Digitized by Kansas Correctional Industries

Deep learning and convolutional neural networks have become the dominant tool for computer vision. These techniques excel at learning complicated representations from data using supervised learning. In particular, image recognition models now out-perform human baselines under constrained settings. However, the science of computer vision aims to build machines which can see. This requires models which can extract richer information than recognition, from images and video. In general, applying these deep learning models from recognition to other problems in computer vision is significantly more challenging. This thesis presents end-to-end deep learning architectures for a number of core computer vision problems; scene understanding, camera pose estimation, stereo vision and video semantic segmentation. Our models outperform traditional approaches and advance state-of-the-art on a number of challenging computer vision benchmarks. However, these end-to-end models are often not interpretable and require enormous quantities of training data. To address this, we make two observations: (i) we do not need to learn everything from scratch, we know a lot about the physical world, and (ii) we cannot know everything from data, our models should be aware of what they do not know. This thesis explores these ideas using concepts from geometry and uncertainty. Specifically, we show how to improve end-to-end deep learning models by leveraging the underlying geometry of the problem. We explicitly model concepts such as epipolar geometry to learn with unsupervised learning, which improves performance. Secondly, we introduce ideas from probabilistic modelling and Bayesian deep learning to understand uncertainty in computer vision models. We show how to quantify different types of uncertainty, improving safety for real world applications.

Learning discriminative feature representations has attracted a great deal of attention since it is a critical step to facilitate the subsequent classification, retrieval and recommendation tasks. In this dissertation, besides incorporating prior knowledge about image labels into the image classification as most prevalent feature learning methods currently do, we also explore some other general-purpose priors and verify their effectiveness in the discriminant feature learning. As a more powerful representation can be learned by implementing such general priors, our approaches achieve state-of-the-art results on challenging benchmarks. We elaborate on these general-purpose priors and highlight where we have made novel contributions. We apply sparsity and hierarchical priors to the explanatory factors that describe the data, in order to better discover the data structure. More specifically, in the first approach we propose that we only incorporate sparse priors into the feature learning. To this end, we present a support discrimination dictionary learning method, which finds a dictionary under which the feature representation of images from the same class have a common sparse structure while the size of the overlapped signal support of different classes is minimised. Then we incorporate sparse priors and hierarchical priors into a unified framework, that is capable of controlling the sparsity of the neuron activation in deep neural networks. Our proposed approach automatically selects the most useful low-level features and effectively combines them into more powerful and discriminative features for our specific image classification problem. We also explore priors on the relationships between multiple factors. When multiple independent factors exist in the image generation process and only some of them are of interest to us, we propose a novel multi-task adversarial network to learn a disentangled feature which is optimized with respect to the factor of interest to us, while being distraction factors agnostic. When common factors exist in multiple tasks, leveraging common factors cannot only make the learned feature representation more robust, but also enable the model to generalise from very few labelled samples. More specifically, we address the domain adaptation problem and propose the re-weighted adversarial adaptation network to reduce the feature distribution divergence and adapt the classifier from source to target domains.

Effective classification with imbalanced data is an important area of research, as high class imbalance is naturally inherent in many real-world applications, e.g. anomaly detection. Modeling such skewed data distributions is often very difficult, and non-standard methods are sometimes required to combat these negative effects. These challenges have been studied thoroughly using traditional machine learning algorithms, but very little empirical work exists in the area of deep learning with class imbalanced big data. Following an in-depth survey of deep learning methods for addressing class imbalance, we evaluate various methods for addressing imbalance on the task of detecting Medicare fraud, a big data problem characterized by extreme class imbalance. Case studies herein demonstrate the impact of class imbalance on neural networks, evaluate the efficacy of data-level and algorithm-level methods, and achieve state-of-the-art results on the given Medicare data set. Results indicate that combining under-sampling and over-sampling maximizes both performance and efficiency. / Includes bibliography. / Thesis (M.S.)--Florida Atlantic University, 2019. / FAU Electronic Theses and Dissertations Collection

At the Castle Creek study area in east-central British Columbia a well-exposed section about 450 m wide and 30 m thick in the (Neoproterozoic) Isaac Formation was analyzed to document vertical and lateral changes in a succession of distinctively heterolithic strata. Strata are interpreted to have been deposited on a deep-marine levee that was sandwiched between its genetically related channel on one side and an erosional escarpment sculpted by an older (underlying) channel on the other. Flows that overspilled the channel (incident flow) eventually encountered the escarpment, which then set up a return flow oriented more or less opposite to the incident (from the channel) flow. This created an area of complex flow that became manifested in the sedimentary record as a highly tabular succession of intricately interstratified sand and mud overlain by an anomalously thick, plane-parallel interlaminated sand-mud unit capped finally by a claystone.

This thesis shows how children’s learning is influenced and modified by the teaching environment. The metacognitive, self-regulatory learning behaviours of sixteen kindergarten students were examined in order to determine how students perceive learning, either by adopting deep approaches, where the focus is on understanding and meaning, or surface approaches, where the meeting of institutional demands frequently subjugate the former goals. The data have been analysed within a qualitative paradigm from a phenomenographic perspective. The study addresses three issues: the nature and frequency of the strategic learning behaviours displayed by the students; the contribution strategic behaviours make to the adoption of deep or surface learning approaches; and how metacognitive teaching environments influence higher-order thinking. Findings reveal that where teachers had metcognitive training, the frequency of strategy use increased irrespective of student performance. High achieving students used more strategic behaviours, used them with greater efficiency, and tended to display more of the characteristics of deep approach learners. This study suggests that many of the differential outcomes evident amongst students may be substantially reduced through early and consistent training within a teaching environment conductive to the development of metacognitive, self-regulatory behaviours and deep learning approaches / Doctor of Philosophy (PhD)

Typescript. Copies of three papers (2 published), co-authored by the author, in back. Bibliography: leaves 179-186.

The current study examined cognitive effects of unilateral subthalamic nucleus (STN) deep brain stimulation (DBS) in Parkinson's disease (PD) patients. Neuropsychological evaluations were conducted at baseline and follow-up. Data was collected from 28 unilateral STN DBS patients (15 English- and 13 Spanish-speaking), and 15 English-speaking matched PD control patients. English-speaking DBS patients demonstrated significant declines in verbal fluency and attention/executive function, whereas PD control patients did not experience significant cognitive decline. Cognitive performance did not differ based on side of DBS. Spanish-speaking DBS patients experienced significant declines in verbal fluency, confrontational naming and visuospatial abilities. Among Spanish-speaking DBS patients, older age and later age of disease onset predicted verbal fluency decline, even after controlling for education.

The Mid Ocean Ridge system with its volcanism and related hydrothermal activity has been a subject for many studies since the discovery of high temperature hydrothermal vents at the ridge surfaces in the 1970´s. This thesis focuses on deep sea hydrothermal activity on a superfast spreading ridge, the SouthernEast Pacific Rise (SEPR).The ridge is located in the South Pacific, off the coast of South America, and separates the Nazca Plate and the Pacific Plate. A fossil high temperature hydrothermal zone hosted by a fault was sampled 80 m below the lava/dike transition zone in the Pito Deep (a tectonic window intothe SEPR). Geochemical data from the fault zone indicates that cold (<150°C)and hot (<390°) fluids coexisted at the same time whilst the hydrothermal system was active. A numerical model (HYDROTHERM) developed by the USGS was used to recreate the geological settings in the SEPR in order to try to model the hydrothermal activity and fluid flow. The model solves two governingpartial differential equations numerically, the water component flow equation(Darcy law for flow in porous media) and the thermal energy transport equation(conservation of enthalpy for the water component and the porous media). The result of the modeling indicates that cold seawater can penetrate from the relatively permeable volcanic material into a highly permeable fault zone in the sheeted dike unit. The cooler seawater fluid flows down the fault zone,reheats and flows up again in a narrow upflow zone at the edge of the fracture/sheeted dike boundary. The result is a horizontal temperature gradient created in the fractured zone supporting the theory that hot and cold fluids can coexist in a fault hosted hydrothermal zone.

Deep generative neural networks such as the Deep Belief Network and Deep Boltzmann Machines have been used successfully to model high dimensional visual data. However, they are not robust to common variations such as occlusion and random noise. In this thesis, we explore two strategies for improving the robustness of DBNs. First, we show that a DBN with sparse connections in the first layer is more robust to variations that are not in the training set. Second, we develop a probabilistic denoising algorithm to determine a subset of the hidden layer nodes to unclamp. We show that this can be applied to any feedforward network classifier with localized first layer connections. By utilizing the already available generative model for denoising prior to recognition, we show significantly better performance over the standard DBN implementations for various sources of noise on the standard and Variations MNIST databases.