Thesis (M. Sc.)--University of Hong Kong, 1997. / Includes bibliographical references (leaf 45-48).

L'accroissement rapide des données numériques vidéographiques fait de la compréhension automatiquedes vidéos un enjeu de plus en plus important. Comprendre de manière automatique une vidéo recouvrede nombreuses applications, parmi lesquelles l'analyse du contenu vidéo sur le web, les véhicules autonomes,les interfaces homme-machine. Cette thèse présente des contributions dans deux problèmes majeurs pourla compréhension automatique des vidéos : la détection d'actions supervisée par des données web, et la localisation d'actions humaines.La détection d'actions supervisées par des données web a pour objectif d'apprendre à reconnaître des actions dans des contenus vidéos sur Internet, sans aucune autre supervision. Nous proposons une approche originaledans ce contexte, qui s'appuie sur la synergie entre les données visuelles (les vidéos) et leur description textuelle associée, et ce dans le but d'apprendre des classifieurs pour les événements sans aucune supervision. Plus précisément, nous télechargeons dans un premier temps une base de données vidéos à partir de requêtes construites automatiquement en s'appuyant sur la description textuelle des événéments, puis nous enlevons les vidéos téléchargées pour un événement, et dans laquelle celui-ci n'apparaït pas. Enfin, un classifieur est appris pour chaque événement. Nous montrons l'importance des deux étapes principales, c'est-à-dire la créations des requêtes et l'étape de suppression des vidéos, par des résutatsquantitatifs. Notre approche est évaluée dans des conditions difficiles, où aucune annotation manuelle n'est disponible, dénotées EK0 dans les challenges TrecVid. Nous obtenons l'état de l'art sur les bases de donnéesMED 2011 et 2013.Dans la seconde partie de notre thèse, nous nous concentrons sur la localisation des actions humaines, ce qui implique de reconnaïtre à la fois les actions se déroulant dans la vidéo, comme par exemple "boire" ou "téléphoner", et leur étendues spatio-temporelles. Nous proposons une nouvelle méthode centrée sur la personne, traquant celle-ci dans les vidéos pour en extraire des tubes encadrant le corps entier, même en cas d'occultations ou dissimulations partielles. Deux raisons motivent notre approche. La première est qu'elle permet de gérer les occultations et les changements de points de vue de la caméra durant l'étape de localisation des personnes, car celle-ci estime la position du corps entier à chaque frame. La seconde est que notre approche fournit une meilleure grille de référence que les tubes humains standards (c'est-à-dire les tubes qui n'encadrent que les parties visibles) pour extraire de l'information sur l'action. Le coeur de notre méthode est un réseau de neurones convolutionnel qui apprend à générer des propositions de parties du corps humain. Notre algorithme de tracking connecte les détections temporellement pour extraire des tubes encadrant le corps entier. Nous évaluons notre nouvelle méthode d'extraction de tubes sur une base de données difficile, DALY, et atteignons l'état de l'art. / With the rapid growth of digital video content, automaticvideo understanding has become an increasingly important task. Video understanding spansseveral applications such as web-video content analysis, autonomous vehicles, human-machine interfaces (eg, Kinect). This thesismakes contributions addressing two major problems in video understanding:webly-supervised action detection and human action localization.Webly-supervised action recognition aims to learn actions from video content on the internet, with no additional supervision. We propose a novel approach in this context, which leverages thesynergy between visual video data and the associated textual metadata, to learnevent classifiers with no manual annotations. Specifically, we first collect avideo dataset with queries constructed automatically from textual descriptionof events, prune irrelevant videos with text and video data, and then learn thecorresponding event classifiers. We show the importance of both the main steps of our method, ie,query generation and data pruning, with quantitative results. We evaluate this approach in the challengingsetting where no manually annotated training set is available, i.e., EK0 in theTrecVid challenge, and show state-of-the-art results on MED 2011 and 2013datasets.In the second part of the thesis, we focus on human action localization, which involves recognizing actions that occur in a video, such as ``drinking'' or ``phoning'', as well as their spatial andtemporal extent. We propose a new person-centric framework for action localization that trackspeople in videos and extracts full-body human tubes, i.e., spatio-temporalregions localizing actions, even in the case of occlusions or truncations.The motivation is two-fold. First, it allows us to handle occlusions and camera viewpoint changes when localizing people, as it infers full-body localization. Second, it provides a better reference grid for extracting action information than standard human tubes, ie, tubes which frame visible parts only.This is achieved by training a novel human part detector that scores visibleparts while regressing full-body bounding boxes, even when they lie outside the frame. The core of our method is aconvolutional neural network which learns part proposals specific to certainbody parts. These are then combined to detect people robustly in each frame.Our tracking algorithm connects the image detections temporally to extractfull-body human tubes. We evaluate our new tube extraction method on a recentchallenging dataset, DALY, showing state-of-the-art results.

abstract: With the emergence of edge computing paradigm, many applications such as image recognition and augmented reality require to perform machine learning (ML) and artificial intelligence (AI) tasks on edge devices. Most AI and ML models are large and computational heavy, whereas edge devices are usually equipped with limited computational and storage resources. Such models can be compressed and reduced in order to be placed on edge devices, but they may loose their capability and may not generalize and perform well compared to large models. Recent works used knowledge transfer techniques to transfer information from a large network (termed teacher) to a small one (termed student) in order to improve the performance of the latter. This approach seems to be promising for learning on edge devices, but a thorough investigation on its effectiveness is lacking. The purpose of this work is to provide an extensive study on the performance (both in terms of accuracy and convergence speed) of knowledge transfer, considering different student-teacher architectures, datasets and different techniques for transferring knowledge from teacher to student. A good performance improvement is obtained by transferring knowledge from both the intermediate layers and last layer of the teacher to a shallower student. But other architectures and transfer techniques do not fare so well and some of them even lead to negative performance impact. For example, a smaller and shorter network, trained with knowledge transfer on Caltech 101 achieved a significant improvement of 7.36\% in the accuracy and converges 16 times faster compared to the same network trained without knowledge transfer. On the other hand, smaller network which is thinner than the teacher network performed worse with an accuracy drop of 9.48\% on Caltech 101, even with utilization of knowledge transfer. / Dissertation/Thesis / Masters Thesis Computer Science 2018

At the Castle Creek study area, levee deposits are well-exposed over an area of ~2.6 km wide and ~90 m thick. This provides an opportunity to describe their lateral and vertical lithological changes, and accordingly details about their reservoir geometry and stratal continuity. Here, levee deposits are divided vertically into packages, each consisting of a sand-rich lower part overlain sharply by a mud-rich upper part. Each lower part displays a consistent thickening then thinning trend laterally away from its genetically related channel. The characteristics of these packages suggest that they were controlled by recurring changes in the structure of channellized flows, which in turn was controlled by grain size and grain sorting. This ultimately was controlled by short-term changes in relative sea level. Moreover, some mud- and sand-rich strata are rich in residual carbon suggesting that mid-fan levees can serve as source rocks for hydrocarbon generation, and also reservoirs.

Please read the abstract in the section 00front of this document / Dissertation (M Eng (Mining Engineering))--University of Pretoria, 2007. / Mining Engineering / unrestricted

This thesis tackles the problem of understanding deep neural network with piece- wise linear activation functions. We leverage tropical geometry, a relatively new field in algebraic geometry to characterize the decision boundaries of a single hidden layer neural network. This characterization is leveraged to understand, and reformulate three interesting applications related to deep neural network. First, we give a geo- metrical demonstration of the behaviour of the lottery ticket hypothesis. Moreover, we deploy the geometrical characterization of the decision boundaries to reformulate the network pruning problem. This new formulation aims to prune network pa- rameters that are not contributing to the geometrical representation of the decision boundaries. In addition, we propose a dual view of adversarial attack that tackles both designing perturbations to the input image, and the equivalent perturbation to the decision boundaries.

京都大学 / 0048 / 新制・課程博士 / 博士(情報学) / 甲第22580号 / 情博第717号 / 新制||情||123(附属図書館) / 京都大学大学院情報学研究科社会情報学専攻 / (主査)教授 吉川 正俊, 教授 黒橋 禎夫, 教授 田島 敬史, 教授 田中 克己(京都大学 名誉教授) / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM

Coherent Raman microscopy (CRM) is a powerful nonlinear optical imaging technique based on contrast via Raman active molecular vibrations. CRM has been used in domains ranging from biology to medicine to geology in order to provide quick, sensitive, chemical-specific, and label-free 3D sectioning of samples. The Raman contrast is usually obtained by combining two ultrashort pulse input beams, known as Pump and Stokes, whose frequency difference is adjusted to the Raman vibrational frequency of interest. CRM can be used in conjunction with other imaging modalities such as second harmonic generation, fluorescence, and third harmonic generation microscopy, resulting in a multimodal imaging technique that can capture a massive amount of data. Two fundamental elements are crucial in CRM. First, a laser source which is broadband, stable, rapidly tunable, and low in noise. Second, a strategy for image analysis that can handle denoising and material classification issues in the relatively large datasets obtained by CRM techniques. Stimulated Raman Scattering (SRS) microscopy is a subset of CRM techniques, and this thesis is devoted entirely to it. Although Raman imaging based on a single vibrational resonance can be useful, non-resonant background signals and overlapping bands in SRS can impair contrast and chemical specificity. Tuning over the Raman spectrum is therefore crucial for target identification, which necessitates the use of a broadband and easily tunable laser source. Although supercontinuum generation in a nonlinear fibre could provide extended tunability, it is typically not viable for some CRM techniques, specifically in SRS microscopy. Signal acquisition schemes in SRS microscopy are focused primarily on detecting a tiny modulation transfer between the Pump and Stokes input laser beams. As a result, very low noise source is required. The primary and most important component in hyperspectral SRS microscopy is a low-noise broadband laser source. The second problem in SRS microscopy is poor signal-to-noise (SNR) ratios in some situations, which can be caused by low target-molecule concentrations in the sample and/or scattering losses in deep-tissue imaging, as examples. Furthermore, in some SRS imaging applications (e.g., in vivo), fast imaging, low input laser power or short integration time is required to prevent sample photodamage, typically resulting in low contrast (low SNR) images. Low SNR images also typically suffer from poorly resolved spectral features. Various de-noising techniques have been used to date in image improvement. However, to enable averaging, these often require either previous knowledge of the noise source or numerous images of the same field of view (under better observing conditions), which may result in the image having lower spatial-spectral resolution. Sample segmentation or converting a 2D hyperspectral image to a chemical concentration map, is also a critical issue in SRS microscopy. Raman vibrational bands in heterogeneous samples are likely to overlap, necessitating the use of chemometrics to separate and segment them. We will address the aforementioned issues in SRS microscopy in this thesis. To begin, we demonstrate that a supercontinuum light source based on all normal dispersion (ANDi) fibres generates a stable broadband output with very low incremental source noise. The ANDi fibre output's noise power spectral density was evaluated, and its applicability in hyperspectral SRS microscopy applications was shown. This demonstrates the potential of ANDi fibre sources for broadband SRS imaging as well as their ease of implementation. Second, we demonstrate a deep learning neural net model and unsupervised machine-learning algorithm for rapid and automated de-noising and segmentation of SRS images based on a ten-layer convolutional autoencoder: UHRED (Unsupervised Hyperspectral Resolution Enhancement and De-noising). UHRED is trained in an unsupervised manner using only a single (“one-shot”) hyperspectral image, with no requirements for training on high quality (ground truth) labelled data sets or images.

In this exciting era of artificial intelligence and machine learning, the success of AlphaGo, AlphaZero, and MuZero has generated a great interest in deep reinforcement learning, especially under self-play settings. The methods used by AlphaZero are finding their ways to be more useful than before in many different application areas, such as clinical medicine, intelligent military command decision support systems, and recommendation systems. While specific methods of reinforcement learning with selfplay have found their place in application domains, there is much to be explored from existing reinforcement learning methods not originally intended for self-play settings. This thesis focuses on evaluating performance of existing reinforcement learning techniques in self-play settings. In this research, we trained and evaluated the performance of two deep reinforcement learning algorithms with self-play settings on game environments, such as the games Connect Four and Chess. We demonstrate how a simple on-policy, policy-based method, such as REINFORCE, shows signs of learning, whereas an off-policy value-based method such as Deep Q-Networks does not perform well with self-play settings in the selected environments. The results show that REINFORCE agent wins 85% of the games after training against a random baseline agent and 60% games against the greedy baseline agent in the game Connect Four. The agent’s strength from both techniques was measured and plotted against different baseline agents. We also investigate the impact of selected significant hyper-parameters in the performance of the agents. Finally, we provide our recommendation for these hyper-parameters’ values for training deep reinforcement learning agents in similar environments. / Graduate

Les ribozymes sont des ARN naturels ou artificiels possédant une activité catalytique. Les ribozymes artificiels ont été identifiés in vitro par la méthode SELEX, et plusieurs d'entre eux ont été caractérisés par des études cinétiques. Ces molécules sont impliquées dans des réactions de clivage, de ligation, de modification d'extrémités d'ARN, de polymérisation, de phosphorylation et d'activation de groupements acyl. Parce qu'elle est nécessaire à la traduction, l'aminoacylation des ARN joue un rôle évolutif important dans la transition du monde de l'ARN vers le monde moderne de l'ADN et des protéines, et elle est centrale à l'établissement du code génétique. Plusieurs ribozymes catalysant le transfert d'acides aminés à partir de cofacteurs activants ont pu être isolés et caractérisés depuis une vingtaine d'années, ce qui a documenté la possibilité d'aminoacylation d'ARNt en l'absence des aminoacyl ARNt synthétases. En développant un nouveau protocole SELEX basé sur l'oxydation au périodate, le but de notre travail est de découvrir de nouveau ribozymes d'une taille de l'ordre d'une vingtaine de nucléotides pouvant combiner la catalyse de l'activation des acides aminé et la transestérification. Bien que des molécules catalysant l'une ou l'autre des deux réactions ont été identifiées, aucun ribozyme n'existe à ce jour qui puisse utiliser des acides aminés libres et un cofacteur activant pour réaliser l'aminoacylation en 3' dans un même milieu réactionnel. La sélection de molécules actives dans une approche SELEX exige la présence de régions constantes sur les deux extrémités des séquences pools aléatoires initiaux. Ces régions sont nécessaires pour l'amplification par PCR, mais elles imposent des contraintes importantes pour l'identification de ribozymes car elles peuvent complètement inhiber leur activité par interférence structurelle. Nous présentons un protocole optimisé qui minimise la taille de ces régions constantes. D'autre part, notre nouveau design est très spécifique pour la sélection d'ARN aminoacylés sur l'extrémité 3'. Ce protocole a été utilisé pour réaliser 6 à 7 cycles de sélection avec différents pools, et un enrichissement en séquences spécifiques a pu être mis en évidence. Bien que certains tests avec les pools sélectionnés a révélé une activité possible, des essais avec des séquences spécifiques de ces pools n'ont pour l'instant pas pu confirmer l'activité catalytique recherchée. Un protocole basé sur le même principe de sélection a été utilisé dans une étude parallèle pour identifier les ARN aminoacylés présents dans l'ARN total d'Escherichia coli. Dans ce deuxième travail, note but est d'identifier tous les d'ARN aminoacylés par séquençage massif, avec à la clé la découverte possible de molécules autres que les ARNt et ARNtm. En utilisant les ARNt comme modèle, nous nous sommes aperçus qu'un protocole RNAseq standard n'était pas adapté à cause des bases modifiées présentes sur ces molécules. Nous avons développé et mis au point un nouveau protocole pour l'identification de n'importe quelle séquence aminoacylée en 3'. La nouvelle approche présentée devrait permette l'étude exhaustive de l'aminoacylation de toutes les séquences présentes dans l'ARN total. / Ribozymes are natural or in vitro selected RNA molecules possessing a catalytic activity. Artificial ribozymes have been extensively investigated by in vitro SELEX experiments, and characterized by kinetic assays. Ribozymes are involved in RNA cleavage, ligation, capping, polymerization, phosphorylation and acyl activation. Because it is required for translation, RNA aminoacylation plays an important role in the evolution from the late RNA world to the modern DNA and protein world, and is central to the genetic code. Several ribozymes catalyzing amino acid transfer from various activating groups have already been selected and characterized in the past two decades, documenting the possibility of tRNA aminoacylation in the absence of aminoacyl tRNA synthetase. With a newly designed SELEX protocol based on periodate oxydation, the aim of our investigation is to uncover small ribozymes of the order of 20 nucleotides that could catalyze both amino acid activation and transesterification. Although molecules catalyzing either reaction have been identified, no existing ribozyme could use free amino acids and activating cofactor(s) as substrates for 3' esterification in a single reactional context. The selection of active molecules in a SELEX procedure requires the presence of constant tracks on both ends of the sequences constituting the initial random pools. These tracks are required for PCR amplification, but they impose significant burden to the identification of ribozymes because they can prevent any activity through structural inhibition. We present an optimized protocol that significantly minimizes the size of these constant tracks. At the same time, our newly design protocol is very specific for the selection of 3'-end aminoacylated RNA. Working with this protocol, we performed 6 to 7 cycles of selection with different pools, and observed an enrichement with specific sequences. Although some experiments performed with entire pools did reveal a possible activity, no activity could be so far confirmed with specific sequences. A similar protocol was also applied in a parallel study to identify aminoacylated RNA from total RNA in Escherichia coli. In this other approach, our goal is to possibly identify new classes of aminoacylated RNA while using the deep sequencing technology. Using tRNA to validate our protocol, we realized that a standard RNAseq procedure could not work due to the presence of modified bases. We established a new method for bank preparation to identify any sequence aminoacylated at the 3' end. Ultimately, this new approach will allow us to study the level of aminoacylation of any sequence present in total RNA.