Cancer is a complex, multi-scale disease marked by unchecked cellular growth and proliferation. As a tumor grows, it is known to lose its capacity to maintain a compact structure. This stage of development, known as invasion, is marked by the disaggregation and dispersion of peripheral cells, and the formation of finger-like margins. This thesis provides an overview of three multi-scale models of tumor growth and invasion. The hybrid discrete-continuum (HDC) model couples a cellular automaton approach, which is used to direct the behavior and interactions of individual cells, with a system of reaction-diffusion-chemotaxis equations that describe the micro-environment. The evolutionary hybrid cellular automaton (EHCA) model maintains the core of the HDC approach, but employs an artificial response network to describe cellular dynamics. In contrast to these two, the immersed boundary (IBCell) model describes cells as fully deformable, viscoelastic entities that interact with each other using membrane bound receptors. As part of this thesis, the HDC model has been modified to examine the role of the ECM as a barrier to cellular expansion. The results of these simulations will be presented and discussed in the context of tumor progression.

Digital Elevation Models (DEMs) are fundamental datasets for environmental modelling. They provide the basic data from which terrain indices that represent or influence environmental phenomena are derived, for example slope gradient and hydrological contributing area, and also the source from which specific morphometric features are quantified and characterised, for example mountains and drainage basins. This thesis focuses on the latter, with the aim being to develop an algorithm to characterise the landscape in terms of five morphometric features (peaks, passes, pits, ridges and valleys) and to assess its validity and effectiveness for characterising landform from DEMs. The research in this thesis is divided into two parts. First, an algorithm of morphometric characterisation of landform from OEMs is developed based on a locally fitted quadratic surface and its positional relationship with the analysis window. Five requirements are taken into account within the algorithm: (1) the ideal cases of different morphometric features are simply and clearly defined; (2) the output is spatially continuous to reflect the inherent fuzziness of landform features; (3) the output is easily combined into a multi-scale index across a range of operational scales; (4) the standard general morphometric parameters can be easily quantified due to the easy calculation of first and second order derivatives from the quadratic surface; and (5) the algorithm is applicable to the different data structures used to represent DEMs. An additional benefit of the quadratic surface is the derivation of the R?? goodness-of-fit statistic, which allows both an assessment of the reliability of the results and the complexity of the terrain. Of the five morphometric features identified using the algorithm, valleys are perhaps the most commonly used. Therefore the second part of this thesis is a more detailed comparison between the Multi-Scale Valleyness (MSV) and three existing algorithms (D8, D∞ and MrVBF). D8 and D∞ are global flow accumulation algorithms, and perform well when characterising valley centre lines. However, they do not identify the valley areas themselves, although this is to be expected given their formulation. MrVBF focuses on characterising valley bottoms and so performs well when characterising valleys in broad and topographically flat areas. It does not identify valleys in the steeper upland parts of a catchment, although this too is something to be expected given its formulation. MSV directly characterises valley areas from a geomorphometric point of view, and performs well for both upland and lowland catchments, irrespective of their width. Overall, the results show that the single- and multi-scale terrain indices developed in this research perform well when characterising the five morphometric features. The approach has considerable potential for use in environmental modelling and terrain analysis.

Many people in widely varied fields are exposed to categorical data describing myriad observations. The breadth of applications in which categorical data are used means that many of the people tasked to apply these data have not been trained in data analysis. Visualization of data is often used to alleviate this problem since visualization can convey relevant information in a non-mathematical manner. However, visualizations are frequently static and the tools to create them are largely geared towards quantitative data. It is the purpose of this thesis to demonstrate a method which expands on the parallel coordinates method of visualization and uses a 'Google Maps' style of interaction and view dependent data presentation for visualizing and exploring categorical data that is accessible by non-experts and promotes the use of domain specific knowledge. The parallel coordinates method has enjoyed increasing popularity in recent times, but has several shortcomings. This thesis seeks to address some of these problems in a manner which involves not just addressing the final static image which is generated, but the paradigm of interaction as well.

Abstract This thesis presents extensions to the local binary pattern (LBP) texture analysis operator. The operator is defined as a gray-scale invariant texture measure, derived from a general definition of texture in a local neighborhood. It is made invariant against the rotation of the image domain, and supplemented with a rotation invariant measure of local contrast. The LBP is proposed as a unifying texture model that describes the formation of a texture with micro-textons and their statistical placement rules. The basic LBP is extended to facilitate the analysis of textures with multiple scales by combining neighborhoods with different sizes. The possible instability in sparse sampling is addressed with Gaussian low-pass filtering, which seems to be somewhat helpful. Cellular automata are used as texture features, presumably for the first time ever. With a straightforward inversion algorithm, arbitrarily large binary neighborhoods are encoded with an eight-bit cellular automaton rule, resulting in a very compact multi-scale texture descriptor. The performance of the new operator is shown in an experiment involving textures with multiple spatial scales. An opponent-color version of the LBP is introduced and applied to color textures. Good results are obtained in static illumination conditions. An empirical study with different color and texture measures however shows that color and texture should be treated separately. A number of different applications of the LBP operator are presented, emphasizing real-time issues. A very fast software implementation of the operator is introduced, and different ways of speeding up classification are evaluated. The operator is successfully applied to industrial visual inspection applications and to image retrieval.

Modern nonlinear programming solvers can efficiently handle very large scale optimization problems when accurate derivative information is available. However, black box or derivative free modeling components are often unavoidable in practice when the modeled phenomena may cross length and time scales. This work is motivated by examples in chemical process optimization where most unit operations have well-known equation oriented representations, but some portion of the model (e.g. a complex reactor model) may only be available with an external function call. The concept of a surrogate model is frequently used to solve this type of problem. A surrogate model is an equation oriented approximation of the black box that allows traditional derivative based optimization to be applied directly. However, optimization tends to exploit approximation errors in the surrogate model leading to inaccurate solutions and repeated rebuilding of the surrogate model. Even if the surrogate model is perfectly accurate at the solution, this only guarantees that the original problem is feasible. Since optimality conditions require gradient information, a higher degree of accuracy is required. In this work, we consider the general problem of hybrid glass box/black box optimization, or gray box optimization, with focus on guaranteeing that a surrogate-based optimization strategy converges to optimal points of the original detailed model. We first propose an algorithm that combines ideas from SQP filter methods and derivative free trust region methods to solve this class of problems. The black box portion of the model is replaced by a sequence of surrogate models (i.e. surrogate models) in trust region subproblems. By carefully managing surrogate model construction, the algorithm is guaranteed to converge to true optimal solutions. Then, we discuss how this algorithm can be modified for effective application to practical problems. Performance is demonstrated on a test set of benchmarks as well as a set of case studies relating to chemical process optimization. In particular, application to the oxycombustion carbon capture power generation process leads to significant efficiency improvements. Finally, extensions of surrogate-based optimization to other contexts is explored through a case study with physical properties.

京都大学 / 0048 / 新制・課程博士 / 博士(工学) / 甲第22474号 / 工博第4735号 / 新制||工||1740(附属図書館) / 京都大学大学院工学研究科化学工学専攻 / (主査)教授 山本 量一, 教授 渡辺 宏, 准教授 谷口 貴志 / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

Is there a significant difference in the design of deep networks for the tasks of classifying object-centric images and scenery images? How to design networks that extract the most representative features for scene recognition? To answer these questions, we design studies to examine the scales and richness of image features for scenery image recognition. Three methods are proposed that integrate the scale factor to the deep networks and reveal the fundamental network design strategies. In our first attempt to integrate scale factors into the deep network, we proposed a method that aggregates both the context and multi-scale object information of scene images by constructing a multi-scale pyramid. In our design, integration of object-centric multi-scale networks achieved a performance boost of 9.8%; integration of object- and scene-centric models obtained an accuracy improvement of 5.9% compared with single scene-centric models. We also exploit bringing the attention scheme to the deep network and proposed a Scale Attentive Network (SANet). The SANet streamlines the multi-scale scene recognition pipeline, learns comprehensive scene features at various scales and locations, addresses the inter-dependency among scales, and further assists feature re-calibration as well as the aggregation process. The proposed network achieved a Top-1 accuracy increase by 1.83% on Place365 standard dataset with only 0.12% additional parameters and 0.24% additional GFLOPs using ResNet-50 as the backbone. We further bring the scale factor implicitly into network backbone design by proposing a Deep-Narrow Network and Dilated Pooling module. The Deep-narrow architecture increased the depth of the network as well as decreased the width of the network, which uses a variety of receptive fields by stacking more layers. We further proposed a Dilated Pooling module which expanded the pooling scope and made use of multi-scale features in the pooling operation. By embedding the Dilated Pooling into Deep-Narrow Network, we obtained a Top-1 accuracy boost of 0.40% using less than half of the GFLOPs and parameters compared to benchmark ResNet-50.

Iterative competition between organisms for limited resources gives rise to different social strategies including cooperation. One specific problem in the cooperating but competing species in that cost associated in exhibiting cooperative traits provokes "cheating". Cheaters acquire relatively higher fitness by reaping the benefits of cooperation without contributing towards community beneficial goods. While the relatively fit cheaters can drive the contributors to extinction, the contributors exhibit different strategies to gain preferential benefits of cooperation. The facultative benefit of cooperation to cheaters drives the population to an equilibrium frequency of cooperators and cheaters. Here we develop a multi-scale modeling approach to simulate the dynamics of such cooperation within mixed population of contributors and cheaters. We recursively use genome-scale metabolic models to estimate the fitness of the organism based on the current ecological state. In addition, a series of ordinary differential equations estimate the dynamics of the population and ecological conditions. We use our approach to investigate alternative strategies whereby the cooperating strain may improve its fitness and find that regulation of gene expression is superior to modulation of enzyme activity in our system.

L’objectif de ce travail de recherche est de déterminer l’influence de la présence de particules thermoplastiques dans une intercouche sur le comportement mécanique d’un stratifié. L’ajout de cette intercouche entre chaque pli permet d’accroître la ténacité inter-pli, diminuant ainsi le délaminage, principalement lors de chargements hors-plan. Cette thèse se concentre sur les endommagements qui apparaissent suite à des sollicitations dans le plan du stratifié et plus particulièrement la fissuration transverse et le micro-délaminage. En s’appuyant sur des essais mécaniques appropriés, à la fois à l’échelle des particules (essais Compact Tension) et de la structure (essais double entaille), l’interaction entre ces deux endommagements et les particules est mise en évidence. Une analyse numérique locale a été réalisée sur un Volume Elémentaire Représentatif en se basant sur la mécanique de la rupture. Elle a permis d’étudier la sensibilité aux divers paramètres de l’intercouche tels que son épaisseur, sa raideur vis-à-vis de celle de la partie fibreuse du pli et la répartition de ses particules. Les résultats de cette analyse ainsi que l’outil numérique créé pour la réaliser fournissent une aide à la conception des matériaux composite stratifiés. La suite du travail a permis de montrer, après une étape d’identification des paramètres suivie d’une étape de vérification, l’aptitude du méso-modèle issu du laboratoire LMT de l’ENS à prédire le comportement mécanique de plusieurs essais avec des sollicitations dans le plan du stratifié mais également hors-plan. La comparaison systématique des résultats numériques et expérimentaux montre la capacité de ce modèle à prédire le comportement des endommagements des stratifiés avec intercouches. / The aim of this research work is to determine the influence of thermoplastic particles which are inside the interleaf of a laminate on its mechanical behavior. This interleaf between each ply enables the interply toughness to increase and prevents from delamination, especially for out-of-plane loadings. This PhD focuses on damages occurring during in-plane-loading, such as transverse crack and micro-delamination. Appropriate experimental tests have been performed, both at the scale of the particles (Compact Tension) and at the scale of the structure (double notch) in order to describe the interaction between the particles and these two damages. A local numerical study based on fracture mechanics has been performed on a Representative Volume Element. The influence of several parameters of the interleaf, such as its thickness, its stiffness compared to the one the ply and the repartition of the particles was determined. The results of this study constitute a new route towards a material by design of interleaved laminates approach. At the coupon structural scale, the parameters for the simulation using the damage meso model of the laboratory LMT of ENS have been accurately identified. Several experiments for both in-plane and out-of-plane loadings have been simulated using the meso model in order to verify the required model parameters. All experimental results have been compared to the numerical ones, showing that this model is able to describe the damage behavior of interleaved laminates.

No description available.