<br />функцијама.</p> / <p>Matematički modeli za estimaciju teško merljivih veličina nazivaju<br />se soft senzorima. Proces formiranja soft senzora nije trivijalan i<br />kvalitet estimacije teško merljive veličine direktno zavisi od<br />načina formiranja. Nedostaci postojećih algoritama za formiranje<br />sprečavaju automatsku konfiguraciju soft senzora. U ovom radu su<br />realizovani novi algoritmi koji imaju za svrhu automatizaciju<br />konfiguracije soft senzora. Realizovani algoritmi rešavaju<br />probleme pronalaska optimalnog seta ulaza u soft senzor i kašnjenja<br />svakog od njih kao i odabira strukture i načina obuke soft senzora<br />zasnovanih na veštačkim neuronskim mrežama sa radijalno baziranim<br />funkcijama.</p> / <p>Mathematical models that are used for estimation of variables that can not be<br />measured in real time are called soft sensors. Creation of soft sensor is a<br />complex process and quality of estimation depends on the way soft sensor is<br />created. Restricted applicability of existing algorithms is preventing automatic<br />configuration of soft sensors. This paper presents new algorithms that are<br />providing automatic configuration of soft sensors. Presented algorithms are<br />capable of determing optimal subset of soft sensor inputs and their time<br />delays, as well as optimal architecture and automatic training of the soft<br />sensors that are based on artificial radial basis function networks.</p>

Se propusieron innovaciones en la predicción espacio y espacio-temporal, a partir de métodos geoestadísticos y de funciones de base radial (RBF), considerando métodos basados en distancias. En este sentido, por medio de las distancias entre las variables explicativas, incorporadas específicamente en la regresión basada en distancias, se propusieron modificaciones en: el método kriging universal y en la interpolación con splines espacial y espacio-temporal usando las RBF. El método basado en la distancia se utiliza en un modelo Geoestadístico para estimar la tendencia y la estructura de covarianza. Esta estrategia aprovecha al máximo la información existente, debido a la relación entre las observaciones, mediante el uso de una descomposición espectral de una distancia seleccionada y las coordenadas principales correspondientes. Para el método propuesto kriging universal basado en distancias (DBUK), se realizó un estudio de simulación que permitió comparar la capacidad predictiva del método tradicional kriging universal con respecto a kriging universal basado en distancias; mientras que en la interpolación con Splines espacial y espacio-temporal, los estudios de simulación permitieron comparar el funcionamiento de las funciones de base radial espaciales y espaciotemporales, considerando en la tendencia las coordenadas principales generadas a partir de las variables explicativas mixtas mediante el uso del método basado en distancias. El método propuesto DBUK muestra, tanto en las simulaciones como en las aplicaciones, ventajas en la reducción del error con respecto al método clásico de krigeado universal. Esta reducción de los errores se asocia a una mejor modelización de la tendencia y a un menor error en el ajuste y modelado del variograma, al considerar las coordenadas principales obtenidas a partir de las variables explicativas mixtas. Entre muchas otras posibles causas, el error es generado por omisión de variables y por considerar formas funcionales incorrectas. El estudio de simulación muestra que el método propuesto DBUK es mejor que el método de krigeado universal tradicional ya que se encontró una notoria reducción del error, asociada a un RMSPE más pequeño, esta reducción en general fue superior al 10%. El método DBUK podrá producir una mejor estimación de la variable regionalizada si el número de coordenadas principales se incrementa. Esto es posible, incluyendo las coordenadas principales más significativas tanto en modelo de tendencia como en el variograma; se presenta una aplicación que ilustra este hecho. Los métodos propuestos de interpolación espacial basada en distancias con RBF (DBSIRBF) e interpolación espacio-temporal basada en distancias con RBF (DBSTIRBF) analizados mediante una estructura de krigeado considerando en la tendencia las coordenadas principales, presentan un buen funcionamiento al trabajar con vecindarios grandes, indicando en general que se tendrá un menor error asociado a un RMSPE más pequeño En diversos estudios, la detección de variabilidad entre zonas es una tarea muy difícil, y por lo cual los métodos propuestos DBUK, DBSIRBF y DBSTIRBF son útiles de acuerdo a los resultados obtenidos en la tesis, ya que aprovechan al máximo la información existente asociada a las variables explicativas. Aunque la correlación de las variables explicativas puede ser baja con respecto a la variable respuesta, el punto clave en los métodos propuestos es la correlación entre las coordenadas principales (construida con las variables explicativas) y la variable respuesta. Los métodos propuestos se aplicaron a datos agronómicos (Concentración de calcio medido a una profundidad de 0-20 cm de Brasil) y climatológicos (Temperaturas medias diarias de la Tierra en Croacia en el año 2008). Los resultados de validación cruzada “leave-one-out” mostraron un buen rendimiento de los predictores propuestos, lo cual indica que se pueden utilizar como métodos alternos y validos a los tradicionales para el modelado de variables correlacionadas espacialmente y espacio-temporalmente, considerando siempre covariables en la remoción de la tendencia. / Space-time geostatistical analysis based on distances and splines with applications. Innovations were proposed in the space and space-time prediction, based on geostatistical methods and radial basis function (RBF), considering distance-based methods. In this sense, through the distances between the explanatory variables, specifically incorporated in the regression based on distances, changes were proposed in: the universal kriging and interpolation with space and space-time splines using RBF. The distance-based method is used in a geostatistical model to estimate the trend and the covariance structure. This strategy takes full advantage of existing information, because of the relationship between the observations, using a spectral decomposition of a selected distance and the corresponding principal coordinates. For the universal kriging method proposed based on distances (DBUK), we performed a simulation study, which allowed to compare the predictive capacity of traditional universal kriging over universal kriging based on distances. The simulation study shows that the proposed method DBUK, is better than the traditional universal kriging method and was found a marked reduction of error associated with a smaller RMSPE, this reduction was generally greater than 10%. Spatial and spatio-temporal spline interpolation in simulation studies possible to compare the performance of space and spatio-temporal radial basis functions, considering the trend in the principal coordinates generated from the mixed explanatory variables using the method based distances. The proposed spatial interpolation methods based on distances with RBF (DBSIRBF) and spatio-temporal interpolation based on distances RBF (DBSTIRBF) analyzed through kriging structure whereas in the trend the principal coordinates, show good performance when working with large neighborhoods, indicating that in general will have less error associated with a smaller RMSPE. The key point in the proposed methods is the correlation between the principal coordinates (constructed with the explanatory variables) and the response variable. The proposed methods were applied to agronomic data (concentration of calcium measured at a depth of 0-20 cm from Brazil) and climatological (average daily temperature of the Earth in Croatia in 2008). The results of cross-validation "leave-one-out" showed a good performance of the proposed predictors, indicating that can be used as alternative methods to traditional and valid for the modeling of spatially correlated variables in space and time, always considering covariates in the removal of the trend.

No description available.

No description available.

Σκοπός της διδακτορικής διατριβής είναι η ανάπτυξη κατάλληλων μεθόδων ψηφιακής επεξεργασίας εικόνας και αναγνώρισης προτύπων με τις οποίες θα προσδιορίζονται βιομετρικές και διαγνωστικές παράμετροι μέσω της αλληλεπίδρασης φωτονίων στο ορατό και υπέρυθρο φάσμα. Πιο συγκεκριμένα επιλύεται ένα αντίστροφο πρόβλημα σκέδασης ΗΜ ακτινοβολίας από ένα ανθρώπινο, υγιές και απαραμόρφωτο ερυθρό αιμοσφαίριο. Παρουσιάζονται μέθοδοι εκτίμησης και αναγνώρισης των γεωμετρικών χαρακτηριστικών απαραμόρφωτων υγιών ερυθρών αιμοσφαιρίων με χρήση εικόνων που προσομοιώνουν φαινόμενα σκέδασης ηλεκτρομαγνητικής ακτινοβολίας που διέρχεται από προσανατολισμένα ερυθρά αιμοσφαίρια. Η διαδικασία της ανάκτησης της πληροφορίας περιλαμβάνει, εξαγωγή χαρακτηριστικών με χρήση δισδιάστατων μετασχηματισμών, κανονικοποίηση των χαρακτηριστικών και την χρήση νευρωνικών δικτύων για την εκτίμηση των γεωμετρικών ιδιοτήτων του ερυθροκυττάρου. Παράλληλα σχεδιάστηκε και αξιολογήθηκε σύστημα αναγνώρισης των γεωμετρικών χαρακτηριστικών των ερυθρών αιμοσφαιρίων. Οι εικόνες σκέδασης δημιουργήθηκαν προσομοιώνοντας το πρόβλημα εμπρόσθιας σκέδασης ενός επίπεδου ηλεκτρομαγνητικού (ΗΜ) κύματος, χρησιμοποιώντας την μέθοδο των συνοριακών στοιχείων, λαμβάνοντας υπόψη τόσο την αξονοσυμμετρική γεωμετρία του ερυθροκυττάρου όσο και τις μη αξονοσυμμετρικές οριακές συνθήκες του προβλήματος. Η επίλυση του εν λόγω προβλήματος πραγματοποιήθηκε στα 632.8 nm και εν συνεχεία επεκτάθηκε σε 12 διακριτά ίσου βήματος μήκη κύματος από 432.8 nm έως 1032.8 nm. Επίσης, προτάθηκε μία νέα πειραματική διάταξη για την απόκτηση πολλαπλών εικόνων σκέδασης και την εκτίμηση των γεωμετρικών χαρακτηριστικών των ερυθρών αιμοσφαιρίων, αποτελούμενη από μία πολυχρωματική πηγή φωτός (Led) και πολλαπλά χρωματικά φίλτρα. Επίσης κατασκευάστηκε μέθοδος επίλυσης του σημαντικού προβλήματος εύρεσης της περιεκτικότητας του διαλύματος σε ερυθρά αιμοσφαίρια διαφορετικών μεγεθών στην περίπτωση απόκτησης πολλαπλών εικόνων σκέδασης από διαφορετικές φωτοδιόδους και πολλαπλά χρωματικά φίλτρα. Στα πειράματα αξιολόγησης της μεθόδου που προτείνεται με εικόνες προσομοίωσης δείχνεται ότι είναι ικανή η εύρεση της αναλογίας των ερυθρών αιμοσφαιρίων με πολύ μεγάλη ακρίβεια ακόμα και στη περίπτωση όπου στις εικόνες έχει προστεθεί λευκός κανονικός θόρυβος. Η βασική μεθοδολογία που παρουσιάζεται στην παρούσα δια-τριβή μπορεί να χρησιμοποιηθεί για την αναγνώριση παθολογικών αιμοσφαιρίων ή να χρησιμοποιηθεί στην αναγνώριση μικροσωματιδίων σε υγρά ή αέρια. / The aim of this PhD thesis is the development of digital image processing and pattern recognition methods to estimate biometric and diagnostic parameters using scattering phenomena in the visible and infrared spectrum. More concretely, several reverse scattering problems of EM radiation from a human, healthy and undistorted Red Blood Cell (RBC) is solved. Methods of estimation and recognition of geometrical characteristics of healthy and undistorted RBCs using simulating images are presented. The information retrieval process includes, features extraction using two-dimensional integral transforms, features normalization, and Neural Networks for estimation of three major RBC geometrical proper-ties. Using the same features set, a recognition system of the geometric characteristics of RBCs was developed and evaluated. The scattering images were created simulating the forward scattering problem of a plane electromagnetic wave using the Boundary Element Method, taking into account both axisymmetric geometry of the scatterer and the non-axisymmetric boundary conditions of the problem. Initially, the problem is solved at 632.8 nm and consequently the same problem was solved at 12 different wavelengths, from 432.8 to 1032.8 nm equally spaced. Also, a new device for acquisition of scattering images from RBCs-flow, consisting of a multi-color light source (Led) was proposed, for RBC size estimation and recognition. Finally, a system for the estimation of different RBCs concentration was developed when scattering images acquired using multiple scattering images acquired from multiple Leds and color filters. The system was evaluated using additive white regular noise.

Face recognition is a field of biometrics that deals with identification of subjects based on features present in the images of their faces. The factors that make face recognition popular and favorite as compared to other biometric methods are easier operation and ability to identify subjects without their knowledge. With these features, face recognition has become an integral part of the present day security systems, targeting a smart and secure world. There are various factors that de ne the performance of a face recognition system. The most important among them are recognition accuracy of algorithm used and time taken for recognition. Recognition accuracy of the face recognition algorithm gets affected by changes in pose, facial expression and illumination along with occlusions in the images. There have been a number of algorithms proposed to enable recognition under these ambient changes. However, it has been hard to and a single algorithm that can efficiently recognize faces in all the above mentioned conditions. Moreover, achieving real time performance for most of the complex face recognition algorithms on embedded platforms has been a challenge. Real-time performance is highly preferred in critical applications such as identification of crime suspects in public. As available software solutions for FR have significantly large latency in recognizing individuals, they are not suitable for such critical real-time applications. This thesis focuses on real-time aspect of FR, where acceleration of the algorithms is achieved by means of parallel hardware architectures. The major contributions of this work are as follows. We target to design a face recognition system that can identify at most 30 faces in each frame of video at 15 frames per second, which amounts to 450 recognitions per second. In addition, we target to achieve good recognition accuracy along with scalability in terms of database size and input image resolutions. To design a system with these specifications, as a first step, we explore algorithms in literature and come up with a hybrid face recognition algorithm. This hybrid algorithm shows good recognition accuracy on face images with changes in illumination, pose and expressions, and also with occlusions. In addition the computations in the algorithm are modular in nature which are suitable for real-time realizations through parallel processing. The face recognition system consists of a face detection module to detect faces in the input image, which is followed by a face recognition module to identify the detected faces. There are well established algorithms and architectures for face detection in literature which can perform detection at 15 frames per second on video frames. Detected faces of different sizes need to be scaled to the size specified by the face recognition module. To meet the real-time constraints, we propose a hardware architecture for real-time bi-cubic convolution interpolation with dynamic scaling factors. To recognize the resized faces in real-time, a scalable parallel pipelined architecture is designed for the hybrid algorithm which can perform 450 recognitions per second on a database containing grayscale images of at most 450 classes on Virtex 6 FPGA. To provide flexibility and programmability, we extend this design to REDEFINE, a multi-core massively parallel reconfigurable architecture. In this design, we come up with FR specific programmable cores termed Scalable Unit for Region Evaluation (SURE) capable of performing modular computations in the hybrid face recognition algorithm. We replicate SUREs in each tile of REDEFINE to construct a face recognition module termed REDEFINE for Face Recognition using SURE Homogeneous Cores (REFRESH). There is a need to learn new unseen faces on-line in practical face recognition systems. Considering this, for real-time on-line learning of unseen face images, we design tiny processors termed VOP, Processor for Vector Operations. VOPs function as coprocessors to process elements under each tile of REDEFINE to accelerate micro vector operations appearing in the synaptic weight computations. We also explore deep neural networks which operate similar to the processing in human brain and capable of working on very large face databases. We explore the field of Random matrix theory to come up with a solution for synaptic weight initialization in deep neural networks for better classification . In addition, we perform design space exploration of hardware architecture for deep convolution networks and conclude with directions for future work.