Soft Tissue Sarcomas (STS) are among the most dangerous diseases, with a 50% mortality rate in the USA in 2016. Heterogeneous responses to the treatments of the same sub-type of STS as well as intra-tumor heterogeneity make the study of biopsies imprecise. Radiologists make efforts to find non-invasive approaches to gather useful and important information regarding characteristics and behaviors of STS tumors, such as aggressiveness and recurrence. Quantitative image analysis is an approach to integrate information extracted using data science, such as data mining and machine learning with biological an clinical data to assist radiologists in making the best recommendation on clinical trials and the course of treatment. The new methods in “Radiomics" extract meaningful features from medical imaging data for diagnostic and prognostic goals. Furthermore, features extracted from Convolutional Neural Networks (CNNs) are demonstrating very powerful and robust performance in computer aided decision systems (CADs). Also, a well-known computer vision approach, Bag of Visual Words, has recently been applied on imaging data for machine learning purposes such as classification of different types of tumors based on their specific behavior and phenotype. These approaches are not fully and widely investigated in STS. This dissertation provides novel versions of image analysis based on Radiomics and Bag of Visual Words integrated with deep features to quantify the heterogeneity of entire STS as well as sub-regions, which have predictive and prognostic imaging features, from single and multi-sequence Magnetic Resonance Imaging (MRI). STS are types of cancer which are rarely touched in term of quantitative cancer analysis versus other type of cancers such as lung, brain and breast cancers. This dissertation does a comprehensive analysis on available data in 2D and multi-slice to predict the behavior of the STS with regard to clinical outcomes such as recurrence or metastasis and amount of tumor necrosis. The experimental results using Radiomics as well as a new ensemble of Bags of Visual Words framework are promising with 91.66% classification accuracy and 0.91 AUC for metastasis, using ensemble of Bags of Visual Words framework integrated with deep features, and 82.44% classification accuracy with 0.63 AUC for necrosis progression, using Radiomics framework, in tests on the available datasets.

Human activity recognition (HAR) based on time series data is the problem of classifying various patterns. Its widely applications in health care owns huge commercial benefit. With the increasing spread of smart devices, people have strong desires of customizing services or product adaptive to their features. Deep learning models could handle HAR tasks with a satisfied result. However, training a deep learning model has to consume lots of time and computation resource. Consequently, developing a HAR system effectively becomes a challenging task. In this study, we develop a solid HAR system using Convolutional Neural Network based on transfer learning, which can eliminate those barriers.

Classification of emotions plays a very important role in affective computing and has real-world applications in fields as diverse as entertainment, medical, defense, retail, and education. These applications include video games, virtual reality, pain recognition, lie detection, classification of Autistic Spectrum Disorder (ASD), analysis of stress levels, and determining attention levels. This vast range of applications motivated us to study automatic emotion recognition which can be done by using facial expression, speech, and physiological data. A person’s physiological signals such are heart rate, and blood pressure are deeply linked with their emotional states and can be used to identify a variety of emotions; however, they are less frequently explored for emotion recognition compared to audiovisual signals such as facial expression and voice. In this thesis, we investigate a multimodal approach to emotion recognition using physiological signals by showing how these signals can be combined and used to accurately identify a wide range of emotions such as happiness, sadness, and pain. We use the deep convolutional neural network for our experiments. We also detail comparisons between gender-specific models of emotion. Our investigation makes use of deep convolutional neural networks, which are the latest state of the art in supervised learning, on two publicly available databases, namely DEAP and BP4D+. We achieved an average emotion recognition accuracy of 98.89\% on BP4D+ and on DEAP it is 86.09\% for valence, 90.61\% for arousal, 90.48\% for liking and 90.95\% for dominance. We also compare our results to the current state of the art, showing the superior performance of our method.

Through the development and implementation of modified learning contexts, the current study encouraged undergraduate teacher education students to modify their approaches to learning by reducing their reliance on surface approaches and progressively adopting deeper approaches. This outcome was considered desirable because students who employed deep approaches would exit the course having achieved higher quality learning than those who relied primarily on surface approaches. It was expected that higher quality learning in a preservice teacher education program would also translate into greater self-confidence in the management of teaching tasks, leading to improvements in students� teaching self-efficacy beliefs. Altered learning contexts were developed through the application of action research methodology involving core members of the teaching team. Learning activities were designed with a focus on co-operative small-group problem-based learning, which included multiple subtasks requiring variable outcome presentation modes. Linked individual reflection was encouraged by personal learning journals and learning portfolios. Students also provided critical analyses of their own learning during the completion of tasks, from both individual and group perspectives. Assessment methods included lecturer, peer and self-assessment, depending on the nature of the learning task. Often these were integrated, so that subtasks within larger ones were assessed using combinations of methods. Learning approach theorists (Biggs, 1993a, 1999; Entwistle, 1986, 1998; Prosser & Trigwell, 1999; Ramsden, 1992, 1997) contend that learning outcomes are directly related to the learning approaches used in their development. They further contend that the approach adopted is largely a result of students� intent, which in turn, is influenced by their perception of the learning context. The present study therefore aimed to develop an integrated and pervasive course-based learning context, constructively aligned (after: Biggs, 1993a, 1996), achievable within the normal constraints of a university program, that would influence students� adoption of deep learning approaches. The cognitive processes students used in response to the altered contexts were interpreted in accordance with self-regulatory internal logic (after: Bandura, 1986, 1991b; Zimmerman, 1989, 1998b). Longitudinal quasi-experimental methods with repeated measures on non-equivalent dependent variables were applied to three cohorts of students. Cohort 1 represented the contrast group who followed a traditional program. Cohort 2 was the main treatment group to whom the modified program was presented. Cohort 3 represented a comparison group that was also presented with the modified program over a shorter period. Student data on learning approach, teaching efficacy and academic attributions were gathered from repeated administrations of the Study Process Questionnaire (Biggs, 1987b), Teacher Efficacy Scale (Gibson & Dembo, 1984) and Multidimensional-Multiattributional Causality Scale (Lefcourt, 1991). In addition, reflective journals, field observations and transcripts of interviews undertaken at the beginning and conclusion of the course, were used to clarify students� approaches to learning and their responses to program modifications. Analyses of learning approaches adopted by Cohorts 1 and 2 revealed that they both began their course predominantly using surface approaches. While students in Cohort 1 completed the course with approximately equal reliance on deep and surface approaches, students in Cohort 2 reported a predominant use of deep approaches on course completion. The relative impact of the modified learning context on students with differing approaches to learning in this cohort were further explained through qualitative data and cluster analyses. The partial replication of the study with Cohort 3, across the first three semesters of their program, produced similar effects to those obtained with Cohort 2. The analyses conducted with teaching efficacy data indicated a similar pattern of development for all cohorts. Little change in either personal or general dimensions was noted in the first half of the program, followed by strong growth in both, in the latter half. While a relationship between learning approach usage and teaching efficacy was not apparent in Cohort 1, developmental path and mediation analyses indicated that the use of deep learning approaches considerably influenced the development of personal teaching efficacy in Cohort 2. The current research suggests that value lies in the construction of learning environments, in teacher education, that enhance students� adoption of deep learning approaches. The nature of the task is complex, multifaceted and context specific, most likely requiring the development of unique solutions in each environment. Nevertheless, this research demonstrates that such solutions can be developed and applied within the prevailing constraints of pre-existing course structures.

Since 2006, deep learning algorithms which rely on deep architectures with several layers of increasingly complex representations have been able to outperform state-of-the-art methods in several settings. Deep architectures can be very efficient in terms of the number of parameters required to represent complex operations which makes them very appealing to achieve good generalization with small amounts of data. Although training deep architectures has traditionally been considered a difficult problem, a successful approach has been to employ an unsupervised layer-wise pre-training step to initialize deep supervised models. First, unsupervised learning has many benefits w.r.t. generalization because it only relies on unlabeled data which is easily found. Second, the possibility to learn representations layer by layer instead of all layers at once improves generalization further and reduces computational time. However, deep learning is a very recent approach and still poses a lot of theoretical and practical questions concerning the consistency of layer-wise learning with many layers and difficulties such as evaluating performance, performing model selection and optimizing layers. In this thesis we first discuss the limitations of the current variational justification for layer-wise learning which does not generalize well to many layers. We ask if a layer-wise method can ever be truly consistent, i.e. capable of finding an optimal deep model by training one layer at a time without knowledge of the upper layers. We find that layer-wise learning can in fact be consistent and can lead to optimal deep generative models. To do this, we introduce the Best Latent Marginal (BLM) upper bound, a new criterion which represents the maximum log-likelihood of a deep generative model where the upper layers are unspecified. We prove that maximizing this criterion for each layer leads to an optimal deep architecture, provided the rest of the training goes well. Although this criterion cannot be computed exactly, we show that it can be maximized effectively by auto-encoders when the encoder part of the model is allowed to be as rich as possible. This gives a new justification for stacking models trained to reproduce their input and yields better results than the state-of-the-art variational approach. Additionally, we give a tractable approximation of the BLM upper-bound and show that it can accurately estimate the final log-likelihood of models. Taking advantage of these theoretical advances, we propose a new method for performing layer-wise model selection in deep architectures, and a new criterion to assess whether adding more layers is warranted. As for the difficulty of training layers, we also study the impact of metrics and parametrization on the commonly used gradient descent procedure for log-likelihood maximization. We show that gradient descent is implicitly linked with the metric of the underlying space and that the Euclidean metric may often be an unsuitable choice as it introduces a dependence on parametrization and can lead to a breach of symmetry. To mitigate this problem, we study the benefits of the natural gradient and show that it can restore symmetry, regrettably at a high computational cost. We thus propose that a centered parametrization may alleviate the problem with almost no computational overhead.

In this research project, the potential of integrating ePortfolios into the reflection and assessment component between a personal trainer and client was examined. Research demonstrates that more effective learning occurs when reflection and engagement are integral components. The problem that this thesis aimed to answer was the importance of enhancing student engagement, in the classroom and beyond. New technology applications like ePortfolios allow for great student reflection and deep learning but are not universally applied. As such, this study explored the potential of integrating an e portfolio as a reflection technology to enhance engagement during learning. While ePortfolios are commonly found in an academic setting, professionals in healthcare have also used ePortfolios with their patients, clients or peers. Could ePortfolios be used by personal trainers, clients, coaches and athletes? If so, how? What would be the outcomes? The focus of this research project was to explore the potential of integrating ePortfolios into the reflection and assessment component between a personal trainer/client. Eight trainers throughout the Ottawa area were interviewed to gain an understanding of their perception on the ways in which trainers could enhance their client’s fitness journey through the use of an ePortfolio and if so, how would the ePortfolio would be used. The results of this research project provided insight as to whether or not reflective technology strategies such as ePportfolios would enhance the client’s fitness journey and if so, how would the ePortfolio be used. Ultimately, this thesis will inform our understanding of how ePortfolios can be transferred from an academic setting to foster deep learning in kinesiology settings.

Through the development and implementation of modified learning contexts, the current study encouraged undergraduate teacher education students to modify their approaches to learning by reducing their reliance on surface approaches and progressively adopting deeper approaches. This outcome was considered desirable because students who employed deep approaches would exit the course having achieved higher quality learning than those who relied primarily on surface approaches. It was expected that higher quality learning in a preservice teacher education program would also translate into greater self-confidence in the management of teaching tasks, leading to improvements in students� teaching self-efficacy beliefs. Altered learning contexts were developed through the application of action research methodology involving core members of the teaching team. Learning activities were designed with a focus on co-operative small-group problem-based learning, which included multiple subtasks requiring variable outcome presentation modes. Linked individual reflection was encouraged by personal learning journals and learning portfolios. Students also provided critical analyses of their own learning during the completion of tasks, from both individual and group perspectives. Assessment methods included lecturer, peer and self-assessment, depending on the nature of the learning task. Often these were integrated, so that subtasks within larger ones were assessed using combinations of methods. Learning approach theorists (Biggs, 1993a, 1999; Entwistle, 1986, 1998; Prosser & Trigwell, 1999; Ramsden, 1992, 1997) contend that learning outcomes are directly related to the learning approaches used in their development. They further contend that the approach adopted is largely a result of students� intent, which in turn, is influenced by their perception of the learning context. The present study therefore aimed to develop an integrated and pervasive course-based learning context, constructively aligned (after: Biggs, 1993a, 1996), achievable within the normal constraints of a university program, that would influence students� adoption of deep learning approaches. The cognitive processes students used in response to the altered contexts were interpreted in accordance with self-regulatory internal logic (after: Bandura, 1986, 1991b; Zimmerman, 1989, 1998b). Longitudinal quasi-experimental methods with repeated measures on non-equivalent dependent variables were applied to three cohorts of students. Cohort 1 represented the contrast group who followed a traditional program. Cohort 2 was the main treatment group to whom the modified program was presented. Cohort 3 represented a comparison group that was also presented with the modified program over a shorter period. Student data on learning approach, teaching efficacy and academic attributions were gathered from repeated administrations of the Study Process Questionnaire (Biggs, 1987b), Teacher Efficacy Scale (Gibson & Dembo, 1984) and Multidimensional-Multiattributional Causality Scale (Lefcourt, 1991). In addition, reflective journals, field observations and transcripts of interviews undertaken at the beginning and conclusion of the course, were used to clarify students� approaches to learning and their responses to program modifications. Analyses of learning approaches adopted by Cohorts 1 and 2 revealed that they both began their course predominantly using surface approaches. While students in Cohort 1 completed the course with approximately equal reliance on deep and surface approaches, students in Cohort 2 reported a predominant use of deep approaches on course completion. The relative impact of the modified learning context on students with differing approaches to learning in this cohort were further explained through qualitative data and cluster analyses. The partial replication of the study with Cohort 3, across the first three semesters of their program, produced similar effects to those obtained with Cohort 2. The analyses conducted with teaching efficacy data indicated a similar pattern of development for all cohorts. Little change in either personal or general dimensions was noted in the first half of the program, followed by strong growth in both, in the latter half. While a relationship between learning approach usage and teaching efficacy was not apparent in Cohort 1, developmental path and mediation analyses indicated that the use of deep learning approaches considerably influenced the development of personal teaching efficacy in Cohort 2. The current research suggests that value lies in the construction of learning environments, in teacher education, that enhance students� adoption of deep learning approaches. The nature of the task is complex, multifaceted and context specific, most likely requiring the development of unique solutions in each environment. Nevertheless, this research demonstrates that such solutions can be developed and applied within the prevailing constraints of pre-existing course structures.

Techniques described to date for the reproducible and noninvasive quantification of regional cardiac function have been largely relegated to research settings due to time-consuming and cumbersome image acquisition and analysis. In this thesis, feature tracking algorithms are developed for 2-D+Time cardiac magnetic resonance (CMR) and 3-D+Time cardiac computed tomography (CCT) image sequences that are easily acquired clinically, while emphasising reproducibility and automation in their design. First, a commercially-implemented CMR feature tracking algorithm for the analysis of steady state free precession (SSFP) cine series is evaluated in patients with hypertrophic cardiomyopathy (HCM) and arrhythmogenic right ventricular cardiomyopathy (ARVC), which primarily affect the left ventricle (LV) and right ventricle (RV), respectively, and functional impairment compared with control populations is found in both cases. The limitations of this implementation are then used to guide development of an automated algorithm for the same purpose, making use of fully convolutional neural networks (CNN) for segmentation and spline registration across all frames simultaneously for tracking. This study is performed in the subjects with HCM, and functional impairment is again identified in disease subjects. Finally, as myocardial contraction is inherently a 3-D phenomenon, a technique is developed for quantification of regional function from 3-D+Time functional CCT studies using simultaneous registration of automatically generated Loop subdivision surface models for tracking. This study is performed in canine mongrels, and compared with the current state of the art technique for CCT functional analysis. This work demonstrates the feasibility of automated, reproducible cardiac functional analysis from CMR and CCT image sequences. While work remains to be done in extending the principles demonstrated and modular components described to fully automated whole-heart analysis, it is hoped that this thesis will accelerate the clinical adoption of regional functional analysis.

Les thèmes principaux abordés dans cette thèse sont l'utilisation de méthodes d'apprentissage actif et d'apprentissage profond dans le contexte du traitement de documents multimodaux. Les contributions proposées dans cette thèse abordent ces deux thèmes. Un système d'apprentissage actif a été introduit pour permettre une annotation plus efficace des émissions de télévision grâce à la propagation des étiquettes, à l'utilisation de données multimodales et à des stratégies de sélection efficaces. Plusieurs scénarios et expériences ont été envisagés dans le cadre de l'identification des personnes dans les vidéos, en prenant en compte l'utilisation de différentes modalités (telles que les visages, les segments de la parole et le texte superposé) et différentes stratégies de sélection. Le système complet a été validé au cours d'un ``test à blanc'' impliquant des annotateurs humains réels.Une deuxième contribution majeure a été l'étude et l'utilisation de l'apprentissage profond (en particulier les réseaux de neurones convolutifs) pour la recherche d'information dans les vidéos. Une étude exhaustive a été réalisée en utilisant différentes architectures de réseaux neuronaux et différentes techniques d'apprentissage telles que le réglage fin (fine-tuning) ou des classificateurs plus classiques comme les SVMs. Une comparaison a été faite entre les caractéristiques apprises (la sortie des réseaux neuronaux) et les caractéristiques plus classiques (``engineered features''). Malgré la performance inférieure des seconds, une fusion de ces deux types de caractéristiques augmente la performance globale.Enfin, l'utilisation d'un réseau neuronal convolutif pour l'identification des locuteurs à l'aide de spectrogrammes a été explorée. Les résultats ont été comparés à ceux obtenus avec d'autres systèmes d'identification de locuteurs récents. Différentes approches de fusion ont également été testées. L'approche proposée a permis d'obtenir des résultats comparables à ceux certains des autres systèmes testés et a offert une augmentation de la performance lorsqu'elle est fusionnée avec la sortie du meilleur système. / The main topics of this thesis include the use of active learning-based methods and deep learning in the context of retrieval of multimodal documents. The contributions proposed during this thesis address both these topics. An active learning framework was introduced, which allows for a more efficient annotation of broadcast TV videos thanks to the propagation of labels, the use of multimodal data and selection strategies. Several different scenarios and experiments were considered in the context of person identification in videos, including using different modalities (such as faces, speech segments and overlaid text) and different selection strategies. The whole system was additionally validated in a dry run involving real human annotators.A second major contribution was the investigation and use of deep learning (in particular the convolutional neural network) for video retrieval. A comprehensive study was made using different neural network architectures and training techniques such as fine-tuning or using separate classifiers like SVM. A comparison was made between learned features (the output of neural networks) and engineered features. Despite the lower performance of the engineered features, fusion between these two types of features increases overall performance.Finally, the use of convolutional neural network for speaker identification using spectrograms is explored. The results are compared to other state-of-the-art speaker identification systems. Different fusion approaches are also tested. The proposed approach obtains comparable results to some of the other tested approaches and offers an increase in performance when fused with the output of the best system.

O aprendizado profundo tem inúmeras aplicações na visão computacional, reconhecimento de fala, processamento de linguagem natural e outras aplicações de interesse comercial. A visão computacional, por sua vez, possui muitas aplicações em áreas distintas, indo desde o entretenimento à aplicações relevantes e críticas. O reconhecimento e manipulação de faces (Snapchat), e a descrição de objetos em fotos (OneDrive) são exemplos de aplicações no entretenimento. Ao passo que, a inspeção industrial, o diagnóstico médico, o reconhecimento de objetos em imagens capturadas por satélites (usadas em missões de resgate e defesa), os carros autônomos e o Sistema Avançado de Auxílio ao Motorista (SAAM) são exemplos de aplicações relevantes e críticas. Algumas das empresas de circuitos integrados mais importantes do mundo, como Xilinx, Intel e Nvidia estão apostando em plataformas dedicadas para acelerar o treinamento e a implementação de algoritmos de aprendizado profundo e outras alternativas de visão computacional para carros autônomos e SAAM devido às suas altas necessidades computacionais. Assim, implementar sistemas de aprendizado profundo que alcançam alto desempenho com o custo de baixa utilização de área e dissipação de potência é um grande desafio. Além do mais, os circuitos eletrônicos para a indústria automotiva devem ser confiáveis mesmo sob efeitos da radiação, defeitos de fabricação e efeitos do envelhecimento. Assim, um gerador automático de VHSIC (Very High Speed Integrated Circuit) Hardware Description Language (VHDL) para Redes Neurais Convolucionais (RNC) foi desenvolvido para reduzir o tempo associado a implementação de algoritmos de aprendizado profundo em hardware. Como estudo de caso, uma RNC foi treinada pela ferramenta Convolutional Architecture for Fast Feature Embedding (Caffe), de modo a classificar 6 classes de placas de trânsito, alcançando uma precisão de cerca de 89,8% no conjunto de dados German Traffic-Sign Recognition Benchmark (GTSRB), que contém imagens de placas de trânsito em cenários complexos. Essa RNC foi implementada num All-Programmable System-on- Chip (APSoC) Zynq-7000, resultando em 313 Frames Por Segundo (FPS) em imagens normalizadas para 32x32, com o APSoC dissipando uma potência de somente 2.057 W, enquanto uma Graphics Processing Unit (GPU) embarcada, em seu modo de operação mínimo, dissipa 10 W. A confiabilidade da RNC proposta foi investigada por injeções de falhas acumuladas e aleatórias por emulação nos bits de configuração da Lógica Programável (LP) do APSoC, alcançando uma confiabilidade de 80,5% sob Single-Bit-Upset (SBU) onde foram considerados ambos os Dados Corrompidos Silenciosos (DCSs) críticos e os casos em que o sistema não respondeu no tempo esperado (time-outs). Em relação às falhas múltiplas, a confiabilidade da RNC decresce exponencialmente com o número de falhas acumuladas. Em vista disso, a confiabilidade da RNC proposta deve ser aumentada através do uso de técnicas de proteção durante o fluxo de projeto. / Deep learning has a plethora of applications in computer vision, speech recognition, natural language processing and other applications of commercial interest. Computer vision, in turn, has many applications in distinct areas, ranging from entertainment applications to relevant and critical applications. Face recognition and manipulation (Snapchat), and object description in pictures (OneDrive) are examples of entertainment applications. Industrial inspection, medical diagnostics, object recognition in images captured by satellites (used in rescue and defense missions), autonomous cars and Advanced Driver-Assistance System (ADAS) are examples of relevant and critical applications. Some of the most important integrated circuit companies around the world, such as Xilinx, Intel and Nvidia are waging in dedicated platforms for accelerating the training and deployment of deep learning and other computer vision algorithms for autonomous cars and ADAS due to their high computational requirement. Thus, implementing a deep learning system that achieves high performance with low area utilization and power consumption costs is a big challenge. Besides, electronic equipment for automotive industry must be reliable even under radiation effects, manufacturing defects and aging effects, inasmuch as if a system failure occurs, a car accident can happen. Thus, a Convolutional Neural Network (CNN) VHSIC (Very High Speed Integrated Circuit) Hardware Description Language (VHDL) automatic generator was developed to reduce the design time associated to the implementation of deep learning algorithms in hardware. As a case study, a CNN was trained by the Convolutional Architecture for Fast Feature Embedding (Caffe) framework, in order to classify 6 traffic-sign classes, achieving an average accuracy of about 89.8% on the German Traffic-Sign Recognition Benchmark (GTSRB) dataset, which contains trafficsigns images in complex scenarios. This CNN was implemented on a Zynq-7000 All- Programmable System-on-Chip (APSoC), achieving about 313 Frames Per Second (FPS) on 32x32-normalized images, with the APSoC consuming only 2.057W, while an embedded Graphics Processing Unit (GPU), in its minimum operation mode, consumes 10W. The proposed CNN reliability was investigated by random piled-up fault injection by emulation in the Programming Logic (PL) configuration bits of the APSoC, achieving 80.5% of reliability under Single-Bit-Upset (SBU) where both critical Silent Data Corruptions (SDCs) and time-outs were considered. Regarding the multiple faults, the proposed CNN reliability exponentially decreases with the number of piled-up faults. Hence, the proposed CNN reliability must be increased by using hardening techniques during the design flow.