>Magister Scientiae - MSc / Aim: To determine the prevalence and diameter of the maxillary end osseous canal which carries the anastomosis of the infra alveolar artery (a branch of the posterior superior alveolar artery) and the infra- orbital artery. Material and methods: Data was analyzed from one hundred archived cone beam computed tomography (CBCT) images. The presence of the end osseous anastomosis in the lateral sinus wall was identified by utilizing axial views. The vessel diameter was also measured in those images where the canal was identified. Results: The maxillary bony canal was identified in 49 (49%) of 100 maxillary sinus.14 (14%) presented on the right hand side, 10 (10%) presented on the left hand side, 25 (25%) had a bilateral presence with a remaining 51 (51%) which cannot be identified on the imaging. From the 49 canals that were identified, 5 canals had a diameter that was 2-3mm wide,19 canals had a diameter that was 1-2mm wide and the remaining 25 had a diameter that was less than 1mm. Conclusion: A sound knowledge of the maxillary sinus vascularity is essential as severe bleeding can occur due to damage of the intra-osseous branch during sinus augmentation procedures. CBCT analysis is required as a pre-requisite for the pre-planning stages during implant treatment to prevent complications such as haemorrhage, sinus perforations or associated vascular anomalies that may arise

<p> Designing self-regulating machines that can see and comprehend various real world objects around it are the main purpose of the AI domain. Recently, there has been marked advancements in the field of deep learning to create state-of-the-art DNNs for various CV applications. It is challenging to deploy these DNNs into resource-constrained micro-controller units as often they are quite memory intensive. Design Space Exploration is a technique which makes CNN/DNN memory efficient and more flexible to be deployed into resource-constrained hardware. MobileNet is small DNN architecture which was designed for embedded and mobile vision, but still researchers faced many challenges in deploying this model into resource limited real-time processors.</p><p> This thesis, proposes three new DNN architectures, which are developed using the Design Space Exploration technique. The state-of-the art MobileNet baseline architecture is used as foundation to propose these DNN architectures in this study. They are enhanced versions of the baseline MobileNet architecture. DSE techniques like data augmentation, architecture tuning, and architecture modification have been done to improve the baseline architecture. First, the Thin MobileNet architecture is proposed which uses more intricate block modules as compared to the baseline MobileNet. It is a compact, efficient and flexible architecture with good model accuracy. To get a more compact models, the KilobyteNet and the Ultra-thin MobileNet DNN architecture is proposed. Interesting techniques like channel depth alteration and hyperparameter tuning are introduced along-with some of the techniques used for designing the Thin MobileNet. All the models are trained and validated from scratch on the CIFAR-10 dataset. The experimental results (training and testing) can be visualized using the live accuracy and logloss graphs provided by the Liveloss package. The Ultra-thin MobileNet model is more balanced in terms of the model accuracy and model size out of the three and hence it is deployed into the NXP i.MX RT1060 embedded hardware unit for image classification application.</p>

<p>In the first chapter, we illustrate a big picture of the printing systems and the concentration of this dissertation. </p><p><br></p><p>In the second chapter, we present a tone-dependent fast error diffusion algorithm for color images, in which the quantizer is based on a simulated linearized printer space and the filter weight function depends on the ratio of the luminance of the current pixel to the maximum luminance value. The pixels are processed according to a serpentine scan instead of the classic raster scan. We compare the results of our algorithm to those achieved using</p> <p>the fixed Floyd-Steinberg weights and processing the image according to a raster scan ordering. In the third chapter, we first design a defect generator to generate the synthetic abnormal</p> <p>printer sounds, and then develop or explore three features for sound-based anomaly detection. In the fourth chapter, we explore six classifiers as our anomaly detection models, and explore or develop six augmentation methods to see whether or not an augmented dataset can improve the model performance. In the fifth chapter, we illustrate the data arrangement and the evaluation methods. Finally, we show the evaluation results based on</p> <p>different inputs, different features, and different classifiers.</p> <p><br></p><p>In the last chapter, we summarize the contributions of this dissertation.</p>

One approach to enabling a larger penetration of renewable sources of energy is the implementation of hybrid power plants. This work presents a process to determine the preliminary optimal configuration of a concentrating solar power-coal hybrid power plant with low solar augmentation, and is demonstrated on a coal power plant in Castle Dale, UT. A representative model is developed and validated against published data for a coal power plant of a different configuration than Hunter Unit 3. The simplifications within the representative model include combining multiple feedwater heaters, combining turbines that operate across the same boundary states, and the mass-average calculation for extraction properties to the combined feedwater heaters. It is shown that the representative model can accurately and consistently simulate a coal power plant. Comparing net power generation and boiler heating estimates from the representative model to the benchmark power plant, the representative model is accurate to within +/- 1% the accepted value from the benchmark power plant. The methods for quantifying solar resource with data from the National Renewable Energy Laboratory are presented with the derivation of an algorithm to simulate a concentrating solar power field arrangement. The solar contribution to electrical power output is estimated using an exergy balance. A simplified financial model is also developed to estimate the solar marginal levelized cost of electricity and payback time using a cash-flow analysis. Estimates for solar resource, solar contribution, and financial performance are consistent with data published by the National Renewable Energy Laboratory or in archival literature. A multi-objective optimization routine is developed consisting of the representative model, the augmentation of solar energy into the solar integration model by means of feedwater heater bypass, solar contribution, levelized cost of electricity, and payback time. Because this study considered complete FWH bypass, higher solar augmentation (>3% of boiler heating) is required for a hybrid design to be considered feasible. However, for higher solar augmentation, the costs are also considerably higher and the financial benefit is insufficient to make any hybrid designs feasible unless a carbon tax is in place. A carbon tax will amplify the financial benefit of hybridization, so optimization results are provided assuming a carbon tax value equivalent to the value used in California's Emissions Trading System (16 USD sh.tn.^-1). The impact of a green energy premium price paid by consumers is also explored in the context of payback time. The resulting optimal design for the Hunter Unit 3 with a carbon tax and no premium is using parabolic trough collector technology at an augment fraction of k=9% to bypass feedwater heater 6. The resulting marginal solar levelized cost of electricity is 9.5 x 10^-4 USD kWh^-1 with an estimated payback time of 25.2 years. This process can be applied to any coal power plant for which operating data and meteorological data are available to evaluate preliminary hybridization feasibility.

Thesis (MEng (Mechanical Engineering))--Cape Peninsula University of Technology, 2019 / The accumulating space debris has been a developing problem for many years. Technological advances led to the creation of nanosatellites, which allows more affordable access to space. As a result, the number of satellite launches is rapidly increasing, which, translates into an increase in debris in the low earth orbit (LEO) and geostationary orbit (GEO). To comply with the Inter-Agency Space Debris Coordination Committee (IADC) requirement of a 25-year maximum orbital lifetime, nanosatellites must have an end of life strategy. Failure to meet these guidelines may not only cause catastrophic collisions but may make future space travel even more challenging. Consequently, orbital lifetime predictions must be completed for nanosatellites. Considering this, the aim of this thesis is to investigate the orbital lifetime predictions for the nanosatellite ZACube-2, and the effects on the orbital lifetime if ZACube-2 is fitted with deorbiting technology, specifically a drag argumentation device. An in-depth literature review regarding the current state of technology pertaining to nanosatellite de-orbiting was conducted. This was followed by studies regarding orbital dynamics and perturbation forces. Four case studies were simulated in NASA’s Debris assessment software (DAS 2.0) using orbital parameters extracted from the two-line element (TLE) file. General information such as launch date and final mass was provided by F’SATI. The Baseline case study presented the orbital lifetime of ZACube-2, without any drag enhancement device. This was followed by case study 1,2 and 3 which represented ZACube-2 when fitted with three different drag enhancement devices. A comparison study indicated a reduction in all three cases. A new inflatable cube de-orbiting device (ICDD) concept was also presented, and the effects it has on the orbital lifetime predictions are showcased in case study three. Two deployment concepts were considered and evaluated against design requirements. Solidworks software was used to model the most suitable concept as well as perform finite element analysis on the structure. Static analysis was followed by natural frequency analysis in which the natural frequencies of the components and assembled structure were extracted. The Soyuz launch vehicle’s sinusoidal testing requirements were used to evaluate the structures survivability under dynamic loading. Based on the finite element , and harmonic analysis it was concluded that the structures will survive the launch conditions of the Soyuz launch vehicle. Furthermore, individual parameters affecting orbital lifetime predictions are also identified, in the form of a mass and cross-sectional sensitivity study and a ballistic coefficient versus orbital time study.

The first case of breast implant-associated anaplastic large cell lymphoma (breast implant ALCL) was described by John Keech and the late Brevator Creech in 1997. In the following 2 decades, much research has led to acceptance of breast implant ALCL as a specific clinicopathologic entity, a process that we bring up to life through the memories of 6 persons who were involved in this progress, although we acknowledge that many others also have contributed to the current state of the art of this disease. Dr. Keech recalls the events that led him and Creech to first report the disease. Ahmet Dogan and colleagues at the Mayo Clinic described a series of 4 patients with breast implant ALCL, and led to increased awareness of breast implant ALCL in the pathology community. Daphne de Jong and colleagues in the Netherlands were the first to provide epidemiologic evidence to support the association between breast implants and ALCL. Garry Brody was one of the first investigators to collect a large number of patients with the disease, present the spectrum of clinical findings, and alert the community of plastic surgeons. Roberto Miranda and L. Jeffrey Medeiros and colleagues studied the pathologic findings of a large number of cases of breast implant ALCL, and published the findings in 2 impactful studies in the medical oncology literature. The recognition and acceptance of this disease by surgeons, epidemiologists, and medical oncologists, working together, has led to subsequent studies on the pathogenesis and optimal therapy of this disease. / Revisión por pares

Indiana University-Purdue University Indianapolis (IUPUI) / Designing self-regulating machines that can see and comprehend various real world objects around it are the main purpose of the AI domain. Recently, there has been marked advancements in the field of deep learning to create state-of-the-art DNNs for various CV applications. It is challenging to deploy these DNNs into resource-constrained micro-controller units as often they are quite memory intensive. Design Space Exploration is a technique which makes CNN/DNN memory efficient and more flexible to be deployed into resource-constrained hardware. MobileNet is small DNN architecture which was designed for embedded and mobile vision, but still researchers faced many challenges in deploying this model into resource limited real-time processors. This thesis, proposes three new DNN architectures, which are developed using the Design Space Exploration technique. The state-of-the art MobileNet baseline architecture is used as foundation to propose these DNN architectures in this study. They are enhanced versions of the baseline MobileNet architecture. DSE techniques like data augmentation, architecture tuning, and architecture modification have been done to improve the baseline architecture. First, the Thin MobileNet architecture is proposed which uses more intricate block modules as compared to the baseline MobileNet. It is a compact, efficient and flexible architecture with good model accuracy. To get a more compact models, the KilobyteNet and the Ultra-thin MobileNet DNN architecture is proposed. Interesting techniques like channel depth alteration and hyperparameter tuning are introduced along-with some of the techniques used for designing the Thin MobileNet. All the models are trained and validated from scratch on the CIFAR-10 dataset. The experimental results (training and testing) can be visualized using the live accuracy and logloss graphs provided by the Liveloss package. The Ultra-thin MobileNet model is more balanced in terms of the model accuracy and model size out of the three and hence it is deployed into the NXP i.MX RT1060 embedded hardware unit for image classification application.

No description available.

Recent research shows that machine learning algorithms are highly susceptible to attacks trying to extract sensitive information about the data used in model training. These attacks called privacy attacks, exploit the model training process. Contemporary defense techniques make alterations to the training algorithm. Such defenses are computationally expensive, cause a noticeable privacy-utility tradeoff, and require control over the training process. This thesis presents a data-centric approach using data augmentations to mitigate privacy attacks. We present privacy-focused data augmentations to change the sensitive data submitted to the model trainer. Compared to traditional defenses, our method provides more control to the individual data owner to protect one's private data. The defense is model-agnostic and does not require the data owner to have any sort of control over the model training. Privacypreserving augmentations are implemented for two attacks namely membership inference and model inversion using two distinct techniques. While the proposed augmentations offer a better privacy-utility tradeoff on CIFAR-10 for membership inference, they reduce the reconstruction rate to ≤ 1% while reducing the classification accuracy by only 2% against model inversion attacks. This is the first attempt to defend model inversion and membership inference attacks using decentralized privacy protection. / Master of Science / Privacy attacks are threats posed to extract sensitive information about the data used to train machine learning models. As machine learning is used extensively for many applications, they have access to private information like financial records, medical history, etc depending on the application. It has been observed that machine learning models can leak the information they contain. As models tend to 'memorize' training data to some extent, even removing the data from the training set cannot prevent privacy leakage. As a result, the research community has focused its attention on developing defense techniques to prevent this information leakage. However, the existing defenses rely heavily on making alterations to the way a machine learning model is trained. This approach is termed as a model-centric approach wherein the model owner is responsible to make changes to the model algorithm to preserve data privacy. By doing this, the model performance is degraded while upholding data privacy. Our work introduces the first data-centric defense which provides the tools to protect the data to the data owner. We demonstrate the effectiveness of the proposed defense in providing protection while ensuring that the model performance is maintained to a great extent.

No description available.