Dissertations / Theses on the topic 'Réseaux génératifs'
Create a spot-on reference in APA, MLA, Chicago, Harvard, and other styles
Consult the top 50 dissertations / theses for your research on the topic 'Réseaux génératifs.'
Next to every source in the list of references, there is an 'Add to bibliography' button. Press on it, and we will generate automatically the bibliographic reference to the chosen work in the citation style you need: APA, MLA, Harvard, Chicago, Vancouver, etc.
You can also download the full text of the academic publication as pdf and read online its abstract whenever available in the metadata.
Browse dissertations / theses on a wide variety of disciplines and organise your bibliography correctly.
Côté, Marc-Alexandre. "Réseaux de neurones génératifs avec structure." Thèse, Université de Sherbrooke, 2017. http://hdl.handle.net/11143/10489.
Full textAzeraf, Elie. "Classification avec des modèles probabilistes génératifs et des réseaux de neurones. Applications au traitement des langues naturelles." Electronic Thesis or Diss., Institut polytechnique de Paris, 2022. https://theses.hal.science/tel-03880848.
Full textMany probabilistic models have been neglected for classification tasks with supervised learning for several years, as the Naive Bayes or the Hidden Markov Chain. These models, called generative, are criticized because the induced classifier must learn the observations' law. This problem is too complex when the number of observations' features is too large. It is especially the case with Natural Language Processing tasks, as the recent embedding algorithms convert words in large numerical vectors to achieve better scores.This thesis shows that every generative model can define its induced classifier without using the observations' law. This proposition questions the usual categorization of the probabilistic models and classifiers and allows many new applications. Therefore, Hidden Markov Chain can be efficiently applied to Chunking and Naive Bayes to sentiment analysis.We go further, as this proposition allows to define the classifier induced from a generative model with neural network functions. We "neuralize" the models mentioned above and many of their extensions. Models so obtained allow to achieve relevant scores for many Natural Language Processing tasks while being interpretable, able to require little training data, and easy to serve
Franceschi, Jean-Yves. "Apprentissage de représentations et modèles génératifs profonds dans les systèmes dynamiques." Electronic Thesis or Diss., Sorbonne université, 2022. http://www.theses.fr/2022SORUS014.
Full textThe recent rise of deep learning has been motivated by numerous scientific breakthroughs, particularly regarding representation learning and generative modeling. However, most of these achievements have been obtained on image or text data, whose evolution through time remains challenging for existing methods. Given their importance for autonomous systems to adapt in a constantly evolving environment, these challenges have been actively investigated in a growing body of work. In this thesis, we follow this line of work and study several aspects of temporality and dynamical systems in deep unsupervised representation learning and generative modeling. Firstly, we present a general-purpose deep unsupervised representation learning method for time series tackling scalability and adaptivity issues arising in practical applications. We then further study in a second part representation learning for sequences by focusing on structured and stochastic spatiotemporal data: videos and physical phenomena. We show in this context that performant temporal generative prediction models help to uncover meaningful and disentangled representations, and conversely. We highlight to this end the crucial role of differential equations in the modeling and embedding of these natural sequences within sequential generative models. Finally, we more broadly analyze in a third part a popular class of generative models, generative adversarial networks, under the scope of dynamical systems. We study the evolution of the involved neural networks with respect to their training time by describing it with a differential equation, allowing us to gain a novel understanding of this generative model
Lavault, Antoine. "Generative Adversarial Networks for Synthesis and Control of Drum Sounds." Electronic Thesis or Diss., Sorbonne université, 2023. http://www.theses.fr/2023SORUS614.
Full textAudio synthesizers are electronic systems capable of generating artificial sounds under parameters depending on their architecture. Even though multiple evolutions have transformed synthesizers from simple sonic curiosities in the 1960s and earlier to the main instruments in modern musical productions, two major challenges remain; the development of a system of sound synthesis with a parameter set coherent with its perception by a human and the design of a universal synthesis method, able to model any source and provide new original sounds. This thesis studies using and enhancing Generative Adversarial Networks (GAN) to build a system answering the previously-mentioned problems. The main objective is to propose a neural synthesizer capable of generating realistic drum sounds controllable by predefined timbre parameters and hit velocity. The first step in the project was to propose an approach based on the latest technological advances at the time of its conception to generate realistic drum sounds. We added timbre control capabilities to this method by exploring a different way from existing solutions, i.e., differentiable descriptors. To give experimental guarantees to our work, we performed evaluation experiments via objective metrics based on statistics and subjective and psychopĥysical evaluations on perceived quality and perception of control errors. These experiments continued to add velocity control to the timbral control. Still, with the idea of pursuing the realization of a versatile synthesizer with universal control, we have created a dataset ex-nihilo composed of drum sounds to create an exhaustive database of sounds accessible in the vast majority of conditions encountered in the context of music production. From this dataset, we present experimental results related to the control of dynamics, one of the critical aspects of musical performance but left aside by the literature. To justify the capabilities offered by the GANs synthesis method, we show that it is possible to marry classical synthesis methods with neural synthesis by exploiting the limits and particularities of GANs to obtain new and musically interesting hybrid sounds
Pagliarini, Silvia. "Modeling the neural network responsible for song learning." Thesis, Bordeaux, 2021. http://www.theses.fr/2021BORD0107.
Full textDuring the first period of their life, babies and juvenile birds show comparable phases of vocal development: first, they listen to their parents/tutors in order to build a neural representation of the experienced auditory stimulus, then they start to produce sound and progressively get closer to reproducing their tutor song. This phase of learning is called the sensorimotor phase and is characterized by the presence of babbling, in babies, and subsong, in birds. It ends when the song crystallizes and becomes similar to the one produced by the adults.It is possible to find analogies between brain pathways responsible for sensorimotor learning in humans and birds: a vocal production pathway involves direct projections from auditory areas to motor neurons, and a vocal learning pathway is responsible for imitation and plasticity. The behavioral studies and the neuroanatomical structure of the vocal control circuit in humans and birds provide the basis for bio-inspired models of vocal learning.In particular, birds have brain circuits exclusively dedicated to song learning, making them an ideal model for exploring the representation of vocal learning by imitation of tutors.This thesis aims to build a vocal learning model underlying song learning in birds. An extensive review of the existing literature is discussed in the thesis: many previous studies have attempted to implement imitative learning in computational models and share a common structure. These learning architectures include the learning mechanisms and, eventually, exploration and evaluation strategies. A motor control function enables sound production and sensory response models either how sound is perceived or how it shapes the reward. The inputs and outputs of these functions lie (1)~in the motor space (motor parameters’ space), (2)~in the sensory space (real sounds) and (3)~either in the perceptual space (a low dimensional representation of the sound) or in the internal representation of goals (a non-perceptual representation of the target sound).The first model proposed in this thesis is a theoretical inverse model based on a simplified vocal learning model where the sensory space coincides with the motor space (i.e., there is no sound production). Such a simplification allows us to investigate how to introduce biological assumptions (e.g. non-linearity response) into a vocal learning model and which parameters influence the computational power of the model the most. The influence of the sharpness of auditory selectivity and the motor dimension are discussed.To have a complete model (which is able to perceive and produce sound), we needed a motor control function capable of reproducing sounds similar to real data (e.g. recordings of adult canaries). We analyzed the capability of WaveGAN (a Generative Adversarial Network) to provide a generator model able to produce realistic canary songs. In this generator model, the input space becomes the latent space after training and allows the representation of a high-dimensional dataset in a lower-dimensional manifold. We obtained realistic canary sounds using only three dimensions for the latent space. Among other results, quantitative and qualitative analyses demonstrate the interpolation abilities of the model, which suggests that the generator model we studied can be used as a motor function in a vocal learning model.The second version of the sensorimotor model is a complete vocal learning model with a full action-perception loop (i.e., it includes motor space, sensory space, and perceptual space). The sound production is performed by the GAN generator previously obtained. A recurrent neural network classifying syllables serves as the perceptual sensory response. Similar to the first model, the mapping between the perceptual space and the motor space is learned via an inverse model. Preliminary results show the influence of the learning rate when different sensory response functions are implemented
Yedroudj, Mehdi. "Steganalysis and steganography by deep learning." Thesis, Montpellier, 2019. http://www.theses.fr/2019MONTS095.
Full textImage steganography is the art of secret communication in order to exchange a secret message. In the other hand, image steganalysis attempts to detect the presence of a hidden message by searching artefacts within an image. For about ten years, the classic approach for steganalysis was to use an Ensemble Classifier fed by hand-crafted features. In recent years, studies have shown that well-designed convolutional neural networks (CNNs) can achieve superior performance compared to conventional machine-learning approaches.The subject of this thesis deals with the use of deep learning techniques for image steganography and steganalysis in the spatialdomain.The first contribution is a fast and very effective convolutional neural network for steganalysis, named Yedroudj-Net. Compared tomodern deep learning based steganalysis methods, Yedroudj-Net can achieve state-of-the-art detection results, but also takes less time to converge, allowing the use of a large training set. Moreover,Yedroudj-Net can easily be improved by using well known add-ons. Among these add-ons, we have evaluated the data augmentation, and the the use of an ensemble of CNN; Both increase our CNN performances.The second contribution is the application of deep learning techniques for steganography i.e the embedding. Among the existing techniques, we focus on the 3-player game approach.We propose an embedding algorithm that automatically learns how to hide a message secretly. Our proposed steganography system is based on the use of generative adversarial networks. The training of this steganographic system is conducted using three neural networks that compete against each other: the embedder, the extractor, and the steganalyzer. For the steganalyzer we use Yedroudj-Net, this for its affordable size, and for the fact that its training does not require the use of any tricks that could increase the computational time.This second contribution defines a research direction, by giving first reflection elements while giving promising first results
Grechka, Asya. "Image editing with deep neural networks." Electronic Thesis or Diss., Sorbonne université, 2023. https://accesdistant.sorbonne-universite.fr/login?url=https://theses-intra.sorbonne-universite.fr/2023SORUS683.pdf.
Full textImage editing has a rich history which dates back two centuries. That said, "classic" image editing requires strong artistic skills as well as considerable time, often in the scale of hours, to modify an image. In recent years, considerable progress has been made in generative modeling which has allowed realistic and high-quality image synthesis. However, real image editing is still a challenge which requires a balance between novel generation all while faithfully preserving parts of the original image. In this thesis, we will explore different approaches to edit images, leveraging three families of generative networks: GANs, VAEs and diffusion models. First, we study how to use a GAN to edit a real image. While methods exist to modify generated images, they do not generalize easily to real images. We analyze the reasons for this and propose a solution to better project a real image into the GAN's latent space so as to make it editable. Then, we use variational autoencoders with vector quantification to directly obtain a compact image representation (which we could not obtain with GANs) and optimize the latent vector so as to match a desired text input. We aim to constrain this problem, which on the face could be vulnerable to adversarial attacks. We propose a method to chose the hyperparameters while optimizing simultaneously the image quality and the fidelity to the original image. We present a robust evaluation protocol and show the interest of our method. Finally, we abord the problem of image editing from the view of inpainting. Our goal is to synthesize a part of an image while preserving the rest unmodified. For this, we leverage pre-trained diffusion models and build off on their classic inpainting method while replacing, at each denoising step, the part which we do not wish to modify with the noisy real image. However, this method leads to a disharmonization between the real and generated parts. We propose an approach based on calculating a gradient of a loss which evaluates the harmonization of the two parts. We guide the denoising process with this gradient
Hamis, Sébastien. "Compression de contenus visuels pour transmission mobile sur réseaux de très bas débit." Electronic Thesis or Diss., Institut polytechnique de Paris, 2020. http://www.theses.fr/2020IPPAS020.
Full textThe field of visual content compression (image, video, 2D/3D graphics elements) has known spectacular achievements for more than twenty years, with the emergence numerous international standards such as JPEG, JPEG2000 for still image compression, or MPEG-1/2/4 for video and 3D graphics content coding.The apparition of smartphones and of their related applications have also benefited from these advances, the image being today ubiquitous in a context of mobility. Nevertheless, image transmission requires reliable and available networks, since such visual data that are inherently bandwidth-intensive. While developed countries benefit today from high-performance mobile networks (3G, 4G...), this is not the case in a certain number of regions of the world, particularly in emerging countries, where communications still rely on 2G SMS networks. Transmitting visual content in such a context becomes a highly ambitious challenge, requiring the elaboration of new, for very low bitrate compression algorithm. The challenge is to ensure images transmission over a narrow bandwidth corresponding to a relatively small set (10 to 20) of SMS (140 bytes per SMS).To meet such constraints, multiple axes of development have been considered. After a state-of-the-art of traditional image compression techniques, we have oriented our research towards deep learning methods, aiming achieve post-treatments over strongly compressed data in order to improve the quality of the decoded content.Our contributions are structures around the creation of a new compression scheme, including existing codecs and a panel of post-processing bricks aiming at enhancing highly compressed content. Such bricks represent dedicated deep neural networks, which perform super-resolution and/or compression artifact reduction operations, specifically trained to meet the targeted objectives. These operations are carried out on the decoder side and can be interpreted as image reconstruction algorithms from heavily compressed versions. This approach offers the advantage of being able to rely on existing codecs, which are particularly light and resource-efficient. In our work, we have retained the BPG format, which represents the state of art in the field, but other compression schemes can also be considered.Regarding the type of neural networks, we have adopted Generative Adversarials Nets-GAN, which are particularly well-suited for objectives of reconstruction from incomplete data. Specifically, the two architectures retained and adapted to our objectives are the SRGAN and ESRGAN networks. The impact of the various elements and parameters involved, such as the super-resolution factors and the loss functions, are analyzed in detail.A final contribution concerns experimental evaluation performed. After showing the limitations of objective metrics, which fail to take into account the visual quality of the image, we have put in place a subjective evaluation protocol. The results obtained in terms of MOS (Mean Opinion Score) fully demonstrate the relevance of the proposed reconstruction approaches.Finally, we open our work to different use cases, of a more general nature. This is particularly the case for high-resolution image processing and for video compression
Kalainathan, Diviyan. "Generative Neural Networks to infer Causal Mechanisms : algorithms and applications." Thesis, Université Paris-Saclay (ComUE), 2019. http://www.theses.fr/2019SACLS516.
Full textCausal 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
Villain, Benjamin. "Nouvelle génération de contrôleur d'accès réseau : une approche par réseaux logiciels." Thesis, Paris 6, 2015. http://www.theses.fr/2015PA066663/document.
Full textThis thesis presents the importance of cross-layer network information for network applications in the context of network access control. The dissertation exposes a novel architecture in which a network access controller is mutualized in the Cloud. This architecture allows to address a key market segment for clients unwilling to buy expensive hardware to control their network. Multiple challenges come into play when hosting the controller remotely. Indeed cross-layer information are no longer available which prevents the controller from correctly controlling users activity. A first implementation to share cross-layer information is presented in chapter 2. It leverages specialized session border controllers to send these data in the application protocol, here HTTP. Then chapter 3 presents an innovative solution for the cross-layering problem which allows to intrumentalize network flows with SDN protocols. The solution focuses on a web portal redirection but is extendable to any kind of protocols. The implementation permits to intercept and modify flows in order to input cross-layer data within another network protocol. This solution was implemented in the OpenDaylight OpenFlow controller and shows great results. The mutualized approach coupled with the SDN cross-layer framework allow to build flexible networks with almost no configuration of on-site equipments. The central network controller reduces the overal cost of the solution by being mutualized among multiple clients. Moreover, having the ability to intrumentalize network traffic in software allows to implement any kind of custom behavior on the runtime
Villain, Benjamin. "Nouvelle génération de contrôleur d'accès réseau : une approche par réseaux logiciels." Electronic Thesis or Diss., Paris 6, 2015. http://www.theses.fr/2015PA066663.
Full textThis thesis presents the importance of cross-layer network information for network applications in the context of network access control. The dissertation exposes a novel architecture in which a network access controller is mutualized in the Cloud. This architecture allows to address a key market segment for clients unwilling to buy expensive hardware to control their network. Multiple challenges come into play when hosting the controller remotely. Indeed cross-layer information are no longer available which prevents the controller from correctly controlling users activity. A first implementation to share cross-layer information is presented in chapter 2. It leverages specialized session border controllers to send these data in the application protocol, here HTTP. Then chapter 3 presents an innovative solution for the cross-layering problem which allows to intrumentalize network flows with SDN protocols. The solution focuses on a web portal redirection but is extendable to any kind of protocols. The implementation permits to intercept and modify flows in order to input cross-layer data within another network protocol. This solution was implemented in the OpenDaylight OpenFlow controller and shows great results. The mutualized approach coupled with the SDN cross-layer framework allow to build flexible networks with almost no configuration of on-site equipments. The central network controller reduces the overal cost of the solution by being mutualized among multiple clients. Moreover, having the ability to intrumentalize network traffic in software allows to implement any kind of custom behavior on the runtime
Wang, Yaohui. "Apprendre à générer des vidéos de personnes." Thesis, Université Côte d'Azur, 2021. http://www.theses.fr/2021COAZ4116.
Full textGenerative Adversarial Networks (GANs) have witnessed increasing attention due to their abilities to model complex visual data distributions, which allow them to generate and translate realistic images. While realistic \textit{video generation} is the natural sequel, it is substantially more challenging w.r.t. complexity and computation, associated to the simultaneous modeling of appearance, as well as motion. Specifically, in inferring and modeling the distribution of human videos, generative models face three main challenges: (a) generating uncertain motion and retaining of human appearance, (b) modeling spatio-temporal consistency, as well as (c) understanding of latent representation. In this thesis, we propose three novel approaches towards generating high-visual quality videos and interpreting latent space in video generative models. We firstly introduce a method, which learns to conditionally generate videos based on single input images. Our proposed model allows for controllable video generation by providing various motion categories. Secondly, we present a model, which is able to produce videos from noise vectors by disentangling the latent space into appearance and motion. We demonstrate that both factors can be manipulated in both, conditional and unconditional manners. Thirdly, we introduce an unconditional video generative model that allows for interpretation of the latent space. We place emphasis on the interpretation and manipulation of motion. We show that our proposed method is able to discover semantically meaningful motion representations, which in turn allow for control in generated results. Finally, we describe a novel approach to combine generative modeling with contrastive learning for unsupervised person re-identification. Specifically, we leverage generated data as data augmentation and show that such data can boost re-identification accuracy
Sapountzis, Nikolaos. "Optimisation au niveau réseau dans le cadre des réseaux hétérogènes nouvelle génération." Electronic Thesis or Diss., Paris, ENST, 2016. http://www.theses.fr/2016ENST0082.
Full textBy 2016, it is well-known that mobile networking has dominated our lives. We use our mobile cell phones for almost everything: from social networking to streaming, finding accommodation or banking. Nevertheless, it seems that operators have not understood yet this domination, since their networks consist of nodes that: (i) suffer from enormous load fluctuations, (ii) waste their resources, and (iii) are blamed to be a major energy-killer worldwide. Such shortcomings hurt: load-balancing, spectral and energy efficiency, respectively. The goal of this dissertation is to carefully study these efficiencies and achieve a good trade-off between them for future mobile 5G heterogeneous networks (HetNets). Towards this direction, we firstly focus on (i) the user and traffic differentiation, emerging from the MTC and IoT applications, and (ii) the RAN. Specifically, we perform appropriate modeling, performance analysis and optimization for a family of objectives, using tools mostly coming from (non) convex optimization, probability and queueing theory. Our initial consideration is on network-layer optimizations (e.g. studying the user association problem). Then, we analytically show that cross-layer optimization is key for the success of future HetNets, as one needs to jointly study other problems coming from the layers below (e.g. the TDD allocation problem from the MAC, or the cross-interference management from the PHY) to avoid performance degradation. Finally, we add the backhaul network into our framework, and consider additional constraints related to the backhaul capacity, backhaul topology, as well as the problem of backhaul TDD allocation
Seddik, Mohamed El Amine. "Random Matrix Theory for AI : From Theory to Practice." Electronic Thesis or Diss., université Paris-Saclay, 2020. http://www.theses.fr/2020UPASG010.
Full textAI nowadays relies largely on using large data and enhanced machine learning methods which consist in developing classification and inference algorithms leveraging large datasets of large sizes. These large dimensions induce many counter-intuitive phenomena, leading generally to a misunderstanding of the behavior of many machine learning algorithms often designed with small data dimension intuitions. By taking advantage of (rather than suffering from) the multidimensional setting, random matrix theory (RMT) is able to predict the performance of many non-linear algorithms as complex as some random neural networks as well as many kernel methods such as Support Vector Machines, semi-supervised classification, principal component analysis or spectral clustering. To characterize the performance of these algorithms theoretically, the underlying data model is often a Gaussian mixture model (GMM) which seems to be a strong assumption given the complex structure of real data (e.g., images). Furthermore, the performance of machine learning algorithms depends on the choice of data representation (or features) on which they are applied. Once again, considering data representations as Gaussian vectors seems to be quite a restrictive assumption. Relying on random matrix theory, this thesis aims at going beyond the simple GMM hypothesis, by studying classical machine learning tools under the hypothesis of Lipschitz-ally transformed Gaussian vectors also called concentrated random vectors, and which are more generic than Gaussian vectors. This hypothesis is particularly motivated by the observation that one can use generative models (e.g., GANs) to design complex and realistic data structures such as images, through Lipschitz-ally transformed Gaussian vectors. This notably suggests that making the aforementioned concentration assumption on data is a suitable model for real data and which is just as mathematically accessible as GMM models. Moreover, in terms of data representation, the concentration framework is compatible with one of the most widely used data representations in practice, namely deep neural nets (DNNs) representations, since they consist in a Lipschitz transformation of the input data (e.g., images). Therefore, we demonstrate through this thesis, leveraging on GANs, the interest of considering the framework of concentrated vectors as a model for real data. In particular, we study the behavior of random Gram matrices which appear at the core of various linear models, kernel matrices which appear in kernel methods and also classification methods which rely on an implicit solution (e.g., Softmax layer in neural networks), with concentrated random inputs. Indeed, these methods are at the heart of many classification, regression and clustering machine learning algorithms. In particular, understanding the behavior of these matrices/methods, for concentrated data, allows us to characterize the performances (on real data if we assimilate them to concentrated vectors) of many machine learning algorithms, such as spectral clustering, SVMs, principal component analysis and transfer learning. Analyzing these methods for concentrated data yields to the surprising result that they have asymptotically the same behavior as for GMM data (with the same first and second order statistics). This result strongly suggest the universality aspect of large machine learning classifiers w.r.t. the underlying data distribution
Wang, Qi. "Statistical Models for Human Motion Synthesis." Thesis, Ecole centrale de Marseille, 2018. http://www.theses.fr/2018ECDM0005/document.
Full textThis thesis focuses on the synthesis of motion capture data with statistical models. Motion synthesis is a task of interest for important application fields such as entertainment, human-computer interaction, robotics, etc. It may be used to drive a virtual character that can be involved in the applications of the virtual reality, animation films or computer games. This thesis focuses on the use of statistical models for motion synthesis with a strong focus on neural networks. From the machine learning point of view designing synthesis models consists in learning generative models. Our starting point lies in two main problems one encounters when dealing with motion capture data synthesis, ensuring realism of postures and motion, and handling the large variability in the synthesized motion. The variability in the data comes first from core individual features, we do not all move the same way but accordingly to our personality, our gender, age, and morphology etc. Moreover there are other short term factors of variation like our emotion, the fact that we are interacting with somebody else, that we are tired etc. Data driven models have been studied for generating human motion for many years. Models are learned from labelled datasets where motion capture data are recorded while actors are performed various activities like walking, dancing, running, etc. Traditional statistical models such as Hidden Markov Models, Gaussian Processes have been investigated for motion synthesis, demonstrating strengths but also weaknesses. Our work focuses in this line of research and concerns the design of generative models for sequences able to take into account some contextual information, which will represent the factors of variation. A first part of the thesis present preliminary works that we realised by extending previous approaches relying on Hidden Markov Models and Gaussian Processes to tackle the two main problems related to realism and variability. We first describe an attempt to extend contextual Hidden Markov Models for handling variability in the data by conditioning the parameters of the models to an additional contextual information such as the emotion which which a motion was performed. We then propose a variant of a traditional method for performing a specific motion synthesis task called Inverse Kinematics, where we exploit Gaussian Processes to enforce realism of each of the postures of a generated motion. These preliminary results show some potential of statistical models for designing human motion synthesis systems. Yet none of these technologies offers the flexibility brought by neural networks and the recent deep learning revolution.The second part of the thesis describes the works we realized with neural networks and deep architectures. It builds on recurrent neural networks for dealing with sequences and on adversarial learning which was introduced very recently in the deep learning community for designing accurate generative models for complex data. We propose a simple system as a basis synthesis architecture, which combines adversarial learning with sequence autoencoders, and that allows randomly generating realistic motion capture sequences. Starting from this architecture we design few conditional neural models that allow to design synthesis systems that one can control up to some extent by either providing a high level information that the generated sequence should match (e.g. the emotion) or by providing a sequence in the style of which a sequence should be generated
Ben, Tanfous Amor. "Représentations parcimonieuses dans les variétés de formes pour la classification et la génération de trajectoires humaines." Thesis, Lille 1, 2019. http://www.theses.fr/2019LIL1I111.
Full textDesigning intelligent systems to understand video content has been a hot research topic in the past few decades since it helps compensate the limited human capabilities of analyzing videos in an efficient way. In particular, human behavior understanding in videos is receiving a huge interest due to its many potential applications. At the same time, the detection and tracking of human landmarks in video streams has gained in reliability partly due to the availability of affordable RGB-D sensors. This infer time-varying geometric data which play an important role in the automatic human motion analysis. However, such analysis remains challenging due to enormous view variations, inaccurate detection of landmarks, large intra- and inter- class variations, and insufficiency of annotated data. In this thesis, we propose novel frameworks to classify and generate 2D/3D sequences of human landmarks. We first represent them as trajectories in the shape manifold which allows for a view-invariant analysis. However, this manifold is nonlinear and thereby standard computational tools and machine learning techniques could not be applied in a straightforward manner. As a solution, we exploit notions of Riemannian geometry to encode these trajectories based on sparse coding and dictionary learning. This not only overcomes the problem of nonlinearity of the manifold but also yields sparse representations that lie in vector space, that are more discriminative and less noisy than the original data. We study intrinsic and extrinsic paradigms of sparse coding and dictionary learning in the shape manifold and provide a comprehensive evaluation on their use according to the nature of the data (i.e., face or body in 2D or 3D). Based on these sparse representations, we present two frameworks for 3D human action recognition and 2D micro- and macro- facial expression recognition and show that they achieve competitive performance in comparison to the state-of-the-art. Finally, we design a generative model allowing to synthesize human actions. The main idea is to train a generative adversarial network to generate new sparse representations that are then transformed to pose sequences. This framework is applied to the task of data augmentation allowing to improve the classification performance. In addition, the generated pose sequences are used to guide a second framework to generate human videos by means of pose transfer of each pose to a texture image. We show that the obtained videos are realistic and have better appearance and motion consistency than a recent state-of-the-art baseline
Kips, Robin. "Neural rendering for improved cosmetics virtual try-on." Electronic Thesis or Diss., Institut polytechnique de Paris, 2022. http://www.theses.fr/2022IPPAT011.
Full textAugmented reality applications have rapidly spread across online retail platforms and social media, allowing consumers to virtually try on a large variety of cosmetics products. However, even though appreciated by consumers, such applications currently offer limited realism compared to real product images. On the other hand, the rapidly emerging field of generative models and neural rendering offers new perspectives that we will study in this work for realistic image synthesis and novel virtual try-on experiences. First, we introduce a novel makeup synthesis method based on generative networks in which the makeup color can be explicitly controlled, similar to a physically-based renderer. Our model obtains photorealistic results on lips and eyes makeup in high resolution. Furthermore, we relax the need for labeled data by introducing a weakly-supervised learning approach for generative-based controllable synthesis.However, GANs methods suffer from limitations for real-time applications. Thus, we propose a neural rendering approach for virtual try-on of cosmetics in real-time on mobile devices. Our approach is based on a novel inverse graphics encoder network that learns to map a single example image into the space of parameters of a computer graphics rendering engine. This model is trained using a self-supervised approach which does not require labeled training data. This method enables new applications where consumers can virtually try-on a novel, unknown cosmetic product from an inspirational reference image on social media. Finally, we propose a novel method for accelerating the digitization of new cosmetics products in virtual try-on applications. Inspired by the field of material capture, we introduced a controlled application and imaging system for cosmetics products. Furthermore, we illustrate how this novel type of cosmetics image can be used to estimate the final appearance of cosmetics on the face using a neural rendering approach. Overall, the novel methods introduced in this thesis improve cosmetics virtual try-on technologies both directly, by introducing more realistic rendering method, and indirectly, allowing novel experiences for consumers, and accelerating the creation of virtual try-on for new cosmetics products
Bréhon, Yannick. "Conception et ingénierie de réseaux nouvelle génération orientés Ethernet." Paris, ENST, 2007. http://www.theses.fr/2007ENST0005.
Full textThe increase in individual and professional customer expectations, as well as the fast technological evolutions, are leading to the design of the Next Generation Internet (NGI). Network optimization allows operators to efficiently deliver high-quality services at reduced costs. Due to the pre-eminence of Ethernet technologies at the customer premises and in the access network segment, and due to its low cost, Ethernet is a natural candidate to serve as the technology of choice of the NGI. In the metropolitan network segment, efforts have led to specifying Ethernet flavors which allow mapping client traffic into Virtual LANs. In this thesis, we investigate how to efficiently assign these VLANs to the Spanning Trees generated by Ethernet’s Spanning Tree Protocol, since this is currently the only way to perform traffic-engineering and network optimization in this network segment. In the core network segment, several initiatives aim at turning Ethernet into a GMPLS-controlled connection-oriented technology (such as Layer 2 LSPs and PBB-TE). In this thesis, a new type of connection for packet- and frame- based, connection-oriented and GMPLS-controlled technologies is introduced and studied: the bus-LSP. Both in single-layer and in multi-layer networks, it provides the operator with significant cost reduction. We provide quantification of this reduction, as well as engineering methods for efficiently deploying bus-LSPs. We also detail the control plane protocols extensions needed to manage these bus-LSPs
Cherti, Mehdi. "Deep generative neural networks for novelty generation : a foundational framework, metrics and experiments." Thesis, Université Paris-Saclay (ComUE), 2018. http://www.theses.fr/2018SACLS029/document.
Full textIn recent years, significant advances made in deep neural networks enabled the creation of groundbreaking technologies such as self-driving cars and voice-enabled personal assistants. Almost all successes of deep neural networks are about prediction, whereas the initial breakthroughs came from generative models. Today, although we have very powerful deep generative modeling techniques, these techniques are essentially being used for prediction or for generating known objects (i.e., good quality images of known classes): any generated object that is a priori unknown is considered as a failure mode (Salimans et al., 2016) or as spurious (Bengio et al., 2013b). In other words, when prediction seems to be the only possible objective, novelty is seen as an error that researchers have been trying hard to eliminate. This thesis defends the point of view that, instead of trying to eliminate these novelties, we should study them and the generative potential of deep nets to create useful novelty, especially given the economic and societal importance of creating new objects in contemporary societies. The thesis sets out to study novelty generation in relationship with data-driven knowledge models produced by deep generative neural networks. Our first key contribution is the clarification of the importance of representations and their impact on the kind of novelties that can be generated: a key consequence is that a creative agent might need to rerepresent known objects to access various kinds of novelty. We then demonstrate that traditional objective functions of statistical learning theory, such as maximum likelihood, are not necessarily the best theoretical framework for studying novelty generation. We propose several other alternatives at the conceptual level. A second key result is the confirmation that current models, with traditional objective functions, can indeed generate unknown objects. This also shows that even though objectives like maximum likelihood are designed to eliminate novelty, practical implementations do generate novelty. Through a series of experiments, we study the behavior of these models and the novelty they generate. In particular, we propose a new task setup and metrics for selecting good generative models. Finally, the thesis concludes with a series of experiments clarifying the characteristics of models that can exhibit novelty. Experiments show that sparsity, noise level, and restricting the capacity of the net eliminates novelty and that models that are better at recognizing novelty are also good at generating novelty
Bitton, Adrien. "Meaningful audio synthesis and musical interactions by representation learning of sound sample databases." Electronic Thesis or Diss., Sorbonne université, 2021. http://www.theses.fr/2021SORUS362.
Full textComputer assisted music extensively relies on audio sample libraries and virtual instruments which provide users an ever increasing amount of contents to produce music with. However, principled methods for large-scale interactions are lacking so that browsing samples and presets with respect to a target sound idea is a tedious and arbitrary process. Indeed, library metadata can only describe coarse categories of sounds but do not meaningfully traduce the underlying acoustic contents and continuous variations in timbre which are key elements of music production and creativity. The recent advances in deep generative modelling show unprecedented successes at learning large-scale unsupervised representations which invert to data as diverse as images, texts and audio. These probabilistic models could be refined to specific generative tasks such as unpaired image translation and semantic manipulations of visual features, demonstrating the ability of learning transformations and representations that are perceptually meaningful. In this thesis, we target efficient analysis and synthesis with auto-encoders to learn low dimensional acoustic representations for timbre manipulations and intuitive interactions for music production. In the first place we adapt domain translation techniques to timbre transfer and propose alternatives to adversarial learning for many-to-many transfers. Then we develop models for explicit modelling of timbre variations and controllable audio sampling using conditioning for semantic attribute manipulations and hierarchical learning to represent both acoustic and temporal variations
Boulic-Bouadjio, Audren. "Génération multi-agents de réseaux sociaux." Thesis, Toulouse 1, 2021. http://www.theses.fr/2021TOU10003.
Full textRana, Aakanksha. "Analyse d'images haute gamme dynamique." Electronic Thesis or Diss., Paris, ENST, 2018. http://www.theses.fr/2018ENST0015.
Full textHigh Dynamic Range (HDR) imaging enables to capture a wider dynamic range and color gamut, thus enabling us to draw on subtle, yet discriminating details present both in the extremely dark and bright areas of a scene. Such property is of potential interest for computer vision algorithms where performance degrades substantially when the scenes are captured using traditional low dynamic range (LDR) imagery. While such algorithms have been exhaustively designed using traditional LDR images, little work has been done so far in contex of HDR content. In this thesis, we present the quantitative and qualitative analysis of HDR imagery for such task-specific algorithms. This thesis begins by identifying the most natural and important questions of using HDR content for low-level feature extraction task, which is of fundamental importance for many high-level applications such as stereo vision, localization, matching and retrieval. By conducting a performance evaluation study, we demonstrate how different HDR-based modalities enhance algorithms performance with respect to LDR on a proposed dataset. However, we observe that none of them can optimally to do so across all the scenes. To examine this sub-optimality, we investigate the importance of task-specific objectives for designing optimal modalities through an experimental study. Based on the insights, we attempt to surpass this sub-optimality by designing task-specific HDR tone-mapping operators (TMOs). In this thesis, we propose three learning based methodologies aimed at optimal mapping of HDR content to enhance the efficiency of local features extraction at each stage namely, detection, description and final matching
Laifa, Oumeima. "A joint discriminative-generative approach for tumour angiogenesis assessment in computational pathology." Electronic Thesis or Diss., Sorbonne université, 2019. http://www.theses.fr/2019SORUS230.
Full textAngiogenesis is the process through which new blood vessels are formed from pre-existing ones. During angiogenesis, tumour cells secrete growth factors that activate the proliferation and migration of endothelial cells and stimulate over production of the vascular endothelial growth factor (VEGF). The fundamental role of vascular supply in tumour growth and anti-cancer therapies makes the evaluation of angiogenesis crucial in assessing the effect of anti-angiogenic therapies as a promising anti-cancer therapy. In this study, we establish a quantitative and qualitative panel to evaluate tumour blood vessels structures on non-invasive fluorescence images and histopathological slide across the full tumour to identify architectural features and quantitative measurements that are often associated with prediction of therapeutic response. We develop a Markov Random Field (MFRs) and Watershed framework to segment blood vessel structures and tumour micro-enviroment components to assess quantitatively the effect of the anti-angiogenic drug Pazopanib on the tumour vasculature and the tumour micro-enviroment interaction. The anti-angiogenesis agent Pazopanib was showing a direct effect on tumour network vasculature via the endothelial cells crossing the whole tumour. Our results show a specific relationship between apoptotic neovascularization and nucleus density in murine tumor treated by Pazopanib. Then, qualitative evaluation of tumour blood vessels structures is performed in whole slide images, known to be very heterogeneous. We develop a discriminative-generative neural network model based on both learning driven model convolutional neural network (CNN), and rule-based knowledge model Marked Point Process (MPP) to segment blood vessels in very heterogeneous images using very few annotated data comparing to the state of the art. We detail the intuition and the design behind the discriminative-generative model, and we analyze its similarity with Generative Adversarial Network (GAN). Finally, we evaluate the performance of the proposed model on histopathology slide and synthetic data. The limits of this promising framework as its perspectives are shown
Ayed, Ibrahim. "Neural Models for Learning Real World Dynamics and the Neural Dynamics of Learning." Electronic Thesis or Diss., Sorbonne université, 2022. http://www.theses.fr/2022SORUS434.
Full textThe work presented in this thesis was initially motivated by the discrepancy between the impressive performances of modern neural networks and the lack of applications to scientific problems for which data abounds. Focusing on evolution problems which are classically modelled through ordinary or partial differential equations~(O/PDEs) naturally brought us to consider the more general problem of representing and learning such equations from raw data with neural networks. This was the inception of the first part of our work. The point of view considered in this first part has a natural counterpart: what about the dynamics induced by the trajectories of the NN's weights during training or by the trajectories of data points within them during inference? Can they be usefully modelled? This question was the core of the second part of our work and, while theoretical tools other than O/PDEs happened to be useful in our analysis, our reasoning and intuition were fundamentally driven by considerations stemming from a dynamical viewpoint
Mallik, Mohammed Tariqul Hassan. "Electromagnetic Field Exposure Reconstruction by Artificial Intelligence." Electronic Thesis or Diss., Université de Lille (2022-....), 2023. https://pepite-depot.univ-lille.fr/ToutIDP/EDENGSYS/2023/2023ULILN052.pdf.
Full textThe topic of exposure to electromagnetic fields has received muchattention in light of the current deployment of the fifth generation(5G) cellular network. Despite this, accurately reconstructing theelectromagnetic field across a region remains difficult due to a lack ofsufficient data. In situ measurements are of great interest, but theirviability is limited, making it difficult to fully understand the fielddynamics. Despite the great interest in localized measurements, thereare still untested regions that prevent them from providing a completeexposure map. The research explored reconstruction strategies fromobservations from certain localized sites or sensors distributed inspace, using techniques based on geostatistics and Gaussian processes.In particular, recent initiatives have focused on the use of machinelearning and artificial intelligence for this purpose. To overcome theseproblems, this work proposes new methodologies to reconstruct EMFexposure maps in a specific urban area in France. The main objective isto reconstruct exposure maps to electromagnetic waves from some datafrom sensors distributed in space. We proposed two methodologies basedon machine learning to estimate exposure to electromagnetic waves. Forthe first method, the exposure reconstruction problem is defined as animage-to-image translation task. First, the sensor data is convertedinto an image and the corresponding reference image is generated using aray tracing-based simulator. We proposed an adversarial network cGANconditioned by the environment topology to estimate exposure maps usingthese images. The model is trained on sensor map images while anenvironment is given as conditional input to the cGAN model.Furthermore, electromagnetic field mapping based on the GenerativeAdversarial Network is compared to simple Kriging. The results show thatthe proposed method produces accurate estimates and is a promisingsolution for exposure map reconstruction. However, producing referencedata is a complex task as it involves taking into account the number ofactive base stations of different technologies and operators, whosenetwork configuration is unknown, e.g. powers and beams used by basestations. Additionally, evaluating these maps requires time andexpertise. To answer these questions, we defined the problem as amissing data imputation task. The method we propose takes into accountthe training of an infinite neural network to estimate exposure toelectromagnetic fields. This is a promising solution for exposure mapreconstruction, which does not require large training sets. The proposedmethod is compared with other machine learning approaches based on UNetnetworks and conditional generative adversarial networks withcompetitive results
Letheule, Nathan. "Apports de l'Apprentissage Profond pour la simulation d'images SAR." Electronic Thesis or Diss., université Paris-Saclay, 2024. https://theses.hal.science/tel-04651643.
Full textSimulation is a valuable tool for many SAR imaging applications, however, large simulated images are not yet realistic enough to fool a radar image expert. This thesis proposes to evaluate to what extent the use of recent advances in deep learning can improve the quality of simulations. As a first step, we propose to define a method for measuring the realism of simulated SAR images by comparing them with real images. The resulting metrics will then be used to evaluate simulation results. Secondly, two simulation frameworks based on deep learning are proposed, with different philosophies. The first does not take into account physical knowledge of the imagery, and proposes to learn the transformation of an optical image into a radar image using a cGAN architecture. The second is based on a physical simulator developed at Onera (EMPRISE), and uses automatic input generation from semantic segmentation of an optical image of the scene, via deep learning. For this last promising avenue, we are looking into the description of the input and its impact on the final simulation result. Finally, we will be proposing ways of enriching the images generated by the physical simulator using deep learning, in particular through diffusion networks and text-to-image approaches
Lu, Jingxian. "L'auto-diagnostic dans les réseaux autonomes : application à la supervision de services multimédia sur réseau IP de nouvelle génération." Thesis, Bordeaux 1, 2011. http://www.theses.fr/2011BOR14461/document.
Full textThe autonomic networks show certain interest to manufacturers and operators of telecommunications. The self-diagnosis, the detection of failure and malfunction, is a critical issue in the context of these networks.We choose based-model diagnosis because it allows an automatic diagnosis, and is suitable to distributed network architecture. This diagnosis is based on an explicit modeling of normal and abnormal behavior of the system. We then use a generic diagnostic algorithm that uses this modeling to perform self-diagnosis. The modeling used is based on causal graph. It is an intuitive and efficient representation of causal relationships between observations and failures.The self-diagnosis algorithm we proposed based on the use of causal graphs. The principle is: when an alarm is triggered, the algorithm is run and, with the causal relationships between alarms and causes, the principal causes will be located. Since the causal graph modeling allows a modular and extensible model, it is possible to separate or merge according to the needs of services and communication architectures. This feature allows us to propose a distributed algorithm that adapts to autonomic network architecture. We have thus proposed a self-diagnosis algorithm that allows for the diagnosis corresponding to the autonomic network architecture to realize a global diagnosis.We have implemented this algorithm on a platform OpenIMS, and we showed that our self-diagnostic algorithm could be used for different types of services. The results of implement correspond to what is expected
Oquab, Maxime. "Convolutional neural networks : towards less supervision for visual recognition." Thesis, Paris Sciences et Lettres (ComUE), 2018. http://www.theses.fr/2018PSLEE061.
Full textConvolutional Neural Networks are flexible learning algorithms for computer vision that scale particularly well with the amount of data that is provided for training them. Although these methods had successful applications already in the ’90s, they were not used in visual recognition pipelines because of their lesser performance on realistic natural images. It is only after the amount of data and the computational power both reached a critical point that these algorithms revealed their potential during the ImageNet challenge of 2012, leading to a paradigm shift in visual recogntion. The first contribution of this thesis is a transfer learning setup with a Convolutional Neural Network for image classification. Using a pre-training procedure, we show that image representations learned in a network generalize to other recognition tasks, and their performance scales up with the amount of data used in pre-training. The second contribution of this thesis is a weakly supervised setup for image classification that can predict the location of objects in complex cluttered scenes, based on a dataset indicating only with the presence or absence of objects in training images. The third contribution of this thesis aims at finding possible paths for progress in unsupervised learning with neural networks. We study the recent trend of Generative Adversarial Networks and propose two-sample tests for evaluating models. We investigate possible links with concepts related to causality, and propose a two-sample test method for the task of causal discovery. Finally, building on a recent connection with optimal transport, we investigate what these generative algorithms are learning from unlabeled data
Yan, Sen. "Personalizing facial expressions by exploring emotional mental prototypes." Electronic Thesis or Diss., CentraleSupélec, 2023. http://www.theses.fr/2023CSUP0002.
Full textFacial expressions are an essential form of nonverbal communication. Nowfacial expression manipulation (FEM) techniques have flooded our daily lives. However,in the application context, there are several requirements that need to be addressed. Diversity: facial expression prototypes should be multiple and different between different users. Flexibility: facial expressions should be personalized, i.e., the system can find the facial expression prototype that can meet the need of the users. Exhaustiveness: most FEM technologies can only deal with the six basic emotions, whereas there are more than 4000 emotion labels. Expertise-free: the FEM system should be controllable by anyone withoutthe need for expert knowledge (e.g., psychol-ogists). Efficiency: the system with interactionshould consider user fatigue. In this thesis, to fulfill all the requirements,we proposed an interdisciplinary approach by combining generative adversarial networks with the psychophysical reverse correlation process. Moreover, we created an interactive microbial genetic algorithm to optimize the entire system
Douwes, Constance. "On the Environmental Impact of Deep Generative Models for Audio." Electronic Thesis or Diss., Sorbonne université, 2023. http://www.theses.fr/2023SORUS074.
Full textIn this thesis, we investigate the environmental impact of deep learning models for audio generation and we aim to put computational cost at the core of the evaluation process. In particular, we focus on different types of deep learning models specialized in raw waveform audio synthesis. These models are now a key component of modern audio systems, and their use has increased significantly in recent years. Their flexibility and generalization capabilities make them powerful tools in many contexts, from text-to-speech synthesis to unconditional audio generation. However, these benefits come at the cost of expensive training sessions on large amounts of data, operated on energy-intensive dedicated hardware, which incurs large greenhouse gas emissions. The measures we use as a scientific community to evaluate our work are at the heart of this problem. Currently, deep learning researchers evaluate their works primarily based on improvements in accuracy, log-likelihood, reconstruction, or opinion scores, all of which overshadow the computational cost of generative models. Therefore, we propose using a new methodology based on Pareto optimality to help the community better evaluate their work's significance while bringing energy footprint -- and in fine carbon emissions -- at the same level of interest as the sound quality. In the first part of this thesis, we present a comprehensive report on the use of various evaluation measures of deep generative models for audio synthesis tasks. Even though computational efficiency is increasingly discussed, quality measurements are the most commonly used metrics to evaluate deep generative models, while energy consumption is almost never mentioned. Therefore, we address this issue by estimating the carbon cost of training generative models and comparing it to other noteworthy carbon costs to demonstrate that it is far from insignificant. In the second part of this thesis, we propose a large-scale evaluation of pervasive neural vocoders, which are a class of generative models used for speech generation, conditioned on mel-spectrogram. We introduce a multi-objective analysis based on Pareto optimality of both quality from human-based evaluation and energy consumption. Within this framework, we show that lighter models can perform better than more costly models. By proposing to rely on a novel definition of efficiency, we intend to provide practitioners with a decision basis for choosing the best model based on their requirements. In the last part of the thesis, we propose a method to reduce the inference costs of neural vocoders, based on quantizated neural networks. We show a significant gain on the memory size and give some hints for the future use of these models on embedded hardware. Overall, we provide keys to better understand the impact of deep generative models for audio synthesis as well as a new framework for developing models while accounting for their environmental impact. We hope that this work raises awareness on the need to investigate energy-efficient models simultaneously with high perceived quality
Maachaoui, Mohamed. "Sécurité et performances des réseaux de nouvelle génération." Phd thesis, Toulouse, INPT, 2015. http://oatao.univ-toulouse.fr/14266/1/Maachaoui.pdf.
Full textRifai, Myriana. "Réseaux virtualisés de prochaine génération basés sur SDN." Thesis, Université Côte d'Azur (ComUE), 2017. http://www.theses.fr/2017AZUR4072/document.
Full textSoftware Defined Networking (SDN) was created to provide network programmability and ease complex configuration. Though SDN enhances network performance, it still faces multiple limitations. In this thesis, we build solutions that form a first step towards creating next-generation SDN based networks. In the first part, we present MINNIE to scale the number of rules of SDN switches far beyond the few thousands rules commonly available in TCAM memory, which permits to handle typical data center traffic at very fine grain. To do so MINNIE dynamically compresses the routing rules installed in the TCAM, increasing the number of rules that can be installed. In the second part, we tackle the degraded performance of short flows and present a coarse grained scheduling prototype that leverages SDN switch statistics to decrease their end-to-end delay. Then, we aim at decreasing the 50ms failure protection interval which is not adapted to current broadband speeds and can lead to degraded Quality of Experience. Our solution PRoPHYS leverages the switch statistics in hybrid networks to anticipate link failures by drastically decreasing the number of packets lost. Finally, we tackle the greening problem where often energy efficiency comes at the cost of performance degradation. We present SENAtoR, our solution that leverages SDN nodes in hybrid networks to turn off network devices without hindering the network performance. Finally, we present SEaMLESS that converts idle virtual machines into virtual network functions (VNF) to enable the administrator to further consolidate the data center by turning off more physical servers and reuse resources (e.g. RAM) that are otherwise monopolized
Hennane, Youssef. "Réseaux électriques en présence de génération d'énergies distribuée." Electronic Thesis or Diss., Université de Lorraine, 2022. http://www.theses.fr/2022LORR0183.
Full textMicrogrids play an important role in the electrification of rural and remote areas because they can operate in islanded mode without being dependent on the main grid, thus facilitating access to electricity for these areas.Microgrids operating in island mode require high energy and power storage elements to ensure reliable power supply of loads due to the intermittent nature of renewable energy sources. This problem is more difficult to solve in microgrids with simple topologies in which sources and loads are connected to a single common point of connection via power lines (single-PCC microgrids). A solution to ensure a higher availability of energy, with smaller storage elements and therefore lower cost, is the implementation of microgrids with mesh structure topologies with several distributed sources of different natures connected to its different connection points (multi-PCC microgrids). One of the challenges for mesh microgrids is to synchronize and connect all the distributed generators while ensuring "plug and play" functionality and respecting the active and reactive power sharing between the different distributed generation units. The most widely used methods to achieve DG active and reactive power sharing are applied at the primary level of microgrid control and are designed based on "Droop control" approaches. However, most of these methods are only effective in single-PCC microgrids and not in multi-PCC mesh microgrids. Another problem is that using Droop control-based methods to control microgrids at the primary level can cause the microgrid voltage and frequency to deviate from their nominal values, affecting the power quality and proper operation of the microgrid.The thesis is divided into three parts. The first part presents the concept of microgrids and a review of the literature on their control strategies. In the second part, we propose a new nonlinear droop control strategy for distributed generators of mesh microgrids, whether they operate in islanded or grid-connected modes. This strategy ensures the secure synchronization of the distributed sources and the accurate power sharing between them. This strategy allows compensating voltage and frequency deviations of an islanded microgrid by deploying an additional secondary control for each of its generators using a single information on the voltage of a pilot node. This control method also allows a smooth transition from islanded to grid-connected mode without affecting the active and reactive power sharing of its sources during synchronization, as well as the control of active and reactive power exchanged with the main grid in grid-connected mode. The third part proposes a consensus-based distributed nonlinear Droop control for accurate sharing of active and reactive powers between distributed sources as well as for frequency and voltage restoration in reconfigurable islanded mesh microgrids. The controllers for primary and secondary controls are locally adjusted and do not require knowledge of the microgrid structure. The efficiency of the proposed controls proposed in this thesis as well as their robustness are proved by simulation using Simscape and are validated by HIL tests. Also, the stability of the systems is studied based on the developed mathematical model of two different mesh microgrids controlled by both proposed distributed controls
Rogez, Vincent. "Modélisation simplifiée de sources de production décentralisée pour des etudes de dynamique des réseaux : application à l'intégration d'une production éolienne dans un réseau de distribution insulaire." Artois, 2004. http://www.theses.fr/2004ARTO0206.
Full textFor about ten years, the electric sector has been confronted with more and more European directives. They request, in particular, a high penetration of renewable energies and especially of wind energy. The insertion of these new technologies in the current power system induces new problems such as power quality, flickers, stability. . . According to these new problems, specific studies must be carried out. Consequently, it is essential to use simplified models of dispersed generation systems. In this thesis, a simplified model of aero-derivative gas turbine is proposed. This gas turbine is very successful and is almost always used jointly with a cogeneration application. Generic models of wind generators are also presented. In particular, the complete architecture ordering of a pitch control wind turbine is in particular developed. Finally, a study of the integration of wind turbines in an insular electrical grid is presented. This study has made it possible to define control laws for the participation of wind production in voltage and frequency control
Crestel, Léopold. "Neural networks for automatic musical projective orchestration." Electronic Thesis or Diss., Sorbonne université, 2018. http://www.theses.fr/2018SORUS625.
Full textOrchestration is the art of composing a musical discourse over a combinatorial set of instrumental possibilities. For centuries, musical orchestration has only been addressed in an empirical way, as a scientific theory of orchestration appears elusive. In this work, we attempt to build the first system for automatic projective orchestration, and to rely on machine learning. Hence, we start by formalizing this novel task. We focus our effort on projecting a piano piece onto a full symphonic orchestra, in the style of notable classic composers such as Mozart or Beethoven. Hence, the first objective is to design a system of live orchestration, which takes as input the sequence of chords played by a pianist and generate in real-time its orchestration. Afterwards, we relax the real-time constraints in order to use slower but more powerful models and to generate scores in a non-causal way, which is closer to the writing process of a human composer. By observing a large dataset of orchestral music written by composers and their reduction for piano, we hope to be able to capture through statistical learning methods the mechanisms involved in the orchestration of a piano piece. Deep neural networks seem to be a promising lead for their ability to model complex behaviour from a large dataset and in an unsupervised way. More specifically, in the challenging context of symbolic music which is characterized by a high-dimensional target space and few examples, we investigate autoregressive models. At the price of a slower generation process, auto-regressive models allow to account for more complex dependencies between the different elements of the score, which we believe to be of the foremost importance in the case of orchestration
Brehon, Yannick. "Conception et Ingénierie de Réseaux Nouvelle Génération Orientés Ethernet." Phd thesis, Télécom ParisTech, 2007. http://pastel.archives-ouvertes.fr/pastel-00002564.
Full textKhawam, Kinda. "Ordonnancement opportuniste dans les réseaux mobiles de nouvelle génération." Phd thesis, Télécom ParisTech, 2006. http://pastel.archives-ouvertes.fr/pastel-00002059.
Full textKhawam, Kinda. "L'ordonnancement opportuniste dans les réseaux mobiles de nouvelle génération." Paris, ENST, 2006. http://www.theses.fr/2006ENST0030.
Full textThe scarce resources in wireless systems compounded by their highly variable and error prone propagation characteristics stress the need for efficient resource management. Scheduling is a key tool to allocate efficiently the radio frequency spectrum. While fading effects have long been combated in wireless networks, primarily devoted to voice calls, they are now seen as an opportunity to increase the capacity of novel wireless networks that incorporate data traffic. For data applications, there is a service flexibility afforded by the delay tolerance of elastic traffic and by their ability to adapt their rate to the variable channel quality. Channel-aware scheduling exploit these characteristics by making use of channel state information to ensure that transmission occurs when radio conditions are most favourable. When users have heterogeneous characteristics and quality of service requirements, channel-aware scheduling becomes a challenging task. In this thesis, channel-aware transmission schemes for supporting downlink non-real time services are proposed and analyzed for novel cellular systems. The proposed schemes are designed for providing various QoS requirements for users while increasing the system global throughput
Fissore, Giancarlo. "Generative modeling : statistical physics of Restricted Boltzmann Machines, learning with missing information and scalable training of Linear Flows." Electronic Thesis or Diss., université Paris-Saclay, 2022. http://www.theses.fr/2022UPASG028.
Full textNeural network models able to approximate and sample high-dimensional probability distributions are known as generative models. In recent years this class of models has received tremendous attention due to their potential in automatically learning meaningful representations of the vast amount of data that we produce and consume daily. This thesis presents theoretical and algorithmic results pertaining to generative models and it is divided in two parts. In the first part, we focus our attention on the Restricted Boltzmann Machine (RBM) and its statistical physics formulation. Historically, statistical physics has played a central role in studying the theoretical foundations and providing inspiration for neural network models. The first neural implementation of an associative memory (Hopfield, 1982) is a seminal work in this context. The RBM can be regarded to as a development of the Hopfield model, and it is of particular interest due to its role at the forefront of the deep learning revolution (Hinton et al. 2006).Exploiting its statistical physics formulation, we derive a mean-field theory of the RBM that let us characterize both its functioning as a generative model and the dynamics of its training procedure. This analysis proves useful in deriving a robust mean-field imputation strategy that makes it possible to use the RBM to learn empirical distributions in the challenging case in which the dataset to model is only partially observed and presents high percentages of missing information. In the second part we consider a class of generative models known as Normalizing Flows (NF), whose distinguishing feature is the ability to model complex high-dimensional distributions by employing invertible transformations of a simple tractable distribution. The invertibility of the transformation allows to express the probability density through a change of variables whose optimization by Maximum Likelihood (ML) is rather straightforward but computationally expensive. The common practice is to impose architectural constraints on the class of transformations used for NF, in order to make the ML optimization efficient. Proceeding from geometrical considerations, we propose a stochastic gradient descent optimization algorithm that exploits the matrix structure of fully connected neural networks without imposing any constraints on their structure other then the fixed dimensionality required by invertibility. This algorithm is computationally efficient and can scale to very high dimensional datasets. We demonstrate its effectiveness in training a multylayer nonlinear architecture employing fully connected layers
Shahid, Mustafizur Rahman. "Deep learning for Internet of Things (IoT) network security." Electronic Thesis or Diss., Institut polytechnique de Paris, 2021. http://www.theses.fr/2021IPPAS003.
Full textThe growing Internet of Things (IoT) introduces new security challenges for network activity monitoring. Most IoT devices are vulnerable because of a lack of security awareness from device manufacturers and end users. As a consequence, they have become prime targets for malware developers who want to turn them into bots. Contrary to general-purpose devices, an IoT device is designed to perform very specific tasks. Hence, its networking behavior is very stable and predictable making it well suited for data analysis techniques. Therefore, the first part of this thesis focuses on leveraging recent advances in the field of deep learning to develop network monitoring tools for the IoT. Two types of network monitoring tools are explored: IoT device type recognition systems and IoT network Intrusion Detection Systems (NIDS). For IoT device type recognition, supervised machine learning algorithms are trained to perform network traffic classification and determine what IoT device the traffic belongs to. The IoT NIDS consists of a set of autoencoders, each trained for a different IoT device type. The autoencoders learn the legitimate networking behavior profile and detect any deviation from it. Experiments using network traffic data produced by a smart home show that the proposed models achieve high performance.Despite yielding promising results, training and testing machine learning based network monitoring systems requires tremendous amount of IoT network traffic data. But, very few IoT network traffic datasets are publicly available. Physically operating thousands of real IoT devices can be very costly and can rise privacy concerns. In the second part of this thesis, we propose to leverage Generative Adversarial Networks (GAN) to generate bidirectional flows that look like they were produced by a real IoT device. A bidirectional flow consists of the sequence of the sizes of individual packets along with a duration. Hence, in addition to generating packet-level features which are the sizes of individual packets, our developed generator implicitly learns to comply with flow-level characteristics, such as the total number of packets and bytes in a bidirectional flow or the total duration of the flow. Experimental results using data produced by a smart speaker show that our method allows us to generate high quality and realistic looking synthetic bidirectional flows
Hadam, Pawel. "Transports nouvelle génération dans les réseaux à très haut débit." Phd thesis, Grenoble INPG, 2005. http://tel.archives-ouvertes.fr/tel-00010643.
Full textSalhani, Mohamad. "Modélisation et simulation des réseaux mobiles de 4ème [quatrième] génération." Phd thesis, Toulouse, INPT, 2008. http://oatao.univ-toulouse.fr/7725/1/salhani.pdf.
Full textTomassilli, Andrea. "Vers les réseaux de nouvelle génération avec SDN et NFV." Thesis, Université Côte d'Azur (ComUE), 2019. http://www.theses.fr/2019AZUR4044.
Full textRecent advances in networks, such as Software Defined Networking (SDN) and Network Function Virtualization (NFV), are changing the way network operators deploy and manage Internet services. On one hand, SDN introduces a logically centralized controller with a global view of the network state. On the other hand, NFV enables the complete decoupling of network functions from proprietary appliances and runs them as software applications on general–purpose servers. In such a way, network operators can dynamically deploy Virtual Network Functions (VNFs). SDN and NFV benefit network operators by providing new opportunities for reducing costs, enhancing network flexibility and scalability, and shortening the time-to-market of new applications and services. Moreover, the centralized routing model of SDN jointly with the possibility of instantiating VNFs on–demand, may open the way for an even more efficient operation and resource management of networks. For instance, an SDN/NFV-enabled network may simplify the Service Function Chain (SFC) deployment and provisioning by making the process easier and cheaper. In this study, we aim at investigating how to leverage both SDN and NFV in order to exploit their potential benefits. We took steps to address the new opportunities offered in terms of network design, network resilience, and energy savings, and the new problems that arise in this new context, such as the optimal network function placement in the network. We show that a symbiosis between SDN and NFV can improve network performance and significantly reduce the network's Capital Expenditure (CapEx) and Operational Expenditure (OpEx)
Redieteab, Getachew. "Optimisation cross-layer des futures générations de réseaux WI-FI." Rennes, INSA, 2012. http://www.theses.fr/2012ISAR0021.
Full textDuring this thesis we have studied and proposed cross-layer optimization techniques, with a focus on the IEEE 802. 11ac standard. A new multichannel aggregation scheme has been proposed to improve performance in collision-prone environments. While testing this solution, we have shown that some functionalities directly involved PHY and MAC layers. A cross-layer simulator, compliant with IEEE 802. 11ac specifications, has thus been implemented. We have then used the implemented cross-layer simulator to evaluate the ‘real’ performance of multiple-user multiple-input, multiple-output (MU-MIMO) and compared the obtained results with those of single-user MIMO (SU-MIMO). The impact of the channel sounding interval of MU-MIMO has particularly been studied. Finally, we have proposed ultra short acknowledgment frames for overhead reduction in machine to machine IEEE 802. 11ah communications
Ju, Min. "Optimisation de la protection des réseaux optiques de nouvelle génération." Thesis, Avignon, 2018. http://www.theses.fr/2018AVIG0226/document.
Full textNetwork survivability is a critical issue for optical networks to maintain resilience against network failures. This dissertation addresses several survivability design issues against single link failure and large-scale disaster failure in optical networks. Twoclassic protection schemes, namely pre-configured Cycles (p-Cycle) protection and path protection, are studied to achieve high protection capacity efficiency while taking intoaccount the equipment cost, power consumption and resource usage. These survivable network design problems are first formulated by mathematical models and then offered scalable solutions by heuristic algorithms or a decomposition approach.We first consider single link failure scenario. To cut the multi-line rates transponderscost in survivable Mixed-Line-Rate (MLR) optical networks, a distance-adaptive andlow Capital Expenditures (CAPEX) cost p-cycle protection scheme is proposed withoutcandidate cycle enumeration. Specifically, path-length-limited p-cycles are designed touse appropriate line rate depending on the transponder cost and transmission reach.A Mixed Integer Linear Programming (MILP) model is formulated to directly generate the optimal p-cycles with the minimum CAPEX cost. Additionally, Graph Partitioning in Average (GPA) algorithm and Estimation of cycle numbers (EI) algorithm are developed to make the proposed MILP model scalable, which are shown to be efficient.Regarding the power consumption in survivable Elastic Optical Networks (EONs),power-efficient directed p-cycle protection scheme for asymmetric traffic is proposed.Owing to the advantage of distinguishing traffic amount in two directions, directedp-cycles consume low power by allocating different Frequency Slots (FSs) and modulation formats for each direction. An MILP model is formulated to minimize total power consumption under constraints of directed cycle generation, spectrum assignment,modulation adaptation and protection capacity allocation. To increase the scalability, the MILP model is decomposed into an improved cycle enumeration and a simplified Integer Linear Programming (ILP) model. We have shown that the directedp-cycles out perform the undirected p-cycles in terms of power consumption and spectrum usage.In order to improve the spectrum usage efficiency in p-cycle protection, a SpectrumShared p-cycle (SS-p-cycle) protection is proposed for survivable EONs with and without spectrum conversion. SS-p-cycles permit to reduce spectrum usage and Spectrum Fragmentation Ratio (SFR) by leveraging potential spectrum sharing among multiplep-cycles that have common link(s). The ILP formulations are designed in both cases of with and without spectrum conversion to minimize the spectrum usage of SS-p-cycleswhich can obtain the optimal solution in small instance, and a time-efficient heuristic algorithm is developed to solve large-scale instances. Simulation results show that SSp-cycles have significant advantages on both spectrum allocation and defragmentation efficiency, and the spectrum conversion does help SS-p-cycle design to acquire better spectrum utilization
Daher, Tony. "Gestion cognitive des réseaux radio auto-organisant de cinquième génération." Thesis, Université Paris-Saclay (ComUE), 2018. http://www.theses.fr/2018SACLT023/document.
Full textThe pressure on operators to improve the network management efficiency is constantly growing for many reasons: the user traffic that is increasing very fast, higher end users expectations, emerging services with very specific requirements. Self-Organizing Networks (SON) concept was introduced by the 3rd Generation Partnership Project as a promising solution to simplify the operation and management of complex networks. Many SON modules are already being deployed in today’s networks. Such networks are known as SON enabled networks, and they have proved to be useful in reducing the complexity of network management. However, SON enabled networks are still far from realizing a network that is autonomous and self-managed as a whole. In fact, the behavior of the SON functions depends on the parameters of their algorithm, as well as on the network environment where it is deployed. Besides, SON objectives and actions might be conflicting with each other, leading to incompatible parameter tuning in the network. Each SON function hence still needs to be itself manually configured, depending on the network environment and the objectives of the operator. In this thesis, we propose an approach for an integrated SON management system through a Cognitive Policy Based SON Management (C-PBSM) approach, based on Reinforcement Learning (RL). The C-PBSM translates autonomously high level operator objectives, formulated as target Key Performance Indicators (KPIs), into configurations of the SON functions. Furthermore, through its cognitive capabilities, the C-PBSM is able to build its knowledge by interacting with the real network. It is also capable of adapting with the environment changes. We investigate different RL approaches, we analyze the convergence time and the scalability and propose adapted solutions. We tackle the problem of non-stationarity in the network, notably the traffic variations, as well as the different contexts present in a network. We propose as well an approach for transfer learning and collaborative learning. Practical aspects of deploying RL agents in real networks are also investigated under Software Defined Network (SDN) architecture
Daher, Tony. "Gestion cognitive des réseaux radio auto-organisant de cinquième génération." Electronic Thesis or Diss., Université Paris-Saclay (ComUE), 2018. http://www.theses.fr/2018SACLT023.
Full textThe pressure on operators to improve the network management efficiency is constantly growing for many reasons: the user traffic that is increasing very fast, higher end users expectations, emerging services with very specific requirements. Self-Organizing Networks (SON) concept was introduced by the 3rd Generation Partnership Project as a promising solution to simplify the operation and management of complex networks. Many SON modules are already being deployed in today’s networks. Such networks are known as SON enabled networks, and they have proved to be useful in reducing the complexity of network management. However, SON enabled networks are still far from realizing a network that is autonomous and self-managed as a whole. In fact, the behavior of the SON functions depends on the parameters of their algorithm, as well as on the network environment where it is deployed. Besides, SON objectives and actions might be conflicting with each other, leading to incompatible parameter tuning in the network. Each SON function hence still needs to be itself manually configured, depending on the network environment and the objectives of the operator. In this thesis, we propose an approach for an integrated SON management system through a Cognitive Policy Based SON Management (C-PBSM) approach, based on Reinforcement Learning (RL). The C-PBSM translates autonomously high level operator objectives, formulated as target Key Performance Indicators (KPIs), into configurations of the SON functions. Furthermore, through its cognitive capabilities, the C-PBSM is able to build its knowledge by interacting with the real network. It is also capable of adapting with the environment changes. We investigate different RL approaches, we analyze the convergence time and the scalability and propose adapted solutions. We tackle the problem of non-stationarity in the network, notably the traffic variations, as well as the different contexts present in a network. We propose as well an approach for transfer learning and collaborative learning. Practical aspects of deploying RL agents in real networks are also investigated under Software Defined Network (SDN) architecture
Varet, Antoine. "Conception, mise en oeuvre et évaluation d'un routeur embarqué pour l'avionique de nouvelle génération." Phd thesis, INSA de Toulouse, 2013. http://tel.archives-ouvertes.fr/tel-00932283.
Full textPogodalla, Sylvain. "Réseaux de preuve et génération pour les grammaires de types logiques." Phd thesis, Institut National Polytechnique de Lorraine - INPL, 2001. http://tel.archives-ouvertes.fr/tel-00112982.
Full textlogiques a essentiellement privilégié le sens de l'analyse - syntaxe vers sémantique. Cette thèse souligne le profit que la génération - sémantique vers syntaxe - tire de l'étroitesse de cette relation.
Elle s'appuie sur l'étude logique de ces modèles grammaticaux et met en avant l'utilisation de la logique linéaire et de ses réseaux de preuve. Autour du calcul de Lambek, un fragment intuitionniste de la logique linéaire non commutative, nous étudions le comportement des extensions de ce calcul en tant que modèles syntaxiques, notamment avec le calcul ordonné. Nous montrons par exemple qu'un fragment de ce dernier permet d'engendrer la même classe de langage que les grammaires d'arbres adjoints.
D'autre part, l'adéquation de la syntaxe, portée par la notion de preuve, à la sémantique de Montague, portée par la notion de lambda-terme, s'illustre dans la correspondance de Curry-Howard. L'utilisation des réseaux de preuve nous permet de montrer que, pour le calcul de Lambek et pour des représentations sémantiques linéaires avec une constante au moins, le problème de génération est décidable et que ces grammaires sont intrinsèquement réversibles. Nous caractérisons les formes sémantiques permettant une réalisation syntaxique polynomiale. Aussi pouvons-nous proposer une méthode complète de génération dans ce cadre.
Ces résultats, de même que l'implémentation dont ils ont fait l'objet, exploitent la théorie de la démonstration sous-jacente et en particulier les réseaux de preuve sous forme de graphes. Nous obtenons ainsi un cadre uniforme pour l'analyse et la génération. Pour le conserver, dans l'optique d'une prise en compte sémantique de termes non linéaires grâce aux connecteurs exponentiels de la logique linéaire, nous donnons une nouvelle syntaxe et un nouveau critère de correction pour les réseaux avec exponentiels sous forme de graphes.
Lagha, Naceur. "Proposition d'une architecture de services pour les réseaux de nouvelle génération." Rennes 1, 2002. http://www.theses.fr/2002REN10132.
Full text