Dissertationen zum Thema „Réseaux de neurones à graphes“
Geben Sie eine Quelle nach APA, MLA, Chicago, Harvard und anderen Zitierweisen an
Machen Sie sich mit Top-50 Dissertationen für die Forschung zum Thema "Réseaux de neurones à graphes" bekannt.
Neben jedem Werk im Literaturverzeichnis ist die Option "Zur Bibliographie hinzufügen" verfügbar. Nutzen Sie sie, wird Ihre bibliographische Angabe des gewählten Werkes nach der nötigen Zitierweise (APA, MLA, Harvard, Chicago, Vancouver usw.) automatisch gestaltet.
Sie können auch den vollen Text der wissenschaftlichen Publikation im PDF-Format herunterladen und eine Online-Annotation der Arbeit lesen, wenn die relevanten Parameter in den Metadaten verfügbar sind.
Sehen Sie die Dissertationen für verschiedene Spezialgebieten durch und erstellen Sie Ihre Bibliographie auf korrekte Weise.
Carboni, Lucrezia. „Graphes pour l’exploration des réseaux de neurones artificiels et de la connectivité cérébrale humaine“. Electronic Thesis or Diss., Université Grenoble Alpes, 2023. http://www.theses.fr/2023GRALM060.
Der volle Inhalt der QuelleThe main objective of this thesis is to explore brain and artificial neural network connectivity from agraph-based perspective. While structural and functional connectivity analysis has been extensivelystudied in the context of the human brain, there is a lack of a similar analysis framework in artificialsystems.To address this gap, this research focuses on two main axes.In the first axis, the main objective is to determine a healthy signature characterization of the humanbrain resting state functional connectivity. To achieve this objective, a novel framework is proposed,integrating traditional graph statistics and network reduction tools, to determine healthy connectivitypatterns. Hence, we build a graph pair-wise comparison and a classifier to identify pathological statesand rank associated perturbed brain regions. Additionally, the generalization and robustness of theproposed framework were investigated across multiple datasets and variations in data quality.The second research axis explores the benefits of brain-inspired connectivity exploration of artificialneural networks (ANNs) in the future perspective of more robust artificial systems development. Amajor robustness issue in ANN models is represented by catastrophic forgetting when the networkdramatically forgets previously learned tasks when adapting to new ones. Our work demonstrates thatgraph modeling offers a simple and elegant framework for investigating ANNs, comparing differentlearning strategies, and detecting deleterious behaviors such as catastrophic forgetting.Moreover, we explore the potential of leveraging graph-based insights to effectively mitigatecatastrophic forgetting, laying a foundation for future research and explorations in this area
Albano, Alice. „Dynamique des graphes de terrain : analyse en temps intrinsèque“. Thesis, Paris 6, 2014. http://www.theses.fr/2014PA066260/document.
Der volle Inhalt der QuelleWe are surrounded by a multitude of interaction networks from different contexts. These networks can be modeled as graphs, called complex networks. They have a community structure, i.e. groups of nodes closely related to each other and less connected with the rest of the graph. An other phenomenon studied in complex networks in many contexts is diffusion. The spread of a disease is an example of diffusion. These phenomena are dynamic and depend on an important parameter, which is often little studied: the time scale in which they are observed. According to the chosen scale, the graph dynamics can vary significantly. In this thesis, we propose to study dynamic processes using a suitable time scale. We consider a notion of relative time which we call intrinsic time, opposed to "traditional" time, which we call extrinsic time. We first study diffusion phenomena using intrinsic time, and we compare our results with an extrinsic time scale. This allows us to highlight the fact that the same phenomenon observed at two different time scales can have a very different behavior. We then analyze the relevance of the use of intrinsic time scale for detecting dynamic communities. Comparing communities obtained according extrinsic and intrinsic scales shows that the intrinsic time scale allows a more significant detection than extrinsic time scale
Albano, Alice. „Dynamique des graphes de terrain : analyse en temps intrinsèque“. Electronic Thesis or Diss., Paris 6, 2014. http://www.theses.fr/2014PA066260.
Der volle Inhalt der QuelleWe are surrounded by a multitude of interaction networks from different contexts. These networks can be modeled as graphs, called complex networks. They have a community structure, i.e. groups of nodes closely related to each other and less connected with the rest of the graph. An other phenomenon studied in complex networks in many contexts is diffusion. The spread of a disease is an example of diffusion. These phenomena are dynamic and depend on an important parameter, which is often little studied: the time scale in which they are observed. According to the chosen scale, the graph dynamics can vary significantly. In this thesis, we propose to study dynamic processes using a suitable time scale. We consider a notion of relative time which we call intrinsic time, opposed to "traditional" time, which we call extrinsic time. We first study diffusion phenomena using intrinsic time, and we compare our results with an extrinsic time scale. This allows us to highlight the fact that the same phenomenon observed at two different time scales can have a very different behavior. We then analyze the relevance of the use of intrinsic time scale for detecting dynamic communities. Comparing communities obtained according extrinsic and intrinsic scales shows that the intrinsic time scale allows a more significant detection than extrinsic time scale
Limnios, Stratis. „Graph Degeneracy Studies for Advanced Learning Methods on Graphs and Theoretical Results Edge degeneracy: Algorithmic and structural results Degeneracy Hierarchy Generator and Efficient Connectivity Degeneracy Algorithm A Degeneracy Framework for Graph Similarity Hcore-Init: Neural Network Initialization based on Graph Degeneracy“. Thesis, Institut polytechnique de Paris, 2020. http://www.theses.fr/2020IPPAX038.
Der volle Inhalt der QuelleExtracting Meaningful substructures from graphs has always been a key part in graph studies. In machine learning frameworks, supervised or unsupervised, as well as in theoretical graph analysis, finding dense subgraphs and specific decompositions is primordial in many social and biological applications among many others.In this thesis we aim at studying graph degeneracy, starting from a theoretical point of view, and building upon our results to find the most suited decompositions for the tasks at hand.Hence the first part of the thesis we work on structural results in graphs with bounded edge admissibility, proving that such graphs can be reconstructed by aggregating graphs with almost-bounded-edge-degree. We also provide computational complexity guarantees for the different degeneracy decompositions, i.e. if they are NP-complete or polynomial, depending on the length of the paths on which the given degeneracy is defined.In the second part we unify the degeneracy and admissibility frameworks based on degree and connectivity. Within those frameworks we pick the most expressive, on the one hand, and computationally efficient on the other hand, namely the 1-edge-connectivity degeneracy, to experiment on standard degeneracy tasks, such as finding influential spreaders.Following the previous results that proved to perform poorly we go back to using the k-core but plugging it in a supervised framework, i.e. graph kernels. Thus providing a general framework named core-kernel, we use the k-core decomposition as a preprocessing step for the kernel and apply the latter on every subgraph obtained by the decomposition for comparison. We are able to achieve state-of-the-art performance on graph classification for a small computational cost trade-off.Finally we design a novel degree degeneracy framework for hypergraphs and simultaneously on bipartite graphs as they are hypergraphs incidence graph. This decomposition is then applied directly to pretrained neural network architectures as they induce bipartite graphs and use the coreness of the neurons to re-initialize the neural network weights. This framework not only outperforms state-of-the-art initialization techniques but is also applicable to any pair of layers convolutional and linear thus being applicable however needed to any type of architecture
Hafidi, Hakim. „Robust machine learning for Graphs/Networks“. Electronic Thesis or Diss., Institut polytechnique de Paris, 2023. http://www.theses.fr/2023IPPAT004.
Der volle Inhalt der QuelleThis thesis addresses advancements in graph representation learning, focusing on the challengesand opportunities presented by Graph Neural Networks (GNNs). It highlights the significanceof graphs in representing complex systems and the necessity of learning node embeddings that capture both node features and graph structure. The study identifies key issues in GNNs, such as their dependence on high-quality labeled data, inconsistent performanceacross various datasets, and susceptibility to adversarial attacks.To tackle these challenges, the thesis introduces several innovative approaches. Firstly, it employs contrastive learning for node representation, enabling self-supervised learning that reduces reliance on labeled data. Secondly, a Bayesian-based classifier isproposed for node classification, which considers the graph’s structure to enhance accuracy. Lastly, the thesis addresses the vulnerability of GNNs to adversarialattacks by assessing the robustness of the proposed classifier and introducing effective defense mechanisms.These contributions aim to improve both the performance and resilience of GNNs in graph representation learning
Hérault, Laurent. „Réseaux de neurones récursifs pour l'optimisation combinatoire : application à la théorie des graphes et à la vision par ordinateur“. Grenoble INPG, 1991. http://www.theses.fr/1991INPG0019.
Der volle Inhalt der QuelleLachaud, Guillaume. „Extensions and Applications of Graph Neural Networks“. Electronic Thesis or Diss., Sorbonne université, 2023. http://www.theses.fr/2023SORUS434.
Der volle Inhalt der QuelleGraphs are used everywhere to represent interactions between entities, whether physical such as atoms, molecules or people, or more abstract such as cities, friendships, ideas, etc. Amongst all the methods of machine learning that can be used, the recent advances in deep learning have made graph neural networks the de facto standard for graph representation learning. This thesis can be divided in two parts. First, we review the theoretical underpinnings of the most powerful graph neural networks. Second, we explore the challenges faced by the existing models when training on real world graph data. The powerfulness of a graph neural network is defined in terms of its expressiveness, i.e., its ability to distinguish non isomorphic graphs; or, in an equivalent manner, its ability to approximate permutation invariant and equivariant functions. We distinguish two broad families of the most powerful models. We summarise the mathematical properties as well as the advantages and disadvantages of these models in practical situations. Apart from the choice of the architecture, the quality of the graph data plays a crucial role in the ability to learn useful representations. Several challenges are faced by graph neural networks given the intrinsic nature of graph data. In contrast to typical machine learning methods that deal with tabular data, graph neural networks need to consider not only the features of the nodes but also the interconnectedness between them. Due to the connections between nodes, training neural networks on graphs can be done in two settings: in transductive learning, the model can have access to the test features in the training phase; in the inductive setting, the test data remains unseen. We study the differences in terms of performance between inductive and transductive learning for the node classification task. Additionally, the features that are fed to a model can be noisy or even missing. In this thesis we evaluate these challenges on real world datasets, and we propose a novel architecture to perform missing data imputation on graphs. Finally, while graphs can be the natural way to describe interactions, other types of data can benefit from being converted into graphs. In this thesis, we perform preliminary work on how to extract the most important parts of skin lesion images that could be used to create graphs and learn hidden relations in the data
Pineau, Edouard. „Contributions to representation learning of multivariate time series and graphs“. Electronic Thesis or Diss., Institut polytechnique de Paris, 2020. http://www.theses.fr/2020IPPAT037.
Der volle Inhalt der QuelleMachine learning (ML) algorithms are designed to learn models that have the ability to take decisions or make predictions from data, in a large panel of tasks. In general, the learned models are statistical approximations of the true/optimal unknown decision models. The efficiency of a learning algorithm depends on an equilibrium between model richness, complexity of the data distribution and complexity of the task to solve from data. Nevertheless, for computational convenience, the statistical decision models often adopt simplifying assumptions about the data (e.g. linear separability, independence of the observed variables, etc.). However, when data distribution is complex (e.g. high-dimensional with nonlinear interactions between observed variables), the simplifying assumptions can be counterproductive. In this situation, a solution is to feed the model with an alternative representation of the data. The objective of data representation is to separate the relevant information with respect to the task to solve from the noise, in particular if the relevant information is hidden (latent), in order to help the statistical model. Until recently and the rise of modern ML, many standard representations consisted in an expert-based handcrafted preprocessing of data. Recently, a branch of ML called deep learning (DL) completely shifted the paradigm. DL uses neural networks (NNs), a family of powerful parametric functions, as learning data representation pipelines. These recent advances outperformed most of the handcrafted data in many domains.In this thesis, we are interested in learning representations of multivariate time series (MTS) and graphs. MTS and graphs are particular objects that do not directly match standard requirements of ML algorithms. They can have variable size and non-trivial alignment, such that comparing two MTS or two graphs with standard metrics is generally not relevant. Hence, particular representations are required for their analysis using ML approaches. The contributions of this thesis consist of practical and theoretical results presenting new MTS and graphs representation learning frameworks.Two MTS representation learning frameworks are dedicated to the ageing detection of mechanical systems. First, we propose a model-based MTS representation learning framework called Sequence-to-graph (Seq2Graph). Seq2Graph assumes that the data we observe has been generated by a model whose graphical representation is a causality graph. It then represents, using an appropriate neural network, the sample on this graph. From this representation, when it is appropriate, we can find interesting information about the state of the studied mechanical system. Second, we propose a generic trend detection method called Contrastive Trend Estimation (CTE). CTE learns to classify pairs of samples with respect to the monotony of the trend between them. We show that using this method, under few assumptions, we identify the true state underlying the studied mechanical system, up-to monotone scalar transform.Two graph representation learning frameworks are dedicated to the classification of graphs. First, we propose to see graphs as sequences of nodes and create a framework based on recurrent neural networks to represent and classify them. Second, we analyze a simple baseline feature for graph classification: the Laplacian spectrum. We show that this feature matches minimal requirements to classify graphs when all the meaningful information is contained in the structure of the graphs
Kalainathan, Diviyan. „Generative Neural Networks to infer Causal Mechanisms : algorithms and applications“. Thesis, Université Paris-Saclay (ComUE), 2019. http://www.theses.fr/2019SACLS516.
Der volle Inhalt der QuelleCausal discovery is of utmost importance for agents who must plan, reason and decide based on observations; where mistaking correlation with causation might lead to unwanted consequences. The gold standard to discover causal relations is to perform experiments.However, experiments are in many cases expensive, unethical, or impossible to realize. In these situations, there is a need for observational causal discovery, that is, the estimation of causal relations from observations alone.Causal discovery in the observational data setting traditionally involves making significant assumptions on the data and on the underlying causal model.This thesis aims to alleviate some of the assumptions made on the causal models by exploiting the modularity and expressiveness of neural networks for causal discovery, leveraging both conditional independences and simplicity of the causal mechanisms through two algorithms.Extensive experiments on both simulated and real-world data and a throughout theoretical anaylsis prove the good performance and the soundness of the proposed approaches
Boulnois, Philippe. „Contribution à l'étude de différentes architectures de réseaux de neurones artificiels réalisant une transcription graphèmes-phonèmes pour le français“. Compiègne, 1994. http://www.theses.fr/1994COMPD675.
Der volle Inhalt der QuelleFaucheux, Cyrille. „Segmentation supervisée d'images texturées par régularisation de graphes“. Thesis, Tours, 2013. http://www.theses.fr/2013TOUR4050/document.
Der volle Inhalt der QuelleIn this thesis, we improve a recent image segmentation algorithm based on a graph regularization process. The goal of this method is to compute an indicator function that satisfies a regularity and a fidelity criteria. Its particularity is to represent images with similarity graphs. This data structure allows relations to be established between similar pixels, leading to non-local processing of the data. In order to improve this approach, combine it with another non-local one: the texture features. Two solutions are developped, both based on Haralick features. In the first one, we propose a new fidelity term which is based on the work of Chan and Vese and is able to evaluate the homogeneity of texture features. In the second method, we propose to replace the fidelity criteria by the output of a supervised classifier. Trained to recognize several textures, the classifier is able to produce a better modelization of the problem by identifying the most relevant texture features. This method is also extended to multiclass segmentation problems. Both are applied to 2D and 3D textured images
Hubert, Nicolas. „Mesure et enrichissement sémantiques des modèles à base d'embeddings pour la prédiction de liens dans les graphes de connaissances“. Electronic Thesis or Diss., Université de Lorraine, 2024. http://www.theses.fr/2024LORR0059.
Der volle Inhalt der QuelleKnowledge graph embedding models (KGEMs) have gained considerable traction in recent years. These models learn a vector representation of knowledge graph entities and relations, a.k.a. knowledge graph embeddings (KGEs). This thesis specifically explores the advancement of KGEMs for the link prediction (LP) task, which is of utmost importance as it underpins several downstream applications such as recommender systems. In this thesis, various challenges around the use of KGEMs for LP are identified: the scarcity of semantically rich resources, the unidimensional nature of evaluation frameworks, and the lack of semantic considerations in prevailing machine learning-based approaches. Central to this thesis is the proposition of novel solutions to these challenges. Firstly, the thesis contributes to the development of semantically rich resources: mainstream datasets for link prediction are enriched using schema-based information, EducOnto and EduKG are proposed to overcome the paucity of resources in the educational domain, and PyGraft is introduced as an innovative open-source tool for generating synthetic ontologies and knowledge graphs. Secondly, the thesis proposes a new semantic-oriented evaluation metric, Sem@K, offering a multi-dimensional perspective on model performance. Importantly, popular models are reassessed using Sem@K, which reveals essential insights into their respective capabilities and highlights the need for multi-faceted evaluation frameworks. Thirdly, the thesis delves into the development of neuro-symbolic approaches, transcending traditional machine learning paradigms. These approaches do not only demonstrate improved semantic awareness but also extend their utility to diverse applications such as recommender systems. In summary, the present work not only redefines the evaluation and functionality of knowledge graph embedding models but also sets the stage for more versatile, interpretable AI systems, underpinning future explorations at the intersection of machine learning and symbolic reasoning
Pasdeloup, Bastien. „Extending convolutional neural networks to irregular domains through graph inference“. Thesis, Ecole nationale supérieure Mines-Télécom Atlantique Bretagne Pays de la Loire, 2017. http://www.theses.fr/2017IMTA0048/document.
Der volle Inhalt der QuelleThis manuscript sums up our work on extending convolutional neuralnetworks to irregular domains through graph inference. It consists of three main chapters, each giving the details of a part of a methodology allowing the definition of such networks to process signals evolving on graphs with unknown structures.First, graph inference from data is explored, in order to provide a graph modeling the support of the signals to classify. Second, translation operators that preserve neighborhood properties of the vertices are identified on the inferred graph. Third, these translations are used to shift a convolutional kernel on the graph in order to define a convolutional neural network that is adapted to the input data.We have illustrated our methodology on a dataset of images. While not using any particular knowledge on the signals, we have been able to infer a graph that is close to a grid. Translations on this graph resemble Euclidean translations. Therefore, this has allowed us to define an adapted convolutional neural network that is very close what one would obtain when using the information that signals are images. This network, trained on the initial data, has out performed state of the art methods by more than 13 points, while using a very simple and easily improvable architecture.The method we have introduced is a generalization of convolutional neural networks. As a matter of fact, they can be seen as aparticularization of our approach in the case where the graph is a grid. Our work thus opens the way to numerous perspectives, as it provides an efficient way to build networks that are adapted to the data
Rosar, Kós Lassance Carlos Eduardo. „Graphs for deep learning representations“. Thesis, Ecole nationale supérieure Mines-Télécom Atlantique Bretagne Pays de la Loire, 2020. http://www.theses.fr/2020IMTA0204.
Der volle Inhalt der QuelleIn recent years, Deep Learning methods have achieved state of the art performance in a vast range of machine learning tasks, including image classification and multilingual automatic text translation. These architectures are trained to solve machine learning tasks in an end-to-end fashion. In order to reach top-tier performance, these architectures often require a very large number of trainable parameters. There are multiple undesirable consequences, and in order to tackle these issues, it is desired to be able to open the black boxes of deep learning architectures. Problematically, doing so is difficult due to the high dimensionality of representations and the stochasticity of the training process. In this thesis, we investigate these architectures by introducing a graph formalism based on the recent advances in Graph Signal Processing (GSP). Namely, we use graphs to represent the latent spaces of deep neural networks. We showcase that this graph formalism allows us to answer various questions including: ensuring generalization abilities, reducing the amount of arbitrary choices in the design of the learning process, improving robustness to small perturbations added to the inputs, and reducing computational complexity
Bekkouch, Imad Eddine Ibrahim. „Auxiliary learning & Adversarial training pour les études des manuscrits médiévaux“. Electronic Thesis or Diss., Sorbonne université, 2024. http://www.theses.fr/2024SORUL014.
Der volle Inhalt der QuelleThis thesis is at the intersection of musicology and artificial intelligence, aiming to leverage AI to help musicologists with repetitive work, such as object searching in the museum's manuscripts. We annotated four new datasets for medieval manuscript studies: AMIMO, AnnMusiconis, AnnVihuelas, and MMSD. In the second part, we improve object detectors' performances using Transfer learning techniques and Few Shot Object Detection.In the third part, we discuss a powerful approach to Domain Adaptation, which is auxiliary learning, where we train the model on the target task and an extra task that allows for better stabilization of the model and reduces over-fitting.Finally, we discuss self-supervised learning, which does not use extra meta-data by leveraging the adversarial learning approach, forcing the model to extract domain-independent features
Osman, Ousama. „Méthodes de diagnostic en ligne, embarqué et distribué dans les réseaux filaires complexes“. Thesis, Université Clermont Auvergne (2017-2020), 2020. http://www.theses.fr/2020CLFAC038.
Der volle Inhalt der QuelleThe research conducted in this thesis focuses on the diagnosis of complex wired networks using distributed reflectometry. It aims to develop new distributed diagnostic techniques for complex networks that allow data fusion as well as communication between reflectometers to detect, locate and characterize electrical faults (soft and hard faults). This collaboration between reflectometers solves the problem of fault location ambiguity and improves the quality of diagnosis. The first contribution is the development of a graph theory-based method for combining data between distributed reflectometers, thus facilitating the location of the fault. Then, the amplitude of the reflected signal is used to identify the type of fault and estimate its impedance. The latter is based on the regeneration of the signal by compensating for the degradation suffered by the diagnosis signal during its propagation through the network. The second contribution enables data fusion between distributed reflectometers in complex networks affected by multiple faults. To achieve this objective, two methods have been proposed and developed: the first is based on genetic algorithms (GA) and the second is based on neural networks (RN). These tools combined with distributed reflectometryallow automatic detection, location, and characterization of several faults in different types and topologies of wired networks. The third contribution proposes the use of information-carrying diagnosis signal to integrate communication between distributed reflectometers. It properly uses the phases of the MCTDR multi-carrier signal to transmit data. This communication ensures the exchange of useful information (such as fault location and amplitude) between reflectometers on the state of the cables, thus enabling data fusion and unambiguous fault location. Interference problems between the reflectometers are also addressed when they simultaneously inject their test signals into the network. These studies illustrate the efficiency and applicability of the proposed methods. They also demonstrate their potential to improve the performance of the current wired diagnosis systems to meet the need and the problem of detecting and locating faults that manufacturers and users face today in electrical systems to improve their operational safety
Chen, Dexiong. „Modélisation de données structurées avec des machines profondes à noyaux et des applications en biologie computationnelle“. Thesis, Université Grenoble Alpes, 2020. http://www.theses.fr/2020GRALM070.
Der volle Inhalt der QuelleDeveloping efficient algorithms to learn appropriate representations of structured data, including sequences or graphs, is a major and central challenge in machine learning. To this end, deep learning has become popular in structured data modeling. Deep neural networks have drawn particular attention in various scientific fields such as computer vision, natural language understanding or biology. For instance, they provide computational tools for biologists to possibly understand and uncover biological properties or relationships among macromolecules within living organisms. However, most of the success of deep learning methods in these fields essentially relies on the guidance of empirical insights as well as huge amounts of annotated data. Exploiting more data-efficient models is necessary as labeled data is often scarce.Another line of research is kernel methods, which provide a systematic and principled approach for learning non-linear models from data of arbitrary structure. In addition to their simplicity, they exhibit a natural way to control regularization and thus to avoid overfitting.However, the data representations provided by traditional kernel methods are only defined by simply designed hand-crafted features, which makes them perform worse than neural networks when enough labeled data are available. More complex kernels inspired by prior knowledge used in neural networks have thus been developed to build richer representations and thus bridge this gap. Yet, they are less scalable. By contrast, neural networks are able to learn a compact representation for a specific learning task, which allows them to retain the expressivity of the representation while scaling to large sample size.Incorporating complementary views of kernel methods and deep neural networks to build new frameworks is therefore useful to benefit from both worlds.In this thesis, we build a general kernel-based framework for modeling structured data by leveraging prior knowledge from classical kernel methods and deep networks. Our framework provides efficient algorithmic tools for learning representations without annotations as well as for learning more compact representations in a task-driven way. Our framework can be used to efficiently model sequences and graphs with simple interpretation of predictions. It also offers new insights about designing more expressive kernels and neural networks for sequences and graphs
Tiano, Donato. „Learning models on healthcare data with quality indicators“. Electronic Thesis or Diss., Lyon 1, 2022. http://www.theses.fr/2022LYO10182.
Der volle Inhalt der QuelleTime series are collections of data obtained through measurements over time. The purpose of this data is to provide food for thought for event extraction and to represent them in an understandable pattern for later use. The whole process of discovering and extracting patterns from the dataset is carried out with several extraction techniques, including machine learning, statistics, and clustering. This domain is then divided by the number of sources adopted to monitor a phenomenon. Univariate time series when the data source is single and multivariate time series when the data source is multiple. The time series is not a simple structure. Each observation in the series has a strong relationship with the other observations. This interrelationship is the main characteristic of time series, and any time series extraction operation has to deal with it. The solution adopted to manage the interrelationship is related to the extraction operations. The main problem with these techniques is that they do not adopt any pre-processing operation on the time series. Raw time series have many undesirable effects, such as noisy points or the huge memory space required for long series. We propose new data mining techniques based on the adoption of the most representative features of time series to obtain new models from the data. The adoption of features has a profound impact on the scalability of systems. Indeed, the extraction of a feature from the time series allows for the reduction of an entire series to a single value. Therefore, it allows for improving the management of time series, reducing the complexity of solutions in terms of time and space. FeatTS proposes a clustering method for univariate time series that extracts the most representative features of the series. FeatTS aims to adopt the features by converting them into graph networks to extract interrelationships between signals. A co-occurrence matrix merges all detected communities. The intuition is that if two time series are similar, they often belong to the same community, and the co-occurrence matrix reveals this. In Time2Feat, we create a new multivariate time series clustering. Time2Feat offers two different extractions to improve the quality of the features. The first type of extraction is called Intra-Signal Features Extraction and allows to obtain of features from each signal of the multivariate time series. Inter-Signal Features Extraction is used to obtain features by considering pairs of signals belonging to the same multivariate time series. Both methods provide interpretable features, which makes further analysis possible. The whole time series clustering process is lighter, which reduces the time needed to obtain the final cluster. Both solutions represent the state of the art in their field. In AnomalyFeat, we propose an algorithm to reveal anomalies from univariate time series. The characteristic of this algorithm is the ability to work among online time series, i.e. each value of the series is obtained in streaming. In the continuity of previous solutions, we adopt the functionality of revealing anomalies in the series. With AnomalyFeat, we unify the two most popular algorithms for anomaly detection: clustering and recurrent neural network. We seek to discover the density area of the new point obtained with clustering
Elagouni, Khaoula. „Combining neural-based approaches and linguistic knowledge for text recognition in multimedia documents“. Thesis, Rennes, INSA, 2013. http://www.theses.fr/2013ISAR0013/document.
Der volle Inhalt der QuelleThis thesis focuses on the recognition of textual clues in images and videos. In this context, OCR (optical character recognition) systems, able to recognize caption texts as well as natural scene texts captured anywhere in the environment have been designed. Novel approaches, robust to text variability (differentfonts, colors, sizes, etc.) and acquisition conditions (complex background, non uniform lighting, low resolution, etc.) have been proposed. In particular, two kinds of methods dedicated to text recognition are provided:- A segmentation-based approach that computes nonlinear separations between characters well adapted to the localmorphology of images;- Two segmentation-free approaches that integrate a multi-scale scanning scheme. The first one relies on a graph model, while the second one uses a particular connectionist recurrent model able to handle spatial constraints between characters.In addition to the originalities of each approach, two extra contributions of this work lie in the design of a character recognition method based on a neural classification model and the incorporation of some linguistic knowledge that enables to take into account the lexical context.The proposed OCR systems were tested and evaluated on two datasets: a caption texts video dataset and a natural scene texts dataset (namely the public database ICDAR 2003). Experiments have demonstrated the efficiency of our approaches and have permitted to compare their performances to those of state-of-the-art methods, highlighting their advantages and limits
Wang, Lianfa. „Improving the confidence of CFD results by deep learning“. Electronic Thesis or Diss., Université Paris sciences et lettres, 2024. http://www.theses.fr/2024UPSLM008.
Der volle Inhalt der QuelleComputational Fluid Dynamics (CFD) has become an indispensable tool for studying complex flow phenomena in both research and industry over the years. The accuracy of CFD simulations depends on various parameters – geometry, mesh, schemes, solvers, etc. – as well as phenomenological knowledge that only an expert CFD engineer can configure and optimize. The objective of this thesis is to propose an AI assistant to help users, whether they are experts or not, to better choose simulation options and ensure the reliability of results for a target flow phenomenon. In this context, deep learning algorithms are explored to identify the characteristics of flows computed on structured and unstructured meshes of complex geometries. Initially, convolutional neural networks (CNNs), known for their ability to extract patterns from im-ages, are used to identify flow phenomena such as vortices and thermal stratification on structured 2D meshes. Although the results obtained on structured meshes are satisfactory, CNNs can only be applied to structured meshes. To overcome this limitation, a graph-based neural network (GNN) framework is proposed. This framework uses the U-Net architecture and a hierarchy of successively refined graphs through the implementation of a multigrid method (AMG) inspired by the one used in the Code_Saturne CFD code. Subsequently, an in-depth study of kernel functions was conducted according to identification accuracy and training efficiency criteria to better filter the different phenomena on unstructured meshes. After comparing available kernel functions in the literature, a new kernel function based on the Gaussian mixture model was proposed. This function is better suited to identifying flow phenomena on unstructured meshes. The superiority of the proposed architecture and kernel function is demonstrated by several numerical experiments identifying 2D vortices and its adaptability to identifying the characteristics of a 3D flow
Tardif, Malo. „Proximal sensing and neural network processes to assist in diagnosis of multi-symptom grapevine diseases“. Electronic Thesis or Diss., Bordeaux, 2023. http://www.theses.fr/2023BORD0369.
Der volle Inhalt der QuelleGrapevine is a plant susceptible to numerous diseases. Some of these diseases can lead to significant yield losses and the death of the infected grapevine. Among these diseases, some present symptoms of different nature on various organs of the same vine. Their diagnosis, typically performed by experts, is even more complex as many confounding factors are present. This research focuses on the development of methodologies for acquiring, annotating, and processing data related to multi-symptom grapevine diseases to study their automated diagnosis. Two groups of diseases are targeted: grapevine yellows such as Flavescence dorée (FD) and grapevine trunk diseases (GTDs) with Eutypa and Botryosphaeria diebacks as specific diseases.RGB image acquisitions were conducted directly in grapevine rows to build datasets for each disease type. The dataset for FD covers five different grape varieties and takes into account many diseases that have symptoms similar to FD, referred to as confounding diseases. The GTDs dataset includes images of a single grape variety and no confounding disease. Three methods for the automatic diagnosis of these diseases are proposed, compared, and discussed. The first method, inspired by state-of-the-art techniques, uses a convolutional neural network-based classifier applied to raw images (method A). The results show that this methodology delivers good results on datasets containing very few confounding diseases. Precision (p) and recall (r) of (p=0.94, r=0.92) are achieved for classifying images of grapevines affected by GTDs, while they are (p=0.87, r=0.84) for classifying images of vines affected by FD in a dataset containing 16% of confounding disease images.To improve these results, two methods were developed, both consisting of two steps: (1) individual symptom detection using a detection algorithm composed of neural convolutional layers and a neural segmentation algorithm; (2) diagnosis based on the association of detected symptoms, either using a Random Forest classifier (method B) or a graph neural network (method C). The results of these two methodologies on the dataset containing 16% of confounding disease images for FD are (p=0.86, r=0.90) for method B and (p=0.90, r=0.96) for method C. These results demonstrate the better effectiveness of two-step methodologies in distinguishing confounding diseases from targeted diseases. They also highlight the relevance of embedded RGB imaging combined with artificial intelligence approaches for diagnosing these diseases.Finally, these three methods are tested on whole-block acquisitions to establish their validity in real-world use cases. The results highlight the advantages of the two-step methodology based on symptom association by graph, the significant contribution of considering the surrounding vines and both sides of the vines during their automated diagnosis, and emphasize the challenges of real-world application of these methodologies
Riva, Mateus. „Spatial Relational Reasoning in Machine Learning : Deep Learning and Graph Clustering“. Electronic Thesis or Diss., Institut polytechnique de Paris, 2022. http://www.theses.fr/2022IPPAT043.
Der volle Inhalt der QuelleThis thesis studies the capabilities of machine learning methods for reasoning on spatial relationships, with a particular focus on directional relationships, and the use of prior relational information by these methods. There are many works in the field of applying knowledge on relationships to machine learning methods. However, this body of work still leaves several open questions. Throughout this thesis, we explore, investigate and attempt to explain different research questions linked to this field.We propose an improvement to the training of CNNs via a regularisation loss function based on relational information. To this end, we propose two novel loss functions which reward relationship satisfaction during CNN training, and design synthetic experiments to showcase their impact. While the proposed loss functions show improvements over an unmodified baseline in specific, strict synthetic scenarios, the impact on more ``generic'' training scenarios is less significant. This result is not easily explainable, as neural network training is a significantly opaque process, and as such, a deeper exploration is required to understand how a CNN learns (or fails to learn) to reason using relational information.To further understanding of how a CNN can learn to reason using relational information, we propose a wide array of distinct synthetic experiments. We explore the processes which enable, facilitate, or hinder ``standard'' CNN reasoning on relationships. We propose a fundamental experience to demonstrate that a basic, unmodified CNN is capable of relational reasoning in some scenarios. Next, we explore which relationships are learned by the CNN, by performing inference on scenes where the prior relationships are disturbed, by recording the difference in the results, and by training and testing CNNs on synthetic data with more or less relationships available. We then investigate the limits placed on relational reasoning by the network receptive field, as well as deepen our analysis on situations where the amount of training data is insufficient. Finally, we explore at which moment during training relationships are satisfied, as a proxy for understanding at which moment the relationships themselves are learned.Following a graph-clustering approach to the usage of relational information, we explore prior relationships in a different machine learning context, that of community discovery on graphs. We formulate graph clustering as an inexact matching problem between the graph to be clustered and a model graph which encodes prior knowledge on how the communities or clusters relate to each other. We compare this approach with traditional graph clustering approaches on a set of synthetic graphs, to showcase the advantages of a relational-aware approach, as well as on real graphs
Ambroise, Corentin. „Structure-aware neural networks in the study of multi-modal population cohorts : an application to mental health“. Electronic Thesis or Diss., université Paris-Saclay, 2024. http://www.theses.fr/2024UPAST065.
Der volle Inhalt der QuelleIt is currently acknowledged that relying solely on conventional classification strategies from a single data source is not effective to understand, diagnose or prognose psychiatric syndromes. The classification targets simply rely on clinician labels that alone do not express a very large variability. In 2009, the Research Domain Criteria (RDoC) recommended a more comprehensive approach to study psychiatric disorders by incorporating diverse data types that cover various levels of life organization (e.g., imaging, genetic, symptoms). The proposal suggests that a thorough description of a pathology requires consideration of multiple dimensions, which may be shared across different psychiatric syndromes and even contribute to non-pathological variability. Efficient frameworks for unsupervised learning, specifically designed for multivariate and multimodal approaches, are anticipated to offer methodologies for handling and integrating the kind of datasets advocated by the RDoC. Deep learning allows us to learn in multimodal settings with modality-specific structure and intermodality correlation structure.To model intra-modality structure, we use specific convolutional neural networks that enable to learn from cortical brain measures distributed across a spherical mesh and thus reveal original biomarkers. In this context, we propose 5 data augmentations and apply them in one of the many novel self-supervised learning schemes relying moslty on data augmentation. This work allows deep representation learning to properly initialize network on huge healthy patient cohorts and then transfer them to study clinical pathology of interest in smaller cohorts.On the other hand, we have identified multi-view variational auto encoders as good candidates to integrate multiple modalities. Moreover, we challenge the common assumption that neural networks are not interpretable. We use a digital avatar procedure as an interpretability module capable of reporting the inter-view relationships learned within a multi-view autoencoder. In particular, we integrate this procedure into a novel framework that combines multiple interpretations and utilizes stability selection to identify meaningful and reproducible associations between brain-imaging modalities and behaviour. We apply this framework to exhibit specific brain-behaviour associations present in the transdiagnostic cohort Healthy Brain Network (HBN). The identified brain-behaviour associations establish connections between regional cortical features from structural magnetic resonance imaging and electronic clinical record forms assessing psychiatric symptoms. We show this framework is able to find relevant and stable associations
Qian, Yang. „Conception et Commande d'un Robot d'Assistance à la Personne“. Phd thesis, Ecole Centrale de Lille, 2013. http://tel.archives-ouvertes.fr/tel-00864692.
Der volle Inhalt der QuelleQian, Yang. „Conception et Commande d’un Robot d’Assistance à la Personne“. Thesis, Ecole centrale de Lille, 2013. http://www.theses.fr/2013ECLI0005/document.
Der volle Inhalt der QuelleThe purpose of this thesis is to design, model and control of a personal assistant robot used for domestic tasks. In order to make the robot’s design more efficient, a virtual simulation system is built using dynamic simulation software. The kinematic model is set up based on modified D-H principle. The dynamic model is built using the Lagrange theorem and elaborated in Matlab. We also employ an energy-based approach for modeling and its bond graph notation ensures encapsulation of functionality, extendibility and reusability of each element of the model. A hybrid algorithm of combining the Jacobian pseudoinverse algorithm with Rapidly-Exploring Random Tree method is presented for collision-free path planning of a redundant manipulator. An intelligent robust controller based on neural network is introduced for the coordinated control of a mobile manipulator. This method does not require an accurate model of the robot. Unknown dynamic parameters of the mobile platform and the manipulator are identified and compensated in closed-loop control using RBF neural network. A similar control algorithm is presented for coordinated force/motion control of a mobile manipulator suffering both holonomic and nonholonomic constraints. Kinematics and dynamics of a dexterous hand manipulating an object with known shape by rolling contacts are derived. A computed torque control algorithm is presented to ensure firm grip, avoid slippage and well track a given motion imposed to the object. The validation of models and different control laws were made by the co-simulation Matlab / virtual model
Cattai, Tiziana. „Leveraging brain connectivity networks to detect mental states during motor imagery“. Electronic Thesis or Diss., Sorbonne université, 2021. http://www.theses.fr/2021SORUS081.
Der volle Inhalt der QuelleThe brain is a complex network and we know that inter-areal synchronization and de-synchronization mechanisms are crucial to perform motor and cognitive tasks. Nowadays, brain functional interactions are studied in brain-computer interface BCI) applications with more and more interest. This might have strong impact on BCI systems, typically based on univariate features which separately characterize brain regional activities. Indeed, brain connectivity features can be used to develop alternative BCIs in an effort to improve performance and to extend their real-life applicability. The ambition of this thesis is the investigation of brain functional connectivity networks during motor imagery (MI)-based BCI tasks. It aims to identify complex brain functioning, re-organization processes and time-varying dynamics, at both group and individual level. This thesis presents different developments that sequentially enrich an initially simple model in order to obtain a robust method for the study of functional connectivity networks. Experimental results on simulated and real EEG data recorded during BCI tasks prove that our proposed method well explains the variegate behaviour of brain EEG data. Specifically, it provides a characterization of brain functional mechanisms at group level, together with a measure of the separability of mental conditions at individual level. We also present a graph denoising procedure to filter data which simultaneously preserve the graph connectivity structure and enhance the signal-to-noise ratio. Since the use of a BCI system requires a dynamic interaction between user and machine, we finally propose a method to capture the evolution of time-varying data. In essence, this thesis presents a novel framework to grasp the complexity of graph functional connectivity during cognitive tasks
Wenzek, Didier. „Construction de réseaux de neurones“. Phd thesis, Grenoble INPG, 1993. http://tel.archives-ouvertes.fr/tel-00343569.
Der volle Inhalt der QuelleTsopze, Norbert. „Treillis de Galois et réseaux de neurones : une approche constructive d'architecture des réseaux de neurones“. Thesis, Artois, 2010. http://www.theses.fr/2010ARTO0407/document.
Der volle Inhalt der QuelleThe artificial neural networks are successfully applied in many applications. But theusers are confronted with two problems : defining the architecture of the neural network able tosolve their problems and interpreting the network result. Many research works propose some solutionsabout these problems : to find out the architecture of the network, some authors proposeto use the problem domain theory and deduct the network architecture and some others proposeto dynamically add neurons in the existing networks until satisfaction. For the interpretabilityproblem, solutions consist to extract rules which describe the network behaviour after training.The contributions of this thesis concern these problems. The thesis are limited to the use of theartificial neural networks in solving the classification problem.In this thesis, we present a state of art of the existing methods of finding the neural networkarchitecture : we present a theoritical and experimental study of these methods. From this study,we observe some limits : difficulty to use some method when the knowledges are not available ;and the network is seem as ’black box’ when using other methods. We a new method calledCLANN (Concept Lattice-based Artificial Neural Network) which builds from the training dataa semi concepts lattice and translates this semi lattice into the network architecture. As CLANNis limited to the two classes problems, we propose MCLANN which extends CLANN to manyclasses problems.A new method of rules extraction called ’MaxSubsets Approach’ is also presented in thisthesis. Its particularity is the possibility of extracting the two kind of rules (If then and M-of-N)from an internal structure.We describe how to explain the MCLANN built network result aboutsome inputs
Zakroum, Mehdi. „Machine Learning for the Automation of Cyber-threat Monitoring and Inference“. Electronic Thesis or Diss., Université de Lorraine, 2023. http://www.theses.fr/2023LORR0108.
Der volle Inhalt der QuelleOver the past few decades, cyber-threats have known a significant increase and continue to grow exponentially. Network operators and security practitioners are constantly striving to automate their defense strategies against large-scale cyber incidents and smaller-scale peculiar events targeting their networks. Improving the monitoring of security events and detecting attacks at an early stage are key features to prevent against eventual damages or at least alleviate their impact. The traffic captured by network sensors such as network telescopes, also known as darknets, constitute a rich source of cybersecurity intelligence. The data recorded by such sensors include different types of traffic ranging from benign traffic like regular scans performed by organizations for statistical purpose, to malicious cyber incidents like worms spread, vulnerability scans, and backscatter packets that come as a side effect spoofed source of Denial of Service attacks. These data could be leveraged to automate and improve cyber-threat monitoring solutions and attack modeling and prediction. To this end, this thesis combines research works on the salient topics of cyber-threat monitoring and cyber-attack classification and forecasting
Voegtlin, Thomas. „Réseaux de neurones et auto-référence“. Lyon 2, 2002. http://theses.univ-lyon2.fr/documents/lyon2/2002/voegtlin_t.
Der volle Inhalt der QuelleThe purpose of this thesis is to present a class of unsupervised learning algorithms for recurrent networks. In the first part (chapters 1 to 4), I propose a new approach to this question, based on a simple principle: self-reference. A self-referent algorithm is not based on the minimization of an objective criterion, such as an error function, but on a subjective function, that depends on what the network has previously learned. An example of a supervised recurrent network where learning is self-referent is the Simple Recurrent Network (SRN) by Elman (1990). In the SRN, self-reference is applied to the supervised error back-propagation algorithm. In this aspect, the SRN differs from other generalizations of back-propagation to recurrent networks, that use an objective criterion, such as Back-Propagation Through Time, or Real-Time Recurrent Learning. In this thesis, I show that self-reference can be combined with several well-known unsupervised learning methods: the Self-Organizing Map (SOM), Principal Components Analysis (PCA), and Independent Components Analysis (ICA). These techniques are classically used to represent static data. Self-reference allows one to generalize these techniques to time series, and to define unsupervised learning algorithms for recurrent networks
Teytaud, Olivier. „Apprentissage, réseaux de neurones et applications“. Lyon 2, 2001. http://theses.univ-lyon2.fr/documents/lyon2/2001/teytaud_o.
Der volle Inhalt der QuelleCôté, Marc-Alexandre. „Réseaux de neurones génératifs avec structure“. Thèse, Université de Sherbrooke, 2017. http://hdl.handle.net/11143/10489.
Der volle Inhalt der QuelleJodouin, Jean-François. „Réseaux de neurones et traitement du langage naturel : étude des réseaux de neurones récurrents et de leurs représentations“. Paris 11, 1993. http://www.theses.fr/1993PA112079.
Der volle Inhalt der QuelleBrette, Romain. „Modèles Impulsionnels de Réseaux de Neurones Biologiques“. Phd thesis, Université Pierre et Marie Curie - Paris VI, 2003. http://tel.archives-ouvertes.fr/tel-00005340.
Der volle Inhalt der QuelleTardif, Patrice. „Autostructuration des réseaux de neurones avec retards“. Thesis, Université Laval, 2007. http://www.theses.ulaval.ca/2007/24240/24240.pdf.
Der volle Inhalt der QuelleMaktoobi, Sheler. „Couplage diffractif pour réseaux de neurones optiques“. Thesis, Bourgogne Franche-Comté, 2020. http://www.theses.fr/2020UBFCD019.
Der volle Inhalt der QuellePhotonic networks with high performance can be considered as substrates for future computing systems. In comparison with electronics, photonic systems have substantial privileges, for instance the possibility of a fully parallel implementation of networks. Recently, neural networks have moved into the center of attention of the photonic community. One of the most important requirements for parallel large-scale photonic networks is to realize the connectivities. Diffraction is considered as a method to process the connections between the nodes (coupling) in optical neural networks. In the current thesis, we evaluate the scalability of a diffractive coupling in more details as follow:First, we begin with a general introductions for artificial intelligence, machine learning, artificial neural network and photonic neural networks. To establish a working neural network, learning rules are an essential part to optimize a configuration for obtaining a low error from the system, hence learning rules are introduced (Chapter 1). We investigate the fundamental concepts of diffractive coupling in our spatio-temporal reservoir. In that case, theory of diffraction is explained. We use an analytical scheme to provide the limits for the size of diffractive networks which is a part of our photonic neural network (Chapter 2). The concepts of diffractive coupling are investigated experimentally by two different experiments to confirm the analytical limits and to obtain maximum number of nodes which can be coupled in the photonic network (Chapter 3). Numerical simulations for such an experimental setup is modeled in two different schemes to obtain the maximum size of network numerically, which approaches a surface of 100 mm2 (Chapter 4). Finally, the complete photonic neural network is demonstrated. We design a spatially extended reservoir for 900 nodes. Consequently, our system generalizes the prediction for the chaotic Mackey–Glass sequence (Chapter 5)
Ouali, Jamel. „Architecture intégrée flexible pour réseaux de neurones“. Grenoble INPG, 1991. http://www.theses.fr/1991INPG0035.
Der volle Inhalt der QuelleTogni, Olivier. „Force des graphes : indice optique des réseaux“. Bordeaux 1, 1998. http://www.theses.fr/1998BOR10596.
Der volle Inhalt der QuelleBigot, Pascal. „Utilisation des réseaux de neurones pour la télégestion des réseaux techniques urbains“. Lyon 1, 1995. http://www.theses.fr/1995LYO10036.
Der volle Inhalt der QuelleFraisse, Pierre. „Cycles et facteurs dans les graphes : application de la théorie des graphes aux réseaux de Pétri“. Paris 11, 1985. http://www.theses.fr/1985PA112100.
Der volle Inhalt der QuelleThis thesis is constituted by several chapters. The first one deals with the existence of certain cycles in graphs of large degree. It gives sufficient conditions for the existence of cycles of length greater than a given number m, of dominating cycles, of circuits containing a set of s vertices and of length at most 2s. The second one gives sufficient conditions for the existence of f-factors in graphs, with conditions of independence number and connectivity, of number of edges, and also by assuming the existence of f-factors in some subgraphs. The third deals with covering of edges and vertices of a graph by cycles, the sum of the length of the cycles being minimal. Finally, the fourth is an application of the graph theory to Petri nets. It gives sufficient conditions for liveness on a class of nets
Koiran, Pascal. „Puissance de calcul des réseaux de neurones artificiels“. Lyon 1, 1993. http://www.theses.fr/1993LYO19003.
Der volle Inhalt der QuelleGraïne, Slimane. „Inférence grammaticale régulière par les réseaux de neurones“. Paris 13, 1994. http://www.theses.fr/1994PA132020.
Der volle Inhalt der QuelleLe, Fablec Yann. „Prévision de trajectoires d'avions par réseaux de neurones“. Toulouse, INPT, 1999. http://www.theses.fr/1999INPT034H.
Der volle Inhalt der QuelleCorne, Christophe. „Parallélisation de réseaux de neurones sur architecture distribuée“. Mulhouse, 1999. http://www.theses.fr/1999MULH0583.
Der volle Inhalt der QuelleFernandez, Brillet Lucas. „Réseaux de neurones CNN pour la vision embarquée“. Thesis, Université Grenoble Alpes, 2020. http://www.theses.fr/2020GRALM043.
Der volle Inhalt der QuelleRecently, Convolutional Neural Networks have become the state-of-the-art soluion(SOA) to most computer vision problems. In order to achieve high accuracy rates, CNNs require a high parameter count, as well as a high number of operations. This greatly complicates the deployment of such solutions in embedded systems, which strive to reduce memory size. Indeed, while most embedded systems are typically in the range of a few KBytes of memory, CNN models from the SOA usually account for multiple MBytes, or even GBytes in model size. Throughout this thesis, multiple novel ideas allowing to ease this issue are proposed. This requires to jointly design the solution across three main axes: Application, Algorithm and Hardware.In this manuscript, the main levers allowing to tailor computational complexity of a generic CNN-based object detector are identified and studied. Since object detection requires scanning every possible location and scale across an image through a fixed-input CNN classifier, the number of operations quickly grows for high-resolution images. In order to perform object detection in an efficient way, the detection process is divided into two stages. The first stage involves a region proposal network which allows to trade-off recall for the number of operations required to perform the search, as well as the number of regions passed on to the next stage. Techniques such as bounding box regression also greatly help reduce the dimension of the search space. This in turn simplifies the second stage, since it allows to reduce the task’s complexity to the set of possible proposals. Therefore, parameter counts can greatly be reduced.Furthermore, CNNs also exhibit properties that confirm their over-dimensionment. This over-dimensionement is one of the key success factors of CNNs in practice, since it eases the optimization process by allowing a large set of equivalent solutions. However, this also greatly increases computational complexity, and therefore complicates deploying the inference stage of these algorithms on embedded systems. In order to ease this problem, we propose a CNN compression method which is based on Principal Component Analysis (PCA). PCA allows to find, for each layer of the network independently, a new representation of the set of learned filters by expressing them in a more appropriate PCA basis. This PCA basis is hierarchical, meaning that basis terms are ordered by importance, and by removing the least important basis terms, it is possible to optimally trade-off approximation error for parameter count. Through this method, it is possible to compress, for example, a ResNet-32 network by a factor of ×2 both in the number of parameters and operations with a loss of accuracy <2%. It is also shown that the proposed method is compatible with other SOA methods which exploit other CNN properties in order to reduce computational complexity, mainly pruning, winograd and quantization. Through this method, we have been able to reduce the size of a ResNet-110 from 6.88Mbytes to 370kbytes, i.e. a x19 memory gain with a 3.9 % accuracy loss.All this knowledge, is applied in order to achieve an efficient CNN-based solution for a consumer face detection scenario. The proposed solution consists of just 29.3kBytes model size. This is x65 smaller than other SOA CNN face detectors, while providing equal detection performance and lower number of operations. Our face detector is also compared to a more traditional Viola-Jones face detector, exhibiting approximately an order of magnitude faster computation, as well as the ability to scale to higher detection rates by slightly increasing computational complexity.Both networks are finally implemented in a custom embedded multiprocessor, verifying that theorical and measured gains from PCA are consistent. Furthermore, parallelizing the PCA compressed network over 8 PEs achieves a x11.68 speed-up with respect to the original network running on a single PE
He, Bing. „Estimation paramétrique du signal par réseaux de neurones“. Lille 1, 2002. https://pepite-depot.univ-lille.fr/RESTREINT/Th_Num/2002/50376-2002-75.pdf.
Der volle Inhalt der QuellePompougnac, Hugo. „Spécification et compilation de réseaux de neurones embarqués“. Electronic Thesis or Diss., Sorbonne université, 2022. http://www.theses.fr/2022SORUS436.
Der volle Inhalt der QuelleIn this thesis, we propose an approach for the joint specification and compilation of both High-Performance Computing (HPC) and Real-Time Embedded (RTE) aspects of a system. Our approach is based on a formal, algorithmic and tooled integration between two formalisms underlying a large part of works in HPC and RTE fields: the SSA formalism and the synchronous dataflow language Lustre. The SSA formalism is a key component of many HPC compilers, including those used by Machine Learning frameworks such as TensorFlow or PyTorch. The Lustre language is a key component of implementation processes of critical embedded systems in avionics or rail transportation
Aïder, Méziane. „Réseaux d'interconnexion bipartis : colorations généralisées dans les graphes“. Phd thesis, Grenoble 1, 1987. http://tel.archives-ouvertes.fr/tel-00325779.
Der volle Inhalt der QuelleCoupechoux, Emilie. „Analyse de grands graphes aléatoires“. Paris 7, 2012. http://www.theses.fr/2012PA077184.
Der volle Inhalt der QuelleSeveral kinds of real-world networks can be represented by graphs. Since such networks are very large, their detailed topology is generally unknown, and we model them by large random graphs having the same local statistical properties as the observed networks. An example of such properties is the fact that real-world networks are often highly clustered : if two individuals have a friend in common, they are likely to also be each other's friends. Studying random graph models that are both appropriate and tractable from a mathematical point of view is challenging, that is why we consider several clustered random graph models. The spread of epidemics in random graphs can be used to model several kinds of phenomena in real-world networks, as the spread of diseases, or the diffusion of a new technology. The epidemic model we consider depends on the phenomenon we wish to represent :. An individual can contract a disease by a single contact with any of his friends (such contacts being independent),. But a new technology is likely to be adopted by an individual if many of his friends already have the technology in question. We essentially study these two cases. In each case, one wants to know if a small proportion of the population initially infected (or having the technology in question) can propagate the epidemic to a large part of the population
Dhandapani, Yogeshwaran. „Réseaux géométriques aléatoires : connexité et comparaison“. Paris 6, 2010. http://www.theses.fr/2010PA06A621.
Der volle Inhalt der Quelle