Many real-world face recognition tasks are under unconstrained conditions such as off-angle pose variations, illumination variations, facial occlusion, facial expression, etc. In this work, we are focusing on the real-world scenarios where only the periocular region of a face is visible such as in the ISIS case. In Part I of the dissertation, we will showcase the face recognition capability based on the periocular region, which we call the periocular face recognition. We will demonstrate that face matching using the periocular region directly is more robust than the full face in terms of age-tolerant face recognition, expression-tolerant face recognition, pose-tolerant face recognition, as well as contains more cues for determining the gender information of a subject. In this dissertation, we will study direct periocular matching more comprehensively and systematically using both shallow and deep learning methods. Based on this, in Part II and Part III of the dissertation, we will continue to explore an indirect way of carrying out the periocular face recognition: periocular-based full face hallucination, because we want to capitalize on the powerful commercial face matchers and deep learning-based face recognition engines which are all trained on large-scale full face images. The reproducibility and feasibility of re-training for a proprietary facial region, such as the periocular region, is relatively low, due to the nonopen source nature of commercial face matchers as well as the amount of training data and computation power required by the deep learning based models. We will carry out the periocular-based full face hallucination based on two proposed reconstructive dictionary learning methods, including the dimensionally weighted K-SVD (DW-KSVD) dictionary learning approach and its kernel feature space counterpart using Fastfood kernel expansion approximation to reconstruct high-fidelity full face images from the periocular region, as well as two proposed generative deep learning approaches that build upon deep convolutional generative adversarial networks (DCGAN) to generate the full face from the periocular region observations, including the Gang of GANs (GoGAN) method and the discriminant nonlinear many-to-one generative adversarial networks (DNMM-GAN) for applications such as the generative open-set landmark-free frontalization (Golf) for faces and universal face optimization (UFO), which tackles an even broader set of problems than periocular based full face hallucination. Throughout Parts I-III, we will study how to handle challenging realworld scenarios such as unconstrained pose variations, unconstrained illumination conditions, and unconstrained low resolution of the periocular and facial images. Together, we aim to achieve unconstrained periocular face recognition through both direct periocular face matching and indirect periocular-based full face hallucination. In the final Part IV of the dissertation, we will go beyond and explore several new methods in deep learning that are statistically efficient for generalpurpose image recognition. Methods include the local binary convolutional neural networks (LBCNN), the perturbative neural networks (PNN), and the polynomial convolutional neural networks (PolyCNN).

Orientador: Maurício Tatsuei Sakima / Banca: Ary dos Santos Pinto / Banca: Luiz Gonzaga Gandini Junior / Banca: Júlio de Araújo Gurgel / Banca: Arno Locks / Resumo: Este estudo cefalométrico prospectivo foi desenvolvido com o propósito de descrever os resultados de uma das estratégias de correção da mordida profunda. Foram selecionados 19 indivíduos, com faixa etária entre 11 e 15 anos, apresentando más-oclusões de Classe II, Divisão 1, com mordida profunda de no mínimo 4 milímetros. Desses, 9 indivíduos serviram como grupo controle e os outros 10 foram tratados com a mecânica de intrusão da técnica de Ricketts (arco base). Foram colocados implantes metálicos de referência intra-mandibulares, para sobreposições de traçados, em todos os componentes da amostra. Telerradiografias cefalométricas, em norma lateral, para a avaliação do comportamento dos incisivos inferiores, e em 45 graus, para a avaliação dos primeiros pré-molares e primeiros molares inferiores, foram tomadas no início do tratamento e após o nivelamento da curva de Spee do arco inferior para o grupo experimental, e após aproximadamente 6 meses no grupo controle. Os resultados na região de incisivos inferiores indicaram que houve intrusão dos incisivos inferiores e também um deslocamento horizontal para lingual dos três pontos estudados (borda incisal, centro de resistência e ápice radicular). Não houve deslocamento vertical (extrusivo) nos primeiros pré-molares e nos primeiros molares causados pelo tratamento. Os primeiros pré-molares demonstraram uma inclinação para distal com o fulcro próximo ao ápice, apesar de nenhum acessório ter sido colocado nesses dentes. Na região de molares, houve uma inclinação distal da coroa e mesial de raiz, com o fulcro desse movimento próximo ao centro de resistência. / Abstract: The purpose of this prospective study was to avaliate the results of treatment strategie for deep overbite correction. Nineteen Class II, Division 1, with deep overbite individuals (age 11 to 15 years) were selected. Nine cases were used as a control group and the others were trated with the bioprogressive technique (Ricketts) for correction of vertical malocclusion. Metallic implants were used for superimpositions. Lateral cephalometric radiographs were used for evaluation of lower incisors. Forty five degrees cephalometric radiographs were used for evaluation of lower first bicuspids and first molars. These radiographs were taken before and immediately after leveling of lower arch and about 6 months later for the control group. The results showed that the technique produced highly significant incisor intrusion and a lingual movement of three points inverstigated (incisal edge, center of resistence and root apex). There was no vertical displacement (extrusion) on lower first bicuspid and first molar. A distal inclination was observed on lower first bicuspid, despite of any bracket has been fixed on it. Lower first molars crowns showed a distal movement and the root showed a mesial movement, with center of rotation near the fulcrum. / Doutor

Thesis (MTech (Mechanical Engineering))--Cape Peninsula University of Technology, 2017. / In South Africa, squid fishing vessels need to find and then anchor above benthic squid egg beds to effect viable catches. However, waves acting on the vessel produce a dynamic response on the anchor line. These oscillatory motions produce impact forces of the chain striking the seabed. It is hypothesised that this causes damage to the squid egg bed beneath the vessels. Different mooring systems may cause more or less damage and this is what is investigated in this research. The effect of vessel mooring lines impact on the seabed during squid fishing is investigated using a specialised hydrodynamic tool commercial package ANSYS AQWA models. This study analysed the single-point versus the two-point mooring system’s impact on the seabed. The ANSYS AQWA models were developed for both mooring systems under the influence of the wave and current loads using the 14 and 22 m vessels anchored with various chain sizes. The effect of various wave conditions was investigated as well as the analysis of three mooring line configurations. The mooring chain contact pressure on the seabed is investigated beyond what is output from ANSYS AQWA using ABAQUS finite element analysis. The real-world velocity of the mooring chain underwater was obtained using video analysis. The ABAQUS model was built by varying chain sizes at different impact velocities. The impact pressure and force due to this velocity was related to mooring line impact velocity on the seabed in ANSYS AQWA. Results show the maximum impact pressure of 191 MPa when the 20 mm diameter chain impacts the seabed at the velocity of 8 m/s from video analysis. It was found that the mooring chain impact pressure on the seabed increased with an increase in the velocity of impact and chain size. The ANSYS AQWA impact pressure on the seabed was found to be 170.86 MPa at the impact velocity of 6.4 m/s. The two-point mooring system was found to double the seabed mooring chain contact length compared to the single-point mooring system. Both mooring systems showed that the 14 m vessel mooring line causes the least seabed footprint compared to the 22 m vessel.

Semantics is the study of meaning and here we explore it through three major representations: brain, image and text. Researchers in the past have performed various studies to understand the similarities between semantic features across all the three representations. Distributional Semantic (DS) models or word vectors that are trained on text corpora have been widely used to study the convergence of semantic information in the human brain. Moreover, they have been incorporated into various NLP applications such as document categorization, speech to text and machine translation. Due to their widespread adoption by researchers and industry alike, it becomes imperative to test and evaluate the performance of di erent word vectors models. In this thesis, we publish the second iteration of BrainBench: a system designed to evaluate and benchmark word vectors using brain data by incorporating two new Italian brain datasets collected using fMRI and EEG technology. In the second half of the thesis, we explore semantics in Convolutional Neural Network (CNN). CNN is a computational model that is the state of the art technology for object recognition from images. However, these networks are currently considered a black-box and there is an apparent lack of understanding on why various CNN architectures perform better than the other. In this thesis, we also propose a novel method to understand CNNs by studying the semantic representation through its hierarchical layers. The convergence of semantic information in these networks is studied with the help of DS models following similar methodologies used to study semantics in the human brain. Our results provide substantial evidence that Convolutional Neural Networks do learn semantics from the images, and the features learned by the CNNs correlate to the semantics of the object in the image. Our methodology and results could potentially pave the way for improved design and debugging of CNNs. / Graduate

Orientador: Dirceu Barnabé Raveli / Banca: Mauricio Tatsuei Sakima / Banca: Helio Hissashi Terada / Resumo: Este estudo cefalométrico prospectivo foi desenvolvido com o propósito de descrever os resultados de uma das estratégias de intrusão de caninos. Foram selecionados 19 indivíduos, com faixa etária entre 11 e 15 anos, apresentando más-oclusões de Classe II, Divisão 1, com mordida profunda mínima de 4 milímetros. Desses, 9 indivíduos serviram como grupo controle e os outros 10 foram tratados inicialmente com a mecânica de intrusão da técnica de Ricketts (arco base). Foram colocados implantes intra-mandibulares, para sobreposição de traçados, em todos os componentes da amostra. Teleradiografias cefalométricas, em norma lateral, para a avaliação do comportamento dos incisivos inferiores, e em 45 graus, para avaliação dos caninos inferiores, foram tomadas no início do tratamento e após a intrusão dos caninos no arco inferior para o grupo experimental, e após aproximadamente 6 meses no grupo controle. Os resultados na região dos incisivos inferiores indicaram que houve uma leve vestibularização deste dentes, sem provocar extrusão. Os resultados nos caninos inferiores demonstraram que houve intrusão nos três pontos estudados (ponta de cúspide, centro de resistência e ápice radicular) e uma inclinação para distal do centro de resistência e ápice radicular. / Abstract: The purpose of this prospective study was to evaluate the results of treatment strategies for canines intrusion. Nineteen Class II, Division 1, with deep overbite individuals (age 11 to 15 years) were selected. Nine cases were used as a control group and the others were treated with the bioprogressive technique (Ricketts) for canine intrusion. Metallic implants were used for superimpositions. Lateral cephalometric radiographs were used for evaluation of lower incisors. Forty-five degrees cephalometric radiographs were used for evaluation of canines. These radiographs were taken after lower incisors intrusion and immediately after canines intrusion and about 6 months later for the control group. The results showed that the technique produced highly significant canines intrusion and a distal movement of center of resistence and root apex. There was no vertical displacement (extrusion) on lower incisors and a vestibular inclination was observed. / Mestre

A presente dissertação aborda a modelagem física de sistemas marinho profundo em escala reduzida, uma metodologia que vem contribuindo no entendimento dos processos sedimentares atuantes neste ambiente, principalmente as correntes de turbidez, fluxo gravitacional subaquoso responsável pela formação dos turbiditos, importantes reservatórios de hidrocarbonetos da costa brasileira. A modelagem física 3D empregada neste trabalho aborda a influência da autogênese no comportamento espacial e evolutivo de depósitos análogos gerados por simulações de correntes de turbidez em duas séries de 10 experimentos com parâmetros de controle constantes (vazão, concentração volumétrica de sedimentos, tipo e granulometria das partículas sedimentares), onde em cada série foi utilizada uma concentração de sedimentos diferente: uma com maior concentração – HDTC (high-density turbidity currents) e outra com menor concentração – LDTC (low-density turbidity currents) onde se buscou observar o efeito desta propriedade na construção dos depósitos. Para caracterizar o comportamento geométrico dos depósitos, uma nova abordagem estatística é utilizada a partir de uma análise de variância. Os resultados obtidos apontam que processos autogênicos locais puderam alterar a configuração global dos depósitos. A concentração de sedimentos teve influência direta nas características morfológicas e evolutivas dos depósitos, sendo os experimentos de HDTC os que apresentam uma evolução mais complexa, onde ocorreu um processo de auto-confinamento das correntes, gerando uma morfologia mais diversa. / Autogenic / allogenic controls have been discussed widely because they represent an important parameter in the constructive and evolutionary process of a sedimentary system. To evaluate these controls in submarine fans and analyse its capacity of selforganizing and creating depositional patterns, this work performed fully controlled 3D physical simulations of turbidity currents under ideal autogenic controls (no external influence) with detailed data for the generated deposits. Two series of 10 experiments of high-density turbidity currents (HDTC) and low-density turbidity currents (LDTC) were run, keeping all other input parameters (discharge, volumetric concentration, type and grain size) constant. From statistical and qualitative approach were characterised the geometric elements and morphodynamic behaviour of the deposits (centroid, Length/Width ratio, morphodynamic evolution). The results indicate local autogenic processes change the global setting of the flow evolution and deposits of submarine fans. A morphodynamic evolution generated by HDTC showed complex stages of filling and stacking caused by two types of flow self-channelling. Type I is characterised by flow channelling due to the elevation of levees without lateral avulsion and more efficient sediment transport (longer deposits, with terminal lobes well developed), and Type II is characterised by flow channelling but allows lateral avulsions and involves less efficient sediment transport (shorter deposits with terminal lobes undeveloped). The HDTC deposits showed random behaviour for the length/width ratio and for the centroid of sedimentary bodies and distinct morphological elements (elongated central deposit, fringes and distal lobes). By contrast, the LDTC morphodynamics were simplified without any self-confining process or distinct morphological elements. Finally, the statistical approach showed that the HDTC deposits had a greater variance of geometrical elements in relation to LDTC deposits. The experiments provided evidence that high rates of sediment supply decisively influenced the geometry and morphodynamic of the deposits, as well as they self-organizing capacity.

abstract: Modern machine learning systems leverage data and features from multiple modalities to gain more predictive power. In most scenarios, the modalities are vastly different and the acquired data are heterogeneous in nature. Consequently, building highly effective fusion algorithms is at the core to achieve improved model robustness and inferencing performance. This dissertation focuses on the representation learning approaches as the fusion strategy. Specifically, the objective is to learn the shared latent representation which jointly exploit the structural information encoded in all modalities, such that a straightforward learning model can be adopted to obtain the prediction. We first consider sensor fusion, a typical multimodal fusion problem critical to building a pervasive computing platform. A systematic fusion technique is described to support both multiple sensors and descriptors for activity recognition. Targeted to learn the optimal combination of kernels, Multiple Kernel Learning (MKL) algorithms have been successfully applied to numerous fusion problems in computer vision etc. Utilizing the MKL formulation, next we describe an auto-context algorithm for learning image context via the fusion with low-level descriptors. Furthermore, a principled fusion algorithm using deep learning to optimize kernel machines is developed. By bridging deep architectures with kernel optimization, this approach leverages the benefits of both paradigms and is applied to a wide variety of fusion problems. In many real-world applications, the modalities exhibit highly specific data structures, such as time sequences and graphs, and consequently, special design of the learning architecture is needed. In order to improve the temporal modeling for multivariate sequences, we developed two architectures centered around attention models. A novel clinical time series analysis model is proposed for several critical problems in healthcare. Another model coupled with triplet ranking loss as metric learning framework is described to better solve speaker diarization. Compared to state-of-the-art recurrent networks, these attention-based multivariate analysis tools achieve improved performance while having a lower computational complexity. Finally, in order to perform community detection on multilayer graphs, a fusion algorithm is described to derive node embedding from word embedding techniques and also exploit the complementary relational information contained in each layer of the graph. / Dissertation/Thesis / Doctoral Dissertation Electrical Engineering 2018

abstract: Mixture of experts is a machine learning ensemble approach that consists of individual models that are trained to be ``experts'' on subsets of the data, and a gating network that provides weights to output a combination of the expert predictions. Mixture of experts models do not currently see wide use due to difficulty in training diverse experts and high computational requirements. This work presents modifications of the mixture of experts formulation that use domain knowledge to improve training, and incorporate parameter sharing among experts to reduce computational requirements. First, this work presents an application of mixture of experts models for quality robust visual recognition. First it is shown that human subjects outperform deep neural networks on classification of distorted images, and then propose a model, MixQualNet, that is more robust to distortions. The proposed model consists of ``experts'' that are trained on a particular type of image distortion. The final output of the model is a weighted sum of the expert models, where the weights are determined by a separate gating network. The proposed model also incorporates weight sharing to reduce the number of parameters, as well as increase performance. Second, an application of mixture of experts to predict visual saliency is presented. A computational saliency model attempts to predict where humans will look in an image. In the proposed model, each expert network is trained to predict saliency for a set of closely related images. The final saliency map is computed as a weighted mixture of the expert networks' outputs, with weights determined by a separate gating network. The proposed model achieves better performance than several other visual saliency models and a baseline non-mixture model. Finally, this work introduces a saliency model that is a weighted mixture of models trained for different levels of saliency. Levels of saliency include high saliency, which corresponds to regions where almost all subjects look, and low saliency, which corresponds to regions where some, but not all subjects look. The weighted mixture shows improved performance compared with baseline models because of the diversity of the individual model predictions. / Dissertation/Thesis / Doctoral Dissertation Electrical Engineering 2018

abstract: This thesis research focuses on phylogenetic and functional studies of microbial communities in deep-sea water, an untapped reservoir of high metabolic and genetic diversity of microorganisms. The presence of photosynthetic cyanobacteria and diatoms is an interesting and unexpected discovery during a 16S ribosomal rRNA-based community structure analyses for microbial communities in the deep-sea water of the Pacific Ocean. Both RT-PCR and qRT-PCR approaches were employed to detect expression of the genes involved in photosynthesis of photoautotrophic organisms. Positive results were obtained and further proved the functional activity of these detected photosynthetic microbes in the deep-sea. Metagenomic and metatranscriptomic data was obtained, integrated, and analyzed from deep-sea microbial communities, including both prokaryotes and eukaryotes, from four different deep-sea sites ranging from the mesopelagic to the pelagic ocean. The RNA/DNA ratio was employed as an index to show the strength of metabolic activity of deep-sea microbes. These taxonomic and functional analyses of deep-sea microbial communities revealed a `defensive' life style of microbial communities living in the deep-sea water. Pseudoalteromonas sp.WG07 was subjected to transcriptomic analysis by application of RNA-Seq technology through the transcriptomic annotation using the genomes of closely related surface-water strain Pseudoalteromonas haloplanktis TAC125 and sediment strain Pseudoalteromonas sp. SM9913. The transcriptome survey and related functional analysis of WG07 revealed unique features different from TAC125 and SM9913 and provided clues as to how it adapted to its environmental niche. Also, a comparative transcriptomic analysis of WG07 revealed transcriptome changes between its exponential and stationary growing phases. / Dissertation/Thesis / Ph.D. Civil and Environmental Engineering 2013

Conselho Nacional de Desenvolvimento CientÃfico e TecnolÃgico / Estimar o recalque em estacas à um problema muito complexo, incerto e ainda nÃo totalmente compreendido, devido Ãs muitas incertezas associadas aos fatores que afetam a magnitude desta deformaÃÃo. As RNA sÃo ferramentas que funcionam analogamente ao cÃrebro humano, e sua unidade principal, o neurÃnio artificial, trabalha de maneira semelhante ao neurÃnio biolÃgico. Esta ferramenta alternativa vem sendo aplicada com sucesso em muitos problemas de engenharia geotÃcnica, podendo, portanto ser utilizadas como uma ferramentas alternativas para avaliar recalques em estacas isoladas. Nessa pesquisa as RNA utilizadas foram do tipo perceptron de mÃltiplas camadas, empregando um treinamento supervisionado utilizando o algoritmo de retropropagaÃÃo do erro. O modelo desenvolvido relaciona o recalque em estacas isoladas com as propriedades geomÃtricas das estacas (diÃmetro e comprimento), a estratigrafia e as caracterÃsticas de compacidade, ou consistÃncia dos solos por meio dos resultados obtidos nos ensaios SPT, e a carga atuante, obtidas em provas de carga realizadas em estacas hÃlice contÃnua, cravada metÃlica e escavada. O conjunto de aprendizagem foi composto por 1947 exemplos de entrada e saÃda. Com auxilio do programa QNET2000 foram treinadas e validadas vÃrias arquiteturas de redes neurais. ApÃs comparar o desempenho da curva carga x recalque elaborada com os recalques estimados pelo modelo proposto com a curva carga x recalque resultante da prova de carga estÃtica e com a curva carga x recalque gerada pelo emprego do programa comercial baseado em elementos finitos tridimensionais PLAXIS 3D Foundation, constatou-se que as RNA foram capazes de entender o comportamento das fundaÃÃes profundas do tipo estacas hÃlice contÃnua, escavada e cravada metÃlica, possibilitando dentre outras coisas, a definiÃÃo das cargas de trabalho e cargas limites nas estacas. / Predicting the settlement in deep foundation is a very complex, uncertain and not yet fully understood, due to the many uncertainties associated with factors that affect the magnitude of this deformation. Artificial Neural Network (ANN) is a tool that works similarly to the human brain, its main unit, the artificial neuron, works in a similar way to the biological neuron. This alternative tool has been successfully applied in many geotechnical engineering problems and can therefore be used as an alternative tool to evaluate the behavior of settlement in isolated piles. In this paper, the ANN used were the multilayer perceptron type, employing a supervised training that uses the error back propagation algorithm. The model developed relates settlement in isolated piles with the type and the geometrical properties of the piles (diameter and length), the stratigraphy and characteristics of compactness or consistency of soils by means of the SPT tests results, and the load applied, obtained in static pile load tests performed in continuous helix, steel driven and excavated pile types. The data set used to model consisted of 1.947 samples of input and output. QNET 2000 was the program used to assist the training and validation of various architectures of neural networks. The architecture formed by 10 nodes in the input layer, 28 neurons distributed in 4 intermediate layers and one neuron in the output layer, corresponding to the measured discharge for cutting (A10: 14:8:4:2:1) was the one that showed the best performance, with the correlation coefficient between the estimated settlements and settlements measured during the validation phase of 0.94, such value can be considered satisfactory when considering the prediction of a complex phenomenon. After comparing the performance of the applied load x settlement estimated by model proposed curve with the applied load x settlement measured in static pile load test curve and the applied load x settlement estimated by an elasto-plastic model thru numerical simulation, it was found that the ANN were able to understand the behavior of deep foundations of continuous helix, steel driven and excavated piles type, allowing among other things, the definition of workloads and load limits at the pile.