<p>In this dissertation, we document a few of our recent attempts in bridging the gap between the fast evolving deep learning research and the vast industry needs for dealing with computer vision challenges. More specifically, we developed novel deep-learning-based techniques for the following application-driven computer vision challenges: image super-resolution with quality restoration, motion estimation by optical flow, object detection for shape reconstruction, and object segmentation for motion tracking. Those four topics cover the computer vision hierarchy from the low level where digital images are processed to restore missing information for better human perception, to middle level where certain objects of interest are recognized and their motions are analyzed, finally to high level where the scene captured in the video footage will be interpreted for further analysis. In the process of building the whole-package of  ready-to-deploy solutions, we center our efforts on designing and training the most suitable convolutional neural networks for the particular computer vision problem at hand. Complementary procedures for data collection, data annotation,  post-processing of network outputs tailored for specific application needs, and deployment details will also be discussed where necessary. We hope our work demonstrates the applicability and versatility of convolutional neural networks for real-world computer vision tasks on a broad spectrum, from seeing better to understanding better.</p>

The Super Proton Synchrotron (SPS) injection system plays a fundamental role to preserve the quality of injected high-brightness beams for the Large Hadron Collider (LHC) physics program and to maintain the maximum storable intensity. The present system is the result of years of upgrades and patches of a system not conceived for such intensities and beam qualities. In this study, we first investigate the effect of emittance growth due to amplitude-dependent tune shifts for erroneously injected beams. As a next step, we propose the design of a completely new injection system for the SPS using multi-level numerical optimisation, including realistic hardware assumptions. Methods and pseudo-algorithms of how this hierarchical optimisation framework can be adapted to other situations for optimal accelerator system design are shown. In addition, we explore the benefits of a numerical optimisation framework for the current SPS injection kicker timing system to minimise residual injection oscillations for maximised delivered beam intensity. We also demonstrate how a simple neural network based upon recorded data can approximate the injection system as a surrogate model, allowing for further studies of different optimisation algorithms even without beam time.

In this thesis, a High Efficiency (IE2), 7.5 kW, 4-pole, induction motor (IM) is studied and redesigned using finite element method (FEM) software with the purpose of increasing its efficiency to the new Super Premium Efficiency (IE4) level. First, a pre-existing model of the motor is analyzed and improved. Then, different methods to improve the efficiency of the motor are investigated and the performance of the machine is studied after each modification. Thirdly, a complete optimization of the design is carried out with the goal of developing an IE4 IM compatible with IEC standard frames. Finally, a similar optimization of the motor is performed substituting the aluminum squirrel cage by a copper one. The pros and cons of this change are studied. The results show the feasibility of reaching the IE4 efficiency level by this type of IM with a standard shaft height. Preserving the original cross-section design of the stator, this efficiency level is only reached if a copper squirrel cage is used and if the rotor cross-section is redesigned. However, if the external diameter of the stator is increased and the rotor and stator cross-sections are redesigned, aluminum rotor bars and short-circuit rings can be used to reach the IE4 efficiency level. / I detta examensarbete analyseras en 7.5 kW 4-pol induktionsmotor med verkningsgradsklass IE2 med hjalp av Finita Element Metoden (FEM). Syftet med arbetet ar att utvardera olika metoder for att oka motorns verkningsgrad och att foresla designforandringar som kan mojliggora en uppgradering till verkningsgradsklass IE4. Forst analyseras och justeras en befintlig modell av motorn som sedan anvands for att utvardera effekterna av olika designforandringar. Sedan optimeras motorn for att oka verkningsgraden, detta gors bade med aluminium och med koppar som rotorledarmaterial. Resultaten visar att det ar mojligt att uppna verkningsgradsklassen IE4 genom att anvanda antingen aluminium eller koppar som rotorledarmaterial. For bada fallen kravs att motorns langd okas. Da koppar anvands kravs endast en ny tvarsnittsgeometri for rotorn medan da aluminium anvands behover bade rotor och statorgeometri andras.

While digital displays have continuously increased in resolution, video content produced before these improvements is however stuck at its original resolution, and the use of some form of scaling is needed for a satisfactory viewing experience on high-resolution displays. In recent years, the field of video scaling has taken a leap forward in output quality, due to the adoption of deep learning methods in research. In this paper, we describe a study wherein we train a convolutional neural network for super-resolution, and conduct a large-scale A/B video quality test in order to investigate if SVT video-ondemand viewers prefer video upscaled using a convolutional neural network to video upscaled using the standard bicubic method. Our results show that viewers generally prefer CNNscaled video, but not necessarily for the types of content this technology would primarily be used to scale. We conclude that the technology of deep upscaling shows promise, but also believe that more optimization and flexibility is need for deep scaling to be viable for mainstream use. / Allteftersom bildskärmstekniken förbättras så får mediekonsumenter tillgång till skärmar med allt högre upplösningar; dock är videomaterial som producerats för en viss bildupplösning, fast på denna nivå, och någon form av skalning måste användas för en bra tittarupplevelse på högupplösta skärmar. På senare tid så har videoskalning förändrats, tack vare användandet av djupinlärningsmetoder inom forskningen. I den här rapporten beskriver vi en studie där vi tränade en djup modell för videouppskalning, och sedan utförde ett storskaligt A/B-test, med syftet att undersöka huruvida SVTs onlinetittare föredrar video skalad med djupinlärning över video skalad med konventionella metoder. Våra resultat visar att tittarna föredrog video skalad med djupinlärning, dock inte nödvändigtvis för det material tekniken främst skulle användas med. Vi drar slutsatsen att videoskalning med hjälp av djupinlärning är lovande, men anser också att mer optimering och flexibilitet behövs innan tekniken kan anses mogen för bred adoption.

Videoaufnahmen aus den vergangenen Jahrzehnten stellen kulturelles Erbe dar. Diese sind jedoch nach heutigen Sehgewohnheiten nicht ohne große Einschränkungen für die Wiederausstrahlung geeignet. Das liegt zum einen an längst vergangenen Standards der Videoaufzeichnung, aber zum anderen auch in großem Maße an unkontrolliert gealterten Speichermedien durch inadäquate Aufbewahrung. Oftmals war es lokalen Fernsehsendern technisch nicht möglich ihre Archivbestände unter optimalen klimatischen Bedingungen langlebig zu lagern. Videoarchivdaten nach der Digitalisierung für die Einbindung in heutige Produktionen manuell zu durchsuchen und entsprechend aufzubereiten, ist ein zeitaufwändiger Prozess, den lokale TV-Sender nicht bewältigen können. In unserem Beitrag möchten wir neuartige Methoden der automatisierten Aufbereitung von archivierten VHS-Digitalisaten für die Wiederausstrahlung vorstellen. Dazu zählen vor allem Verfahren zu den Schwerpunkten der Korrektur von Falschfarben (Recoloring) und zur Steigerung der Auflösung von ehemals PAL zu Full-HD und Ultra-HD (Super-Resolution). Zum Einsatz kommen dabei künstliche neuronale Netze, die anders als klassische Verfahren der Bildverarbeitung, semantische Bildkomponenten erfassen und bei der Bearbeitung berücksichtigen können. Mitunter können so deutliche Qualitätsverbesserungen erzielt werden. In unserem Beitrag möchten wir auf Chancen und aktuelle Beschränkungen dieser Technologien eingehen und anhand von digitalisierten Videoarchivdaten deren Funktionsweise demonstrieren.

<p>Applications in materials and biological imaging are currently limited by the ability to collect high-resolution data over large areas in practical amounts of time. One possible solution to this problem is to collect low-resolution data and apply a super-resolution interpolation algorithm to produce a high-resolution image. However, state-of-the-art super-resolution algorithms are typically designed for natural images, require aligned pairing of high and low-resolution training data for optimal performance, and do not directly incorporate a data-fidelity mechanism.</p><p><br></p><p>We present a Multi-Resolution Data Fusion (MDF) algorithm for accurate interpolation of low-resolution SEM and TEM data by factors of 4x and 8x. This MDF interpolation algorithm achieves these high rates of interpolation by first learning an accurate prior model denoiser for the TEM sample from small quantities of unpaired high-resolution data and then balancing this learned denoiser with a novel mismatched proximal map that maintains fidelity to measured data. The method is based on Multi-Agent Consensus Equilibrium (MACE), a generalization of the Plug-and-Play method, and allows for interpolation at arbitrary resolutions without retraining. We present electron microscopy results at 4x and 8x super resolution that exhibit reduced artifacts relative to existing methods while maintaining fidelity to acquired data and accurately resolving sub-pixel-scale features.</p>

The task of upscaling pictures is very ill-posed since it requires the creation of novel data. Any algorithm or model trying to perform this task will have to interpolate and guess the missing pixels in the pictures. Classical algorithms usually result in blurred or pixelated interpolations, especially visible around sharp edges. The reason it could be considered a good idea to use neural networks to upscale pictures is because they can infer context when upsampling different parts of an image. In this report, a special deep learning structure called U-Net is trained on reconstructing high-resolution images from the Div2k dataset. Multiple loss functions are tested and a combination of a GAN-based loss function, simple pixel loss and also a Sobel-based edge loss was used to get the best results. The proposed model scored a PSNR score of 33.11dB compared to Lanczos 30.23dB, one of the best classical algorithms, on the validation dataset.

<p>Narcissism is a personality construct linked to dysfunction in several domains. It encompasses grandiose and vulnerable variants as well as antagonism, agentic extraversion, and neuroticism higher-order factors. Many measures that vary in breadth and length have been constructed to measure narcissism. In recent years, super-short forms have become popular in research settings. Although brief measures hold some advantages, their brevity can come at psychometric costs. The comparative limitations of these short narcissism forms have received relatively little empirical examination. The goal of the current project was to fill this gap by determining the potential costs and benefits of using short measures of narcissism rather than longer measures in an online community sample (<em>N</em>= 473). This examination included assessing short form completion time, psychometric properties, structure, and measurement invariance across gender. Generally, the short forms demonstrated adequate internal consistency, variable convergence with each other, and mostly moderate to strong convergence with long forms. Short forms with long form counterparts performed well in terms of accounting for the variance of their long form counterparts. The short form items used for the bass-ackward analysis successfully replicated the factor structure of narcissism found by Crowe et al. (2019) using longer narcissism measures at both the two- and three-factor level, which showed measurement invariance across gender, generally at the scalar invariance level. Taken together, these findings suggest that it is still likely most advantageous to use the long forms whenever possible but that some short forms could be used when efficiency of survey administration is particularly important without significant psychometric cost.</p>

Several avenues of investigation through which the 'labyrinths' of schizopsychotic diagnosis may be examined, are offered by the consideration of the 'beacons' of symptom-profiles and biomarkers. Neurodevelopmental issues and risk assessment, neurocognitive factors of predictive necessity, supersensitivity in neurotransmitter systems, the implications of prodromal expressions of the disorder, functional dysconnectivity arising from prefrontal to diverse regional patterns and circuits with a neurodevelopmental origin, and heritable gene characteristics are viewed against the backdrop of the schizophrenia spectrum disorders. The associations between adolescent-adult use of cannabis, on the one hand, and, alternatively, the prevalence of chromosomal abnormalities, e.g., GRIK4 and NPAS3, and mental retardation, on the other hand, with the symptom-profiles of schizopsychosis provide further evidence of emerging biomarkers of biological inheritance factors. The involvement of dopamine D1 and D2 receptors, particularly in prefrontal region, with regard to functional integrity of cognitive systems is reviewed. It would appear that considerations of these disorders imply that one essential hub around which much of the neuropathology revolves may be observed in the various expressions of the cognitive and structural insufficiency.

Companies that advertise during the Super Bowl can reach 40 million U.S. households with a 30-second commercial spot, but the cost can exceed $2 million. This research examines Nielsen television ratings and expenses for related commercial spots and suggests that the Super Bowl is not always the best site for introducing new companies or products to the marketplace. ANOVA test results indicate that younger companies may better affect purchase decisions by advertising more frequently during less expensive programming slots.