Notre but dans cette thèse est de construire un système qui réponde à une question en langue naturelle (NL) en représentant sa sémantique comme une forme logique (LF) et ensuite en calculant une réponse en exécutant cette LF sur une base de connaissances. La partie centrale d'un tel système est l'analyseur sémantique qui transforme les questions en formes logiques. Notre objectif est de construire des analyseurs sémantiques performants en apprenant à partir de paires (NL, LF). Nous proposons de combiner des réseaux neuronaux récurrents (RNN) avec des connaissances préalables symboliques exprimées à travers des grammaires hors-contexte (CFGs) et des automates. En intégrant des CFGs contrôlant la validité des LFs dans les processus d'apprentissage et d'inférence des RNNs, nous garantissons que les formes logiques générées sont bien formées; en intégrant, par le biais d'automates pondérés, des connaissances préalables sur la présence de certaines entités dans la LF, nous améliorons encore la performance de nos modèles. Expérimentalement, nous montrons que notre approche permet d'obtenir de meilleures performances que les analyseurs sémantiques qui n'utilisent pas de réseaux neuronaux, ainsi que les analyseurs à base de RNNs qui ne sont pas informés par de telles connaissances préalables / Our goal in this thesis is to build a system that answers a natural language question (NL) by representing its semantics as a logical form (LF) and then computing the answer by executing the LF over a knowledge base. The core part of such a system is the semantic parser that maps questions to logical forms. Our focus is how to build high-performance semantic parsers by learning from (NL, LF) pairs. We propose to combine recurrent neural networks (RNNs) with symbolic prior knowledge expressed through context-free grammars (CFGs) and automata. By integrating CFGs over LFs into the RNN training and inference processes, we guarantee that the generated logical forms are well-formed; by integrating, through weighted automata, prior knowledge over the presence of certain entities in the LF, we further enhance the performance of our models. Experimentally, we show that our approach achieves better performance than previous semantic parsers not using neural networks as well as RNNs not informed by such prior knowledge

Understanding road surface conditions is an important component in active vehicle safety. Estimations can be achieved through image classification using increasingly popular convolutional neural networks (CNNs). In this paper, we explore the effects of multi-task learning by creating CNNs capable of simultaneously performing the two tasks road surface condition classification (RSCC) and road scene semantic segmentation (RSSS). A multi-task network, containing a shared feature extractor (VGG16, ResNet-18, ResNet-101) and two taskspecific network branches, is built and trained using the Road-Conditions and Cityscapes datasets. We reveal that utilizing task-dependent homoscedastic uncertainty in the learning process improvesmulti-task model performance on both tasks. When performing task adaptation, using a small set of additional data labeled with semantic information, we gain considerable RSCC improvements on complex models. Furthermore, we demonstrate increased model generalizability in multi-task models, with up to 12% higher F1-score compared to single-task models.

Most research behind the use of Machine Learning models in the field of Human Activity Recognition focuses mainly on the classification of daily human activities and aerobic exercises. In this study, we focus on the use of 1 accelerometer and 2 gyroscope sensors to build a Deep Learning classifier to recognise 5 different strength exercises, as well as a null class. The strength exercises tested in this research are as followed: Bench press, bent row, deadlift, lateral rises and overhead press. The null class contains recordings of daily activities, such as sitting or walking around the house. The model used in this paper consists on the creation of consecutive overlapping fixed length sliding windows for each exercise, which are processed separately and act as the input for a Deep Convolutional Neural Network. In this study we compare different sliding windows lengths and overlap percentages (step sizes) to obtain the optimal window length and overlap percentage combination. Furthermore, we explore the accuracy results between 1D and 2D Convolutional Neural Networks. Cross validation is also used to check the overall accuracy of the classifiers, where the database used in this paper contains 5 exercises performed by 3 different users and a null class. Overall the models were found to perform accurately for window’s with length of 0.5 seconds or greater and provided a solid foundation to move forward in the creation of a more robust fully integrated model that can recognize a wider variety of exercises.

Phenotypic data is of crucial importance for plant breeding in estimating a plant's biomass. Traits such as leaf area and plant height are known to be correlated with biomass. Image analysis and computer vision methods can automate data analysis for high-throughput phenotyping. Many methods have been proposed for plant phenotyping in controlled environments such as greenhouses. In this thesis, we present multiple methods to estimate traits of the plant crop sorghum from images acquired from UAV and field-based sensors. We describe machine learning techniques to extract the plots of a crop field, a method for leaf counting from low-resolution images, and a statistical model that uses prior information about the field structure to estimate the center of each plant. We also develop a new loss function to train Convolutional Neural Networks (CNNs) to count and locate objects of any type and use it to estimate plant centers. Our methods are evaluated with ground truth of sorghum fields and publicly available datasets and are shown to outperform the state of the art in generic object detection and domain-specific tasks. <br><br>This thesis also examines the use of crowdsourcing information in video analytics. The large number of cameras deployed for public safety surveillance systems requires intelligent processing capable of automatically analyzing video in real time. We incorporate crowdsourcing in an online basis to improve a crowdflow estimation method. We present various approaches to characterize this uncertainty and to aggregate crowdsourcing results. Our techniques are evaluated using publicly available datasets.<br>

<p>Quick and accurate segmentation and object detection of the biomedical image is the starting point of most disease analysis and understanding of biological processes in medical research. It will enhance drug development and advance medical treatment, especially in cancer-related diseases. However, identifying the objects in the CT or MRI images and labeling them usually takes time even for an experienced person. Currently, there is no automatic detection technique for nucleus identification, pneumonia detection, and fetus brain segmentation. Fortunately, as the successful application of artificial intelligence (AI) in image processing, many challenging tasks are easily solved with deep convolutional neural networks. In light of this, in this thesis, the deep learning based object detection and segmentation methods were implemented to perform the nucleus segmentation, lung segmentation, pneumonia detection, and fetus brain segmentation. The semantic segmentation is achieved by the customized U-Net model, and the instance localization is achieved by Faster R-CNN. The reason we choose U-Net is that such a network can be trained end-to-end, which means the architecture of this network is very simple, straightforward and fast to train. Besides, for this project, the availability of the dataset is limited, which makes U-Net a more suitable choice. We also implemented the Faster R-CNN to achieve the object localization. Finally, we evaluated the performance of the two models and further compared the pros and cons of them. The preliminary results show that deep learning based technique outperforms all existing traditional segmentation algorithms. </p>

Reducing the amount of radiation in X-ray computed tomography has been an active area of research in the recent years. The reduction of radiation has the downside of degrading the quality of the CT scans by increasing the ratio of the noise. Therefore, some techniques must be utilized to enhance the quality of images. In this research, we approach the denoising problem using two class of algorithms and we reduce the noise in CT scans that have been acquired with 75% less dose to the patient compared to the normal dose scans. Initially, we implemented wavelet denoising to successfully reduce the noise in low-dose X-ray computed tomography (CT) images. The denoising was improved by finding the optimal threshold value instead of a non-optimal selected value. The mean structural similarity (MSSIM) index was used as the objective function for the optimization. The denoising performance of combinations of wavelet families, wavelet orders, decomposition levels, and thresholding methods were investigated. Results of this study have revealed the best combinations of wavelet orders and decomposition levels for low dose CT denoising. In addition, a new shrinkage function is proposed that provides better denoising results compared to the traditional ones without requiring a selected parameter. Alternatively, convolutional neural networks were employed using different architectures to resolve the same denoising problem. This new approach improved denoising even more in comparison to the wavelet denoising. / Includes bibliography. / Dissertation (Ph.D.)--Florida Atlantic University, 2018. / FAU Electronic Theses and Dissertations Collection

Smartphones and mobile tablets are rapidly growing, and very important nowadays. The most popular mobile operating system since 2012 has been Android. Android is an open source platform that allows developers to take full advantage of both the operating system and the applications itself. However, due to the open source community of an Android platform, some Android developers took advantage of this and created countless malicious applications such as Trojan, Malware, and Ransomware. All which are currently hidden in a large number of benign apps in official Android markets, such as Google PlayStore, and Amazon. Ransomware is a malware that once infected the victim’s device. It will encrypt files, unlock device system, and display a popup message which asks the victim to pay ransom in order to unlock their device or system which may include medical devices that connect through the internet. In this research, we propose to combine permission and API calls, then use Deep Learning techniques to detect ransomware apps from the Android market. Permissions setting and API calls are extracted from each app file by using a python library called AndroGuard. We are using Permissions and API call features to characterize each application, which can identify which application has potential to be ransomware or is benign. We implement our Android Ransomware Detection framework based on Keras, which uses MLP with back-propagation and a supervised algorithm. We used our method with experiments based on real-world applications with over 2000 benign applications and 1000 ransomware applications. The dataset came from ARGUS’s lab [1] which validated algorithm performance and selected the best architecture for the multi-layer perceptron (MLP) by trained our dataset with 6 various of MLP structures. Our experiments and validations show that the MLPs have over 3 hidden layers with medium sized of neurons achieved good results on both accuracy and AUC score of 98%. The worst score is approximately 45% to 60% and are from MLPs that have 2 hidden layers with large number of neurons. / Includes bibliography. / Thesis (M.S.)--Florida Atlantic University, 2018. / FAU Electronic Theses and Dissertations Collection

We are stumbling across a video tsunami flooding our communication channels. The ubiquity of digital cameras and social networks has increased the amount of visual media content generated and shared by people, in particular videos. Cisco reports that 82% of the internet traffic would be in the form of videos by 2022. The computer vision community has embraced this challenge by offering the first building blocks to translate the visual data in segmented video clips into semantic tags. However, users usually require to go beyond tagging at the video level. For example, someone may want to retrieve important moments such as the “first steps of her child” from a large collection of untrimmed videos; or retrieving all the instances of a home-run from an unsegmented video of baseball. In the face of this data deluge, it becomes crucial to develop efficient and scalable algorithms that can intelligently localize semantic visual content in untrimmed videos. In this work, I address three different challenges on the localization of actions in videos. First, I develop deep-based action proposals and detection models that take a video and generate action-agnostic and class-specific temporal segments, respectively. These models retrieve temporal locations with high accuracy in an efficient manner, faster than real-time. Second, I propose the new task to retrieve and localize temporal moments from a collection of videos given a natural language query. To tackle this challenge, I introduce an efficient and effective model that aligns the text query to individual clips of fixed length while still retrieves moments spanning multiple clips. This approach not only allows smooth interactions with users via natural languagequeries but also reduce the index size and search time for retrieving the moments. Lastly, I introduce the concept of actor-supervision that exploits the inherent compo sitionality of actions, in terms of transformations of actors, to achieve spatiotemporal localization of actions without the need of action box annotations. By designing ef ficient models to scan a single video in real-time; retrieve and localizing moments of interest from multiple videos; and an effective strategy to localize actions without resorting in action box annotations, this thesis provides insights that put us closer to the goal of general video understanding.

Manual analysis of medical images such as magnetic resonance imaging (MRI) requires a trained professional, is time-consuming and results may vary between experts. We propose an automated method for kidney segmentation using a convolutional Neural Network (CNN) model based on the U-Net architecture. Investigations are done to compare segmentations between trained experts, inexperienced operators and the Neural Network model, showing near human expert level performance from the Neural Network. Stratified sampling is performed when selecting which subject volumes to perform manual segmentations on to create training data. Experiments are run to test the effectiveness of transfer learning and data augmentation and we show that one of the most important components of a successful machine learning pipeline is larger quantities of carefully annotated data for training.

Submitted by Caroline Xavier (caroline.xavier@pucrs.br) on 2017-06-30T17:22:52Z No. of bitstreams: 1 DIS_ANIBAL_SOLON_HEINSFELD_COMPLETO.pdf: 12807619 bytes, checksum: d11b60094a8bde0d839a6f7a23bbb56c (MD5) / Made available in DSpace on 2017-06-30T17:22:52Z (GMT). No. of bitstreams: 1 DIS_ANIBAL_SOLON_HEINSFELD_COMPLETO.pdf: 12807619 bytes, checksum: d11b60094a8bde0d839a6f7a23bbb56c (MD5) Previous issue date: 2016-03-28 / O Espectro Autista (EA) compreende uma s?rie de desordens no desenvolvimento neurol?gico, caracterizado por defici?ncias sociais e dificuldades de comunica??o, comportamentos repetitivos e atrasos cognitivos. Atualmente, o diagn?stico do EA ? amplamente baseado em medi??es comportamentais, que pode ser demorado, e depende da coopera??o do paciente e da experi?ncia do examinador. Para mitigar esta limita??o, investigamos padr?es neurais que ajudem no diagn?stico de desordens do EA. Nesta disserta??o, usamos t?cnicas de deep learning, a fim de extrair caracter?sticas robustas de neuroimagens de pacientes com autismo. Neuroimagens cont?m cerca de 300.000 pontos espaciais, com aproximadamente 200 medi??es cada. As t?cnicas de deep learning s?o ?teis para extrair caracter?sticas relevantes que diferenciam autistas de n?o-autistas. Ao utilizar denoising autoencoders, uma t?cnica de deep learning espec?fica que visa reduzir a dimensionalidade dos dados, n?s superamos o estado da arte, atingindo 69% de acur?cia, comparado com o melhor resultado encontrado na literatura, com 60% de acur?cia. / Autism Spectrum Disorders (ASD) comprise a range of neurodevelopmental disorders, characterized by social deficits and communication difficulties, repetitive behaviors, and cognitive delays. The diagnosis of ASD is largely based on behavioral measurements, which can be timeconsuming and relies on the patient cooperation and examiner expertise. In order to address this limitation, we aim to investigate neural patterns to help in the diagnosis of ASD. In this dissertation, we use deep learning techniques to extract robust characteristics from neuroimages of autistic subject brain function. Since neuroimage contains about 300,000 spatial points, with approximately 200 temporal measurements each, deep learning techniques are useful in order to extract important features to discriminate ASD subjects from non-ASD. By using denoising autoencoders, a specific deep learning technique that aims to reduce data dimensionality, we surpass the state-of-the-art by achieving 69% of accuracy, compared to 60% using the same dataset.