Designing generalized data-driven distance measures for both ordered and unordered set data is the core focus of the proposed work. An ordered set is a set where time-linear property is maintained when distance between pair of temporal segments. One application in the ordered set is the human gesture analysis from RGBD data. Human gestures are fast becoming the natural form of human computer interaction. This serves as a motivation to modeling, analyzing, and recognition of gestures. The large number of gesture categories such as sign language, traffic signals, everyday actions and also subtle cultural variations in gesture classes makes gesture recognition a challenging problem. As part of generalization, an algorithm is proposed as part of an overlap speech detection application for unordered set. Any gesture recognition task involves comparing an incoming or a query gesture against a training set of gestures. Having one or few samples deters any class statistic learning approaches to classification, as the full range of variation is not covered. Due to the large variability in gesture classes, temporally segmenting individual gestures also becomes hard. A matching algorithm in such scenarios needs to be able to handle single sample classes and have the ability to label multiple gestures without temporal segmentation. Each gesture sequence is considered as a class and each class is a data point on an input space. A pair-wise distances pattern between to gesture frame sequences conditioned on a third (anchor) sequence is considered and is referred to as warp vectors. Such a process is defined as conditional distances. At the algorithmic core we have two dynamic time warping processes, one to compute the warp vectors with the anchor sequences and the other to compare these warp vectors. We show that having class dependent distance function can disambiguate classification process where the samples of classes are close to each other. Given a situation where the model base is large (number of classes is also large); the disadvantage of such a distance would be the computational cost. A distributed version combined with sub-sampling anchor gestures is proposed as speedup strategy. In order to label multiple connected gestures in query we use a simultaneous segmentation and recognition matching algorithm called level building algorithm. We use the dynamic programming implementation of the level building algorithm. The core of this algorithm depends on a distance function that compares two gesture sequences. We propose that, we replace this distance function, with the proposed distances. Hence, this version of level building is called as conditional level building (clb). We present results on a large dataset of 8000 RGBD sequences spanning over 200 gesture classes, extracted from the ChaLearn Gesture Challenge dataset. The result is that there is significant improvement over the underlying distance used to compute conditional distance when compared to conditional distance. As an application of unordered set and non-visual data, overlap speech segment detection algorithm is proposed. Speech recognition systems have a vast variety of application, but fail when there is overlap speech involved. This is especially true in a meeting-room setting. The ability to recognize speaker and localize him/her in the room is an important step towards a higher-level representation of the meeting dynamics. Similar to gesture recognition, a new distance function is defined and it serves as the core of the algorithm to distinguish between individual speech and overlap speech temporal segments. The overlap speech detection problem is framed as outlier detection problem. An incoming audio is broken into temporal segments based on Bayesian Information Criterion (BIC). Each of these segments is considered as node and conditional distance between the nodes are determined. The underlying distances for triples used in conditional distances is the symmetric KL distance. As each node is modeled as a Gaussian, the distance between the two segments or nodes is given by Monte-Carlo estimation of the KL distance. An MDS based global embedding is created based on the pairwise distance between the nodes and RANSAC is applied to compute the outliers. NIST meeting room data set is used to perform experiments on the overlap speech detection. An improvement of more than 20% is achieved with conditional distance based approach when compared to a KL distance based approach.

Notre contribution concerne les trois domaines complémentaires suivants: la différentiation automatique de programmes, l'optimisation de formes pour de grands systèmes, l'adaptation de maillages. Dans le chapitre 1 de la partie 1, nous exposons une méthode de calcul de gradients par Différentiation Automatique pour un problème classique d'optimisation de formes. Nous expliquons comment déduire un gradient exact basé sur un état adjoint sans stocker explicitement le jacobien. Le mode adjoint de la DA que nous proposons utilise beaucoup moins d'espace mémoire. Dans le chapitre 2 de la partie 2, nous proposons une méthode de type SQP pour résoudre une classe de problèmes d'optimisation avec contraintes égalités. Le nouvel algorithme permet une résolution simultanée du système d'optimalité. Cette méthode one shot combine efficacité et robustesse. Dans le chapitre 3 de la partie 2, nous étudions une nouvelle stratégie de préconditionnement pour l'optimisation de formes. Nous construisons un préconditionnement multiniveau additif à partir du principe classique de Bramble-Pasciak-Xu et du principe d'agglomération. Nous spécifions aisément le gain en régularité de notre préconditionneur avec un seul paramètre réel. Dans le chapitre 1 de la partie 3, nous étudions le problème du meilleur maillage adapté pour de l'interpolation pure. La résolution du système d'optimalité donne une expression complètement explicite de la métrique optimale en fonction de la fonction à adapter. Dans le chapitre 2 de la partie 3, nous étendons la méthode du chapitre précédent au problème de l'adaptation de maillage pour EDP. Notre méthode repose sur une analyse a priori rigoureuse puis sur une modélisation.

This thesis presents techniques constructed to aid the radiologists in detecting breast cancer, the second largest cause of cancer deaths for western women. In the first part of the thesis, a computer-aided detection (CAD) system constructed for the detection of stellate lesions is presented. Different segmentation methods and an attempt to incorporate contra-lateral information are evaluated. In the second part, a new method for evaluating such CAD systems is presented based on constructing credible regions for the number of false positive marks per image at a certain desired target sensitivity. This method shows that the resulting regions are rather wide and this explains some of the difficulties encountered by other researchers when trying to compare CAD algorithms on different data sets. In this part an attempt to model the clinical use of CAD as a second look is also made and it shows that applying CAD in sequence to the radiologist in a routine manner, without duly altering the decision criterion of the radiologist, might very well result in suboptimal operating points. Finally, in the third part two dual-energy imaging methods optimized for contrast-enhanced imaging of breast tumors are presented. The first is based on applying an electronic threshold to a photon-counting digital detector to discriminate between high- and low-energy photons. This allows simultaneous acquisition of the high- and low-energy images. The second method is based on the geometry of a scanned multi-slit system and also allows single-shot contrast-enhanced dual-energy mammography by filtering the x-ray beam that reaches different detector lines differently. / QC 20100819

This project examines the how camera movement as a stylistic element is used as a storytelling device in the films of select international filmmakers. The main intention of the study is to trace the changing function of the mobile frame to see how a specific stylistic element develops across different narrative paradigms, national industries and between “early” and contemporary periods of filmmakers. My primary assertion is that the norms guiding the development of the tracking camera expand gradually from normative functions toward figurative uses. In order to be able to differentiate between normative and figurative uses of the tracking camera with conceptual clarity, this project adapts Roland Barthes’s typology about the narrative function of distinctive textual/stylistic units. Barthes’ conceptual framework becomes functional by assigning specific codes (hermeneutic, the semic, the proairetic, the symbolic and the cultural codes) to the interactions of the elements of narration. When transforming and changing the function of stylistic elements across their films, artists respond to a wide range of industrial, technological, aesthetic, cultural factors, from which this study focuses on socio-cultural trends. The underlying assumption of this project holds that the mentioned trends can be detected in the stylistic choices of artists. This study takes a bottom-up route: starting with an analytical interpretation of a specific aesthetic device, it moves towards an explanation that connects camera movement to larger, dynamic signifying systems. The arch of my project traces the relation between normative and figurative textual codes through the prism of camera movement.

We analyze experimental evidence of transport through evanescent waves in graphene, reconciling existing experimental data with theory. We propose novel experimental geometries that provide even more compelling evidence of evanescent waves. We investigate the shot noise generated by evanescent modes in graphene for several experimental setups. For two impurity-free graphene strips kept at the Dirac point by gate potentials, separated by a long highly doped region, we find that the Fano factor takes the universal value F=1/4. For a large superlattice consisting of many strips gated to the Dirac point, interspersed among doped regions, we find F=1/(8ln2). These results differ from the value F=1/3 predicted for a disordered metal, providing an unambiguous experimental signature of evanescent mode transport in graphene. For a graphene nano-ribbon transistor geometry, we explain that the experimentally observed anomalous voltage scale of the shot noise can arise from doping by the contacts to the electrical circuit. These observations provide strong evidence of evanescent mode transport in graphene.

This work was carried out to study the effect of shot-peening on the fatigue behaviour of carbon steels. Differently heat treated medium and high carbon steel specimens were selected. Medium carbon steels, AISI 1141 and AISI 1151, were respectively air cooled and quenched-tempered. A high carbon steel, C70S6 (AISI 1070), was air cooled. The other material was a powder metal (0.5% C) steel. Each group of steels was divided into two. One was shot-peened. The other half remained in their original conditions. All were fatigue tested under fully reversed (R=-1) tension-compression loading conditions. Microhardness tests were carried out on both the grip and gage sections of selected non shot-peened and shot-peened specimens to determine the hardness profile and effect of cycling. Shot-peening was found to be deeper on one side of each specimen. Compressive residual stress profiles and surface roughness measurements were provided. Shot-peening increased the surface roughness from 0.26±0.03µm to 3.60±0.44µm. Compressive residual stresses induced by shot-peening reached a maximum of -463.9MPa at a depth of 0.1mm.The fatigue limit (N≈106 cycles) and microhardness profiles of the non shot-peened and shot-peened specimens were compared to determine the material behaviour changes after shot-peening and cycling. Also their fatigue properties were related to the manufacturing process including heat and surface treatments. Comparing the grip and gage microhardness profiles of each steel showed that neither cyclic softening nor hardening occurred in the non shot-peened condition. Cyclic softening was apparent in the shot-peened regions of all steels except powder metal (PM) steel. The amount of softening in the shot-peened region was 55.0% on the left side and 73.0% on the right in the AISI 1141 steel , 46.0% on the left side and 55.0% on the right in the C70S6AC steel and 31.0% on the right side in AISI 1151QT steel. Softening was accompanied by a decrease in the depth of surface hardness. It is suggested that although the beneficial effects of shot peening, compressive residual stresses and work hardening, were offset by surface roughness, crack initiation was more likely to occur below the surface. Surface roughness was not a significant factor in controlling the fatigue lives of AISI 1141AC and C70S6 steels, since they were essentially the same for the non shot-peened and shot-peened conditions. Shot-peening had very little effect on the push-pull fatigue limit of C70S6 steel (-2.1%), and its effect on AISI 1141AC steel was relatively small (6.0%). However, the influence of shot-peening on the AISI 1151QT and PM steels was more apparent. The fatigue limit of the PM steel increased 14.0% whereas the fatigue limit of the AISI 1151QT steel decreased 11.0% on shot peening.

This thesis aims comparative analysis of the state of the art shot boundary detection algorithms. The major methods that have been used for shot boundary detection such as pixel intensity based, histogram-based, edge-based, and motion vectors based, are implemented and analyzed. A recent method which utilizes &ldquo / graph partition model&rdquo / together with the support vector machine classifier as a shot boundary detection algorithm is also implemented and analyzed. Moreover, a novel graph theoretic concept, &ldquo / dominant sets&rdquo / , is also successfully applied to the shot boundary detection problem as a contribution to the solution domain.

In this study, residual stresses introduced via application of shot peening on the raceways of bearing rings and their effect on the fatigue life was investigated experimentally. For improvement of residual compressive stress state, shot peening operation with different parameters was utilized. Residual stress measurements were conducted via X-ray diffraction technique. Optimization of residual stress state during the production of ball bearings is the main target of this study. Process parameters for shot peening and super-finishing were studied for determination of the parameters that induce the most favorable residual stress state. The fatigue life of ball bearings were determined by life cycle tests and tabulated to show the results of the study. The results of the thesis are believed to help for optimization of residual stress distribution and improvement of service life of ball bearings.

Shot peening is a cold-working process by which residual compressive stresses are being induced in the surface region to increase the fatigue strength and the resistance to stress-corrosion cracking. This study covers non-destructive measurement of surface residual stresses in the shot-peened steel components by a micro-magnetic technique, named as Magnetic Barkhausen Noise (MBN) method. For this purpose, various low alloy steel specimens were prepared by a controlled shot peening process with different intensity, impact angle and coverage values. The measurements showed that a clear relationship exists between residual stresses and the MBN signals. Residual stress values determined by MBN technique were also verified by X-ray diffraction measurements.

Airburst projectiles increase the effectiveness of air defense, by forming clouds of small pellets. In this work, in order to evaluate the effectiveness of airburst projectiles, Single Shot Kill Probability (SSKP) is computed at different burst distances by using three lethality functions defined from different measures of effectiveness. These different measures are target coverage, number of sub-projectile hits on the target and kinetic energy of sub-projectiles after burst. Computations are carried out for two different sub-projectile distribution patterns, namely circular and ring patterns. In this work, for the determination of miss distance, a Monte Carlo simulation is implemented, which uses Modified Point Mass Model (MPMM) trajectory equations. According to the results obtained two different distribution patterns are compared in terms of effectiveness and optimum burst distance of each distribution pattern is determined at different ranges.