BACKGROUND: Though direct oral anticoagulants (DOACs) have become a standard of care in the treatment of acute deep vein thrombosis (DVT), it is our observation that physicians tend to initiate heparin or low-molecular-weight heparin, hereafter called “heparin”, for the treatment of extensive DVT or phlegmasia cerulea dolens (PCD). This might be due to a perception that heparin might relieve DVT-related symptoms more quickly than DOACs. Whether these assumptions are true has not been evaluated. METHODS: We conducted a survey of thrombosis specialists in North America to explore the practical management of anticoagulant therapy in patients with extensive DVT, and the underlying reasons for the selection of heparin over DOACs. A cross-sectional, web-based survey was distributed to thrombosis specialists who are members of four thrombosis societies. RESULTS: Eighty-nine respondents provided consent. Most respondents selected DOACs over heparin in a case scenario representing mild DVT-related symptoms and limited thrombus involvement (81% vs. 19%). Most respondents selected heparin over DOACs in a case scenario representing early stage PCD (84% vs.16.3%) or a patient with high bodyweight (72% vs. 28%). In a case scenario representing extensive DVT, 57.4% of the respondents selected heparin, whereas, 42.6% selected DOACs. In the respondents who selected heparin over DOACs, the major reason was that heparin might relieve DVT-related symptoms more quickly because of its anti-inflammatory effects. DISCUSSION: Severity of DVT-related symptoms, thrombus extent, and bodyweight play a role in the selection of anticoagulant therapy. Despite a lack of evidence to support the hypothesis with respect to which anticoagulant is superior, most thrombosis specialists selected heparin over DOACs in patients with severe DVT-related symptoms and extensive thrombus involvement. Observation of variations in the selection of anticoagulant therapy for the treatment of extensive DVT also indicates that clinical trials in this patient population are needed. / Thesis / Master of Science (MSc)

With the recent advancement in deep learning, there has been increasing interest to apply deep learning algorithms to mobile edge devices (e.g. wireless access points, mobile phones, and self-driving vehicles). Such devices are closer to end-users and data sources compared to cloud data centers, therefore deep learning on the edge leads to several merits: 1) reduce communication overhead (e.g. latency), 2) preserve data privacy (e.g. not leaking sensitive information to cloud service providers), and 3) promote autonomy without the need of continuous network connectivity. However, it also comes with a trade-off that deep learning on the edge often results in less prediction accuracy or longer inference time. How to optimize such a trade-off has drawn a lot of attention among the machine learning and systems research communities. Those communities have explored three main directions: partitioning, caching, and compression to solve the problem. Deep learning model partitioning works in distributed and parallel computing by leveraging computation units (e.g. edge nodes and end devices) of different capabilities to achieve the best of both worlds (accuracy and latency), but the inference time of partitioning is nevertheless lower bounded by the smallest of inference times on edge nodes (or end devices). In contrast, model caching is not limited by such a lower bound. There are two trends of studies in caching, 1) caching the prediction results on the edge node or end device, and 2) caching a partition or less complex model on the edge node or end device. Caching the prediction results usually compromises accuracy, since a mapping function (e.g. a hash function) from the inputs to the cached results often cannot match a complex function given by a full-size neural network. On the other hand, caching a model's partition does not sacrifice accuracy, if we employ a proper partition selection policy. Model compression reduces deep learning model size by e.g. pruning neural network edges or quantizing network parameters. A reduced model has a smaller size and fewer operations to compute on the edge nodes or end device. However, compression usually sacrifices prediction accuracy in exchange for shorter inference time. In this thesis, our contributions to partitioning, caching, and compression are covered with experiments on state-of-the-art deep learning models. In partitioning, we propose TeamNet based on competitive and selective learning schemes. Experiments using MNIST and CIFAR-10 datasets show that on Raspberry Pi and Jetson TX2 (with TensorFlow), TeamNet shortens neural network inference as much as 53% without compromising predictive accuracy. In caching, we propose CacheNet, which caches low-complexity models on end devices and high-complexity (or full) models on edge or cloud servers. Experiments using CIFAR-10 and FVG have shown on Raspberry Pi, Jetson Nano, and Jetson TX2 (with TensorFlow Lite and NCNN), CacheNet is 58-217% faster than baseline approaches that run inference tasks on end devices or edge servers alone. In compression, we propose the logographic subword model for compression in machine translation. Experiments demonstrate that in the tasks of English-Chinese/Chinese-English translation, logographic subword model reduces training and inference time by 11-77% with Theano and Torch. We demonstrate our approaches are promising for applying deep learning models on the mobile edge. / Thesis / Doctor of Philosophy (PhD) / Edge artificial intelligence (EI) has attracted much attention in recent years. EI is a new computing paradigm where artificial intelligence (e.g. deep learning) algorithms are distributed among edge nodes and end devices of computer networks. There are many merits in EI such as shorter latency, better privacy, and autonomy. These advantages motivate us to contribute to EI by developing intelligent solutions including partitioning, caching, and compression.

Deep-marine levees are areally extensive features that border submarine channel systems. Compared to the adjacent channel, where episodes of erosion and bypass are commonplace, levees are mostly depositional features that experience little erosion, and therefore high preservation potential of individual beds, and presumably provide a nearly continuous depositional record of transport events down deep-marine slopes. Nevertheless, despite their size, volumetric prominence, and interpretive significance, deep-marine levees have received much less research attention compared to the adjacent channels. Accordingly, the spatial and temporal evolution of levee stratigraphy is much less well understood, in part because of the typically recessive nature of levee deposits exposed in outcrop in the ancient sedimentary record, and insufficient seismic resolution seismic in the modern. Also, although modern deep-marine levees have been shown to sequester a large proportion of the world’s total buried organic carbon, few studies have attempted to assess carbon deposition and preservation in ancient deep-marine levee deposits. In the Isaac Formation of the Windermere Supergroup (Neoproterozoic) of east-central British Columbia, Canada, well-exposed levee deposits display a systematic organization on several dimensional scales. Levee packages (decameter-scale) are interpreted to be due to cyclic changes in the granulometric makeup of sediment being supplied to the system, whereas bedsets (centimeter- to meter-scale) are interpreted to represent systematic and recurring pulses or surges during a single flow event. Furthermore, physical and geochemical characterization of levee strata at Castle Creek has shown that the unique depositional processes in levees can result in the concentration and enrichment of sedimentary marine organic matter (OM), which occurs mostly in banded, mud-rich sandstones deposited under oxic conditions. Organic carbon occurs primarily as nano-scale coatings on clay particles and uncommon sand-sized organomineralic aggregates and discrete sand-sized amorphous grains. The distribution of this OM in levee strata is controlled by a combination of primary productivity, sea level, and rates of continental runoff and detrital terrigenous influx, which collectively are principally controlled by climate. Understanding the stacking patterns, geochemistry, and organic content of ancient levee deposits is important for assessing sedimentation patterns, depositional processes, event frequency and magnitude, paleoenvironmental conditions, and the evolution of ancient ocean and climate systems.

Driver Assistance and Autonomous Driving features are becoming nearly ubiquitous in new vehicles. The intent of the Driver Assistant features is to assist the driver in making safer decisions. The intent of Autonomous Driving features is to execute vehicle maneuvers, without human intervention, in a safe manner. The overall goal of Driver Assistance and Autonomous Driving features is to reduce accidents, injuries, and deaths with a comforting driving experience. However, different drivers can react differently to advanced automated driving technology. It is therefore important to consider and improve the adaptability of these advances based on driver behavior. In this thesis, a human-centric approach is adopted in order to provide an enriching driving experience. The thesis investigates the natural behavior of drivers when changing lanes in terms of preferences of vehicle kinematics parameters using a real-world driving dataset collected as part of the Second Strategic Highway Research Program (SHRP2). The SHRP2 Naturalistic Driving Study (NDS) set is mined for lane change events. This work develops a way to detect reliable lane changing instances from a huge NDS dataset with more than 5,400,000 data files. The lane changing instances are distinguished from noisy and erroneous data by using machine vision lane tracking system variables such as left lane marker probability and right lane marker probability. We have shown that detected lane changing instances can be validated using only vehicle kinematics data. Kinematic vehicle parameters such as vehicle speed, lateral displacement, lateral acceleration, steering wheel angle, and lane change duration are then extracted and examined from time series data to characterize these lane-changing instances for a given driver. We have shown how these vehicle kinematic parameters change and exhibit patterns during lane change maneuvers for a specific driver. The thesis shows the limitations of analyzing vehicle kinematic parameters separately and develops a novel metric, Lane Change Dynamic Score(LCDS) that shows the collective effect of these vehicle kinematic parameters. LCDS is used to classify each lane change and thereby different driving styles. / Master of Science / The current tendency of car manufacturers is to create vehicles that will offer the user the most comfortable ride possible. The user experience is given a lot of attention to ensure it is up to par. With technological advancements, we are moving closer to an era in which automobiles perform many functions autonomously. However, different drivers may react differently to highly automated driving technologies. Therefore, adapting to different driving styles is critical to increasing the acceptance of autonomous vehicle features. In this work, we examine one of the stressful maneuvers of lane changes. The analysis of various drivers' lane-changing behaviors and the value of personalization are the main subjects of this study based on actual driving scenarios. To achieve this, we have provided an algorithm to identify occurrences of lane-changing from real driving trip data files. Following that, we investigated parameters such as lane change duration, vehicle speed, displacement, acceleration, and steering wheel angle when changing lanes. We have demonstrated the patterns and changes in these vehicle kinematic characteristics that occur when a particular driver performs lane change operations. The thesis shows the limitations of analyzing vehicle kinematic parameters separately and develops a novel metric, Lane Change Dynamic Score(LCDS) that shows the collective effect of these vehicle kinematic parameters. LCDS is used to classify each lane change and thereby different driving styles.

Efficient exploitation and conservation of the oceans poses great technological challenges for scientists and engineers who must develop materials, structures and equipment for use in harsh environment of the oceans. For the applications of materials in marine environment, knowledge of the corrosion properties is essential for selection purposes. Presently, effort is being devoted to exploit deep-sea mineral resources. Deterioration of materials in the deep sea is due to the cumulative effect hydrostatic pressure, temperature, pH, dissolved oxygen, salinity and sea current. For the first time, in-situ corrosion measurements on the effect of deep sea environment on some metallic and composite materials were carried out at depths of 500,1200,3500, and 5100 m for 168,174 and 174 days of exposure in the Indian Ocean. Corrosion rate was obtained from weight loss measurements (mm/year) and surface morphology of as-exposed and cleaned specimens of the above materials was studied under scanning electron microscope and ED AX. Galvanic coupling of steel with zinc, magnesium and aluminium were also studied.. Tensile on metal and alloys and tensile, compressive, flexure and ILSS tests on carbon fibre reinforced composite specimen were performed on exposed specimens. XRD studies were conducted on the corrosion product of materials. In order to correlate the performance of materials in deep-sea environment, seawater current and temperature data were also collected at same period Results reveal that the corrosion behaviour of steels is controlled by dissolved oxygen prevailing and corrosion rate corresponds to dissolved oxygen available at these depth levels. This is due to the fact that oxygen acts as a cathodic deploarizer during corrosion reaction of steels in seawater. Corrosion rate of aluminium increases as the depth increases. This is due to the effect of hydrostatic pressure, which reduces the ionic radii of chlorine ions and facilitates easy penetration of these ions into surface layer. Titanium, titanium alloy (Ti-6A1-4V) and stainless steels did not show any deterioration at all depths studied. Morphology of as exposed and corroded coupons reveal different features. EDS analyses on exposed specimens are analyzed in light of seawater parameters. Carbon fibre reinforced composite did not show any change in properties like tensile, compression flexural and ILSS compared to control (unexposed) specimens. The deposition of calcium carbonate on galvanically coupled mild steel with zinc, aluminium and magnesium corresponds to availability of calcium in the deep ocean. EDS analyses on exposed coupons did not reveal calcium element below the calcium carbonate compensation depth (CCD) at 3800 m in Indian Ocean. Potentiodynamic polarization studies on some metals and alloys indicate that the behaviour of materials in deep-sea environment is a cumulative effect of all oceanographic parameters. Tensile test results on stainless steels SS-304 & SS-316L), titanium and titanium alloy (exposed) specimens did not show any significant change in their tensile properties and is again attributed to the passive film formed on its surface and nearly zero corrosion rate observed. Microbiological investigations on the exposed materials indicate that except carbon fibre reinforced composite all other metals and alloys harboured bacterial colonies. Results have been used to recommend structural materials suitable for the deep-sea applications.

No description available.

The diploma thesis present technology production of draining outlet. This is a rectangular deep drawn part, where basic shape is formed in two towing operations now. The used material is an austenitic stainless steel 17 240 with thickness 1 mm. According to the entry documentation the production technology was not change – current technology (it means drawing without thinning the wall) is most suitable. Based on technological and control calculations two-operations production was determined, which will take place in the united conventional instrument. The hydraulic press with marking Dieffenbacher PO250, which has a magnetic tool clamping, per stroke will perform here both operations. The payback was calculated for yearlong production 50 000 pieces after 0,8 years. The proposed production will make the current production more efficient including economic benefits.

The project looks for a solution to optimalization of rotational deep drawn part with flange production. It deals with sewerage drain, which is produced by conventional tool to two deep drawing operations. The used material is austenitic Cr-Ni stainless steel 1.4301. Suitability of this material for deep drawing operations was considered by means of the tensile test and the Erichsen test. On the bases of these results, serializability of production and theoretic relationships for determination of number of deep drawing operations, it was decided upon one operation serial production, by conventional tool without attenuation thickness of sheet. Production will take place on hydraulic press Dieffenbacher PO250 with magnetic tool fastening. The deep drawing tool was designed and constructed. For series of 2 000 units per year the rate of return was calculated after 4,7 years, while using the punch from the current second deep drawing operation it was calculated after 3,4 years.

No description available.

Deep tectonic tremor has been observed in major subduction zones, strike-slip faults, inland faulting systems, and arc-continent collision environments around the Pacific Rim. However, detailed space-time evolution of its source locations remains enigmatic because of difficulties in detecting and locating tremor accurately. In 2011, we installed two dense, small-aperture seismic arrays aiming to detect ambient tremor source beneath southern Central Range in Taiwan. We recorded continuous waveforms for a total of 134 days, including tremor triggered by the great 2011 Mw9.0 Tohoku earthquake. We use the broadband frequency-wavenumber beamforming and the moving-window grid-search methods to compute array parameters for detecting seismic signals. The obtained array parameters closely match both relocated local earthquakes and triggered tremor bursts located by an envelope cross-correlations method, indicating the robustness of our array technique. We identify tremor signals with coherent waveforms and deep incidence angles and detect tremor for 44 days among the 134-day study period. The total duration is 1,481-minute, which is 3-6 times more than that detected by the envelope cross-correlations method. In some cases, we observe rapid tremor migration with a speed at the order of 40-50 km/hour that is similar to the speed of fast tremor migration along-dip on narrow streaks in Japan and Cascadia. Our results suggest that dense array techniques are capable of capturing detailed spatiotemporal evolutions of tremor behaviors in southern Taiwan.