One of the state-of-the-art highlights of deep learning in the past ten years is the introduction of generative adversarial networks (GANs), which had achieved great success in their ability to generate images comparable to real photos with minimum human intervention. These networks can generalise to a multitude of desired outputs, especially in image-to-image problems and image syntheses. This thesis proposes a computer graphics pipeline for 3D rendering by utilising generative adversarial networks (GANs). This thesis is motivated by regression models and convolutional neural networks (ConvNets) such as U-Net architectures, which can be directed to generate realistic global illumination effects, by using a semi-supervised GANs model (Pix2pix) that is comprised of PatchGAN and conditional GAN which is then accompanied by a U-Net structure. Pix2pix had been chosen for this thesis for its ability for training as well as the quality of the output images. It is also different from other forms of GANs by utilising colour labels, which enables further control and consistency of the geometries that comprises the output image. The series of experiments were carried out with laboratory created image sets, to pursue the possibility of which deep learning and generative adversarial networks can lend a hand to enhance the pipeline and speed up the 3D rendering process. First, ConvNet is applied in combination with Support Vector Machine (SVM) in order to pair 3D objects with their corresponding shadows, which can be applied in Augmenter Reality (AR) scenarios. Second, a GANs approach is presented to generate shadows for non-shadowed 3D models, which can also be beneficial in AR scenarios. Third, the possibility of generating high quality renders of image sequences from low polygon density 3D models using GANs. Finally, the possibility to enhance visual coherence of the output image sequences of GAN by utilising multi-colour labels. The results of the adopted GANs model were able to generate realistic outputs comparable to the lab generated 3D rendered ground-truth and control group output images with plausible scores on PSNR and SSIM similarity index metrices.

This research paper focuses on fraud detection in the financial industry using Generative Adversarial Networks (GANs) in conjunction with Uni and Multi Variate Bayesian Model with Shrinkage Priors (BMSP). The problem addressed is the need for accurate and advanced fraud detection techniques due to the increasing sophistication of fraudulent activities. The methodology involves the implementation of GANs and the application of BMSP for variable selection to generate synthetic fraud samples for fraud detection using the augmented dataset. Experimental results demonstrate the effectiveness of the BMSP GAN approach in detecting fraud with improved performance compared to other methods. The conclusions drawn highlight the potential of GANs and BMSP for enhancing fraud detection capabilities and suggest future research directions for further improvements in the field.

Thesis (PhD)--Stellenbosch University, 2012. / ENGLISH ABSTRACT: Few people realize exactly how complicated the translation of children’s poetry is. Translators do not only have to adhere to the young readers’ desires and satisfy the adult critics, but are constantly confronted with choices concerning the translation of the ‘play-element’ (structure) and the ‘visual element’ (content) of children’s verses, i.e. the translation of cultural elements, figurative language, pun, nonce words, onomatopoeia, alliteration, rhyme and meter. In addition, their translation strategies are continually subjective to and restricted by the visual text (illustrations) in the source texts, which interrelate with the verbal text (verses). In this descriptive, systematic analysis the Afrikaans and French translations of Mother Goose’s nursery rhymes, Dr. Seuss’s rhyming picture books and Roald Dahl’s verse fragments are compared to one another and the source texts in order to identify the various translation strategies and theoretical translation approaches used by the various Afrikaans and French translators, to make concrete observations about the translation of children’s poetry that can be useful for translators, and to establish if it is indeed possible to create translations of children’s verses that remain true to the ‘spirit’ of the original poetic texts, can function as autonomous texts in the target system, and that can supplement the Afrikaans and French children’s literature systems. __________________________________________________________________________ / AFRIKAANSE OPSOMMING: Min mense besef hóé ingewikkeld die vertaling van kinderverse eintlik is. Vertalers moet nie net tegelykertyd aan jong lesers se behoeftes voldoen en volwasse kritici tevrede stel nie, maar word ook deurgaans gekonfronteer met keuses wat betref die vertaling van die spel-element (struktuur) en visuele element (inhoud) van kinderverse, o.a. die vertaling van kultuurgebonde verwysings, beeldspraak, woordspel, neologisme, onomatopee, alliterasie, rym en metrum. Daarbenewens word die vertalers se vertaalstrategieë beïnvloed en beperk omdat die visuele teks (illustrasies) in die brontekste deurgaans met die verbale teks (verse) in gesprek tree. In dié deskriptiewe sisteemondersoek word die Afrikaanse en Franse vertalings van Moeder Gans se kinderrympies, Dr. Seuss se versverhale en Roald Dahl se prosimetriese kinderstories met mekaar en die brontekste vergelyk om die verskillende vertaalstrategieë en teoreties gefundeerde vertaalbenaderings wat deur die onderskeie Afrikaanse en Franse vertalers toegepas is, te identifiseer, konkrete bevindinge oor die vertaling van die kindervers te maak wat vir toekomstige vertalers van praktiese nut kan wees, en te bepaal of dit inderdaad moontlik is om vertalings van kinderverse te skep wat getrou bly aan die ‘gees’ van die oorspronklike gedigtekste, as selfstandige tekste in die doelsisteem kan funksioneer, en die Afrikaanse en Franse kinder- en jeugliteratuursisteme kan aanvul.

Les particules de suie, issues de la combustion incomplète, peuvent, en fonction des conditions de combustion, contenir une part plus ou moins importante de composés organiques (OC/TC). Par ailleurs, dès lors que ces nanoparticules sont émises dans l'atmosphère, des composés organiques volatiles peuvent s'adsorber, formant une gangue autour de ces agrégats fractals. L'impact de cette composition initiale ou de ce « vieillissement atmosphérique » sur les propriétés morphologiques et radiatives de ces particules n'est pas bien connu. Ceci a un impact sur les modèles radiatifs climatiques mais aussi sur l'interprétation des signaux délivrés par les différents diagnostics optiques pouvant être utilisés pour la métrologie des aérosols. En particulier, ce travail vise à juger de la pertinence de l'usage de diagnostics optiques pour caractériser les particules de suie en conditions atmosphériques. Afin d'étudier l'impact des composés organiques initialement présents dans la particule ou adsorbés en post-combustion sur leurs propriétés radiatives, nous avons étudié en laboratoire des suies produites par une flamme de diffusion (miniCAST) pour différentes richesses globales et avons ajouté un revêtement organique d'acide. Dans le but de générer en laboratoire une couche d'acide oléique sur des particules de référence, un dispositif de « coating » a été mis en œuvre et qualifié. Les particules ainsi générées et recouvertes ou non, ont été caractérisées en masse (mesures TEOM), en taille (mesures SMPS) et morphologiquement (densité effective). L'épaisseur de coating ainsi que la restructuration morphologique causée par l'ajout d'une gangue de matière organique ont ainsi été quantifiées. Les propriétés radiatives, ont été mesurées par extinction spectrale (Turbidimétrie) et diffusion (diffusion angulaire et spectrale). Un effort particulier a été mené pour que des mesures expérimentales puissent valider des résultats de calculs numériques préexistants. Par ailleurs, ces différentes techniques de mesures (optiques et non optique) ont conduit à la généralisation de la théorie Rayleigh Debye Gans for Fractal Aggregates (RDG-FA) à des particules de type agrégats fractals polydispersés recouverts d'un revêtement organique (RDG-CFA). Ceci permettant d'appréhender de façon phénoménologique l'impact du coating sur les propriétés radiatives et d'entrevoir une implantation plus aisée dans les codes de simulation climatique ou pour l'interprétation des mesures optiques dans l'atmosphère. Enfin, une attention particulière a été portée sur la technique d'incandescence induite par laser (LII) afin d'étudier la faisabilité de l'application de cette technique aux particules organiques ou ayant interagi avec les composés atmosphériques au cours de leur vieillissement. / Soot particles resulting from incomplete combustion may contain a more or less important part of organic compounds (OC / TC), depending on the combustion conditions. Moreover, once these nanoparticles are emitted into the atmosphere, volatile organic compounds can adsorb, forming a coating around these fractal aggregates. The impact of the initial composition or the atmospheric aging on the morphological and radiative properties of these particles is not well known. This has an impact on the radiative climate models but also on the interpretation of the signais delivered by the different optical diagnostics that can be used for aerosol metrology. In particular, this work aims to evaluate the relevance of the use of optical diagnostics to characterize soot particles in atmospheric conditions. In order to study the impact of organic compounds initially present in the particle or adsorbed in post-combustion on their radiative properties, we have studied soot produced by a diffusion flame (miniCAST) for different global richnesses and added an organic acid coating. In order to produce an oleic acid layer on reference particles, a coating device has been implemented and qualified. Particles thus generated, coated or not, were characterized in mass (TEOM measurements), in size (SMPS measurements) and morphologically (effective density). The coating thickness as well as the morphological restructuring caused by the addition of an organic coating was thus quantified. The radiative properties were measured by spectrally resolved light extinction and scattering (angular and spectrally resolved). A special effort was made to allow experimental measurements to validate pre-existing numerical results. Moreover, these different measurement techniques (optical and non-optical) have led to the generalization of the Rayleigh Debye Gans for Fractal Aggregates (RDG-FA) theory to particles of the polydispersed fractal aggregate type coated with an organic layer (RDG-CFA). This allows to understand phenomenologically the impact of a coating on the radiative properties and to permit an easier implementation in climate simulation codes or for the interpretation of optical measurements in the atmosphere. Finally, special attention was paid to the laser induced incandescence technique (LII), to study the applicability of this technique to organic particles or having interacted with atmospheric compounds during their aging processes.

CNN-based (Convolutional Neural Network) visual object detectors often reach human level of accuracy but need to be trained with large amounts of manually annotated data. Collecting and annotating this data can frequently be time-consuming and financially expensive. Using generative models to augment the data can help minimize the amount of data required and increase detection per-formance. Many state-of-the-art generative models are Generative Adversarial Networks (GANs). This thesis investigates if and how one can utilize image data to generate new data through GANs to train a YOLO-based (You Only Look Once) object detector, and how CAD (Computer-Aided Design) models can aid in this process. In the experiments, different models of GANs are trained and evaluated by visual inspection or with the Fréchet Inception Distance (FID) metric. The data provided by Ericsson Research consists of images of antenna and baseband equipment along with annotations and segmentations. Ericsson Research supplied the YOLO detector, and no modifications are made to this detector. Finally, the YOLO detector is trained on data generated by the chosen model and evaluated by the Average Precision (AP). The results show that the generative models designed in this work can produce RGB images of high quality. However, the quality reduces if binary segmentation masks are to be generated as well. The experiments with CAD input data did not result in images that could be used for the training of the detector. The GAN designed in this work is able to successfully replace objects in images with the style of other objects. The results show that training the YOLO detector with GAN-modified data compared to training with real data leads to the same detection performance. The results also show that the shapes and backgrounds of the antennas contributed more to detection performance than their style and colour.

Die Serie »Nutztiere in Sachsen« umfasst neun Ausgaben: Milchrind, Fleischrind, Schaf, Ziege und Schwein; weiterhin Pferd, Legehenne, Mastgeflügel und Karpfen. Die Faltblätter informieren in kompakter Form über Bedeutung und Bestände der Nutztiere in Sachsen und geben einen Überblick über Rassen, Zuchtziele, Leistungen sowie über Haltungsformen, Fütterung und Produktqualität. Sie richten sich in erster Linie an Personen, welche ohne spezielle Vorkenntnisse an der Nutztierhaltung interessiert sind. Redaktionsschluss: 30.09.2017

Today we are surrounded by IOT devices that constantly generate different kinds of data about its environment and its users. Much of this data could be useful for different research purposes and development, but a lot of this collected data is privacy-sensitive for the individual person. To protect the individual's privacy, we have data protection laws. But these restrictions by laws also dramatically reduce the amount of data available for research and development. Therefore it would be beneficial if we could find a work around that respects people's privacy without breaking the laws while still maintaining the usefulness of data. The purpose of this thesis is to show how we can generate privacy-aware data from a dataset by using Generative Adversarial Networks (GANS) and Differential Privacy (DP), that maintains data utility. This is useful because it allows for the sharing of privacy-preserving data, so that the data can be used in research and development with concern for privacy. GANS is used for generating synthetic data. DP is an anonymization technique of data. With the combination of these two techniques, we generate synthetic-privacy-aware data from an existing open-source Fitbit dataset. The specific type of GANS model that is used is called CTGAN and differential privacy is achieved with the help of gaussian noise. The results from the experiments performed show many similarities between the original dataset and the experimental datasets. The experiments performed very well at the Kolmogorov Smirnov test, with the lowest P-value of all experiments sitting at 0.92. The conclusion that is drawn is that this is another promising methodology for creating privacy-aware-synthetic data, that maintains reasonable data utility while still utilizing DP techniques to achieve data privacy.

Abstractive text summarization aims to condense long textual documents into a short, human-readable form while preserving the most important information from the source document. A common approach to training summarization models is by using maximum likelihood estimation with the teacher forcing strategy. Despite its popularity, this method has been shown to yield models with suboptimal performance at inference time. This work examines how using alternative, task-specific training signals affects the performance of summarization models. Two novel training signals are proposed and evaluated as part of this work. One, a novelty metric, measuring the overlap between n-grams in the summary and the summarized article. The other, utilizing a discriminator model to distinguish human-written summaries from generated ones on a word-level basis. Empirical results show that using the mentioned metrics as rewards for policy gradient training yields significant performance gains measured by ROUGE scores, novelty scores and human evaluation. / Abstraktiv textsammanfattning syftar på att korta ner långa textdokument till en förkortad, mänskligt läsbar form, samtidigt som den viktigaste informationen i källdokumentet bevaras. Ett vanligt tillvägagångssätt för att träna sammanfattningsmodeller är att använda maximum likelihood-estimering med teacher-forcing-strategin. Trots dess popularitet har denna metod visat sig ge modeller med suboptimal prestanda vid inferens. I det här arbetet undersöks hur användningen av alternativa, uppgiftsspecifika träningssignaler påverkar sammanfattningsmodellens prestanda. Två nya träningssignaler föreslås och utvärderas som en del av detta arbete. Den första, vilket är en ny metrik, mäter överlappningen mellan n-gram i sammanfattningen och den sammanfattade artikeln. Den andra använder en diskrimineringsmodell för att skilja mänskliga skriftliga sammanfattningar från genererade på ordnivå. Empiriska resultat visar att användandet av de nämnda mätvärdena som belöningar för policygradient-träning ger betydande prestationsvinster mätt med ROUGE-score, novelty score och mänsklig utvärdering.

<p>Generative Adversarial Networks (GANs) have consistently defined the state-of-the-art in the generative modelling of high-quality images in several applications.  The images generated using GANs, however, do not lend themselves to being directly used in supervised learning tasks without first being curated through annotations.  This dissertation investigates how to carry out automatic on-the-fly segmentation of GAN-generated images and how this can be applied to the problem of producing high-quality simulated data for X-ray based security screening.  The research exploits the hidden layer properties of GAN models in a self-supervised learning framework for the automatic one-shot segmentation of images created by a style-based GAN.  The framework consists of a novel contrastive learner that is based on a Sinkhorn distance-based clustering algorithm and that learns a compact feature space for per-pixel classification of the GAN-generated images.  This facilitates faster learning of the feature vectors for one-shot segmentation and allows on-the-fly automatic annotation of the GAN images.  We have tested our framework on a number of standard benchmarks (CelebA, PASCAL, LSUN) to yield a segmentation performance that not only exceeds the semi-supervised baselines by an average wIoU margin of 1.02 % but also improves the inference speeds by a factor of 4.5.  This dissertation also presents BagGAN, an extension of our framework to the problem domain of X-ray based baggage screening.  BagGAN produces annotated synthetic baggage X-ray scans to train machine-learning algorithms for the detection of prohibited items during security screening.  We have compared the images generated by BagGAN with those created by deterministic ray-tracing models for X-ray simulation and have observed that our GAN-based baggage simulator yields a significantly improved performance in terms of image fidelity and diversity.  The BagGAN framework is also tested on the PIDRay and other baggage screening benchmarks to produce segmentation results comparable to their respective baseline segmenters based on manual annotations.</p>

Musical timbre transfer is the task of re-rendering the musical content of a given source using the rendering style of a target sound. The source keeps its musical content, e.g., pitch, microtiming, orchestration, and syncopation. I specifically focus on the task of transferring the style of percussive patterns extracted from polyphonic audio using a MelGAN-VC model [57] by training acoustic properties for each genre. Evaluating audio style transfer is challenging and typically requires user studies. An analytical methodology based on supervised and unsupervised learning including visualization for evaluating musical timbre transfer is proposed. The proposed methodology is used to evaluate the MelGAN-VC model for musical timbre transfer of drum tracks. The method uses audio features to analyze results of the timbre transfer based on classification probability from Random Forest classifier. And K-means algorithm can classify unlabeled instances using audio features and style-transformed results are visualized by t-SNE dimensionality reduction technique, which is helpful for interpreting relations between musical genres and comparing results from the Random Forest classifier. / Graduate