Teses / dissertações sobre o tema "Informatique Neuromorphique"
Crie uma referência precisa em APA, MLA, Chicago, Harvard, e outros estilos
Veja os 32 melhores trabalhos (teses / dissertações) para estudos sobre o assunto "Informatique Neuromorphique".
Ao lado de cada fonte na lista de referências, há um botão "Adicionar à bibliografia". Clique e geraremos automaticamente a citação bibliográfica do trabalho escolhido no estilo de citação de que você precisa: APA, MLA, Harvard, Chicago, Vancouver, etc.
Você também pode baixar o texto completo da publicação científica em formato .pdf e ler o resumo do trabalho online se estiver presente nos metadados.
Veja as teses / dissertações das mais diversas áreas científicas e compile uma bibliografia correta.
Suri, Manan. "Technologies émergentes de mémoire résistive pour les systèmes et application neuromorphique". Phd thesis, Université de Grenoble, 2013. http://tel.archives-ouvertes.fr/tel-00935190.
Texto completo da fonteMartinenghi, Romain. "Démonstration opto-électronique du concept de calculateur neuromorphique par Reservoir Computing". Thesis, Besançon, 2013. http://www.theses.fr/2013BESA2052/document.
Texto completo da fonteReservoir Computing (RC) is a currently emerging new brain-inspired computational paradigm, which appeared in theearly 2000s. It is similar to conventional recurrent neural network (RNN) computing concepts, exhibiting essentiallythree parts: (i) an input layer to inject the information in the computing system; (ii) a central computational layercalled the Reservoir; (iii) and an output layer which is extracting the computed result though a so-called Read-Outprocedure, the latter being determined after a learning and training step. The main originality compared to RNNconsists in the last part, which is the only one concerned by the training step, the input layer and the Reservoir beingoriginally randomly determined and fixed. This specificity brings attractive features to RC compared to RNN, in termsof simplification, efficiency, rapidity, and feasibility of the learning, as well as in terms of dedicated hardwareimplementation of the RC scheme. This thesis is indeed concerned by one of the first a hardware implementation of RC,moreover with an optoelectronic architecture.Our approach to physical RC implementation is based on the use of a sepcial class of complex system for the Reservoir,a nonlinear delay dynamics involving multiple delayed feedback paths. The Reservoir appears thus as a spatio-temporalemulation of a purely temporal dynamics, the delay dynamics. Specific design of the input and output layer are shownto be possible, e.g. through time division multiplexing techniques, and amplitude modulation for the realization of aninput mask to address the virtual nodes in the delay dynamics. Two optoelectronic setups are explored, one involving awavelength nonlinear dynamics with a tunable laser, and another one involving an intensity nonlinear dynamics with anintegrated optics Mach-Zehnder modulator. Experimental validation of the computational efficiency is performedthrough two standard benchmark tasks: the NARMA10 test (prediction task), and a spoken digit recognition test(classification task), the latter showing results very close to state of the art performances, even compared with purenumerical simulation approaches
Mesquida, Thomas. "Méthode de calcul et implémentation d’un processeur neuromorphique appliqué à des capteurs évènementiels". Thesis, Université Grenoble Alpes (ComUE), 2018. http://www.theses.fr/2018GREAT117/document.
Texto completo da fonteStudying how our nervous system and sensory mechanisms work lead to the creation of event-driven sensors. These sensors follow the same principles as our eyes or ears for example. This Ph.D. focuses on the search for bio-inspired low power methods enabling processing data from this new kind of sensor. Contrary to legacy sensors, our retina and cochlea only react to the perceived activity in the sensory environment. The artificial “retina” and “cochlea” implementations we call dynamic sensors provide streams of events comparable to neural spikes. The quantity of data transmitted is closely linked to the presented activity, which decreases the redundancy in the output data. Moreover, not being forced to follow a frame-rate, the created events provide increased timing resolution. This bio-inspired support to convey data lead to the development of algorithms enabling visual tracking or speaker recognition or localization at the auditory level, and neuromorphic computing environment implementation. The work we present rely on these new ideas to create new processing solutions. More precisely, the applications and hardware developed rely on temporal coding of the data in the spike stream provided by the sensors
Novembre, Christophe. "Architectures des systèmes de l'information adaptées aux technologies nanométriques et/ou moléculaires : développement d'un composant moléculaire neuromorphique". Lille 1, 2007. https://pepite-depot.univ-lille.fr/LIBRE/Th_Num/2007/50376-2007-Novembre.pdf.
Texto completo da fonteShahsavari, Mahyar. "Unconventional computing using memristive nanodevices : from digital computing to brain-like neuromorphic accelerator". Thesis, Lille 1, 2016. http://www.theses.fr/2016LIL10203/document.
Texto completo da fonteBy 2020, there will be 50 to 100 billion devices connected to the Internet. Two domains of hot research to address these high demands of data processing are the Internet of Things (IoT) and Big Data. The demands of these new applications are increasing faster than the development of new hardware particularly because of the slowdown of Moore's law. The main reason of the ineffectiveness of the processing speed is the memory wall or Von Neumann bottleneck which is coming from speed differences between the processor and the memory. Therefore, a new fast and power-efficient hardware architecture is needed to respond to those huge demands of data processing. In this thesis, we introduce novel high performance architectures for next generation computing using emerging nanotechnologies such as memristors. We have studied unconventional computing methods both in the digital and the analog domains. However, the main focus and contribution is in Spiking Neural Network (SNN) or neuromorphic analog computing. In the first part of this dissertation, we review the memristive devices proposed in the literature and study their applicability in a hardware crossbar digital architecture. At the end of part~I, we review the Neuromorphic and SNN architecture. The second part of the thesis contains the main contribution which is the development of a Neural Network Scalable Spiking Simulator (N2S3) suitable for the hardware implementation of neuromorphic computation, the introduction of a novel synapse box which aims at better learning in SNN platforms, a parameter exploration to improve performance of memristor-based SNN, and finally a study of the application of deep learning in SNN
Lévi, Timothée. "Méthodologie de développement d'une bibliothèque d'IP-AMS en vue de la conception automatisée de systèmes sur puces analogiques et mixtes : application à l'ingénierie neuromorphique". Bordeaux 1, 2007. http://www.theses.fr/2007BOR13480.
Texto completo da fonteAbderrahmane, Nassim. "Impact du codage impulsionnel sur l’efficacité énergétique des architectures neuromorphiques". Thesis, Université Côte d'Azur, 2020. http://www.theses.fr/2020COAZ4082.
Texto completo da fonteNowadays, Artificial Intelligence (AI) is a widespread concept applied to many fields such as transportation, medicine and autonomous vehicles. The main AI algorithms are artificial neural networks, which can be divided into two families: Spiking Neural Networks (SNNs), which are bio-inspired models resulting from neuroscience, and Analog Neural Networks (ANNs), which result from machine learning. The ANNs are experiencing unprecedented success in research and industrial fields, due to their recent successes in many application contexts such as image classification and object recognition. However, they require considerable computational capacity for their deployment which is not adequate to very constrained systems such as 'embedded systems'. To overcome these limitations, many researchers are interested in brain-inspired computing, which would be the perfect alternative to conventional computers based on the Von Neumann architecture (CPU/GPU). This paradigm meets computing performance but not energy efficiency requirements. Hence, it is necessary to design neuromorphic hardware circuits adaptable to parallel and distributed computing. In this context, we have set criteria in terms of accuracy and hardware implementation cost to differentiate the two neural families (SNNs and ANNs). In the case of simple network topologies, we conducted a study that has shown that the spiking models have significant gains in terms of hardware cost when compared to the analog networks, with almost similar prediction accuracies. Therefore, the objective of this thesis is to design a generic neuromorphic architecture that is based on spiking neural networks. To this end, we have set up a three-level design flow for exploring and implementing neuromorphic architectures.In an energy efficiency context, a thorough exploration of different neural coding paradigms for neural data representation in SNNs has been carried out. Moreover, new derivative versions of rate-based coding have been proposed that aim to get closer to the activity produced by temporal coding, which is characterized by a reduced number of spikes propagating in the network. In this way, the number of spikes can be reduced so that the number of events to be processed in the SNNs gets smaller. The aim in doing this approach is to reduce the hardware architecture's energy consumption. The proposed coding approaches are: First Spike, which is characterized using at most one single spike to present an input data, and Spike Select, which allows to regulate and minimize the overall spiking activity in the SNN.In the RTL design exploration, we quantitatively compared three SNN architectural models having different levels of computing parallelism and multiplexing. Using Spike Select coding results in a distribution regulation of the spiking data, with most of them generated within the first layer and few of them propagate into the deep layers. Such distribution benefits from a so-called 'hybrid architecture' that includes a fully-parallel part for the first layer and multiplexed parts to the other layers. Therefore, combining the Spike Select and the Hybrid Architecture would be an effective solution for embedded AI applications, with an efficient hardware and latency trade-off.Finally, based on the architectural and neural choices resulting from the previous exploration, we have designed a final event-based architecture dedicated to SNNs supporting different neural network types and sizes. The architecture supports the most used layers: convolutional, pooling and fully-connected. Using this architecture, we will be able to compare analog and spiking neural networks on realistic applications and to finally conclude about the use of SNNs for Embedded Artificial Intelligence
Lagorce, Xavier. "Computational methods for event-based signals and applications". Thesis, Paris 6, 2015. http://www.theses.fr/2015PA066434/document.
Texto completo da fonteComputational Neurosciences are a great source of inspiration for data processing and computation. Nowadays, how great the state of the art of computer vision might be, it is still way less performant that what our brains or the ones from other animals or insects are capable of. This thesis takes on this observation to develop new computational methods for computer vision and generic computation relying on data produced by event-based sensors such as the so called “silicon retinas”. These sensors mimic biology and are used in this work because of the sparseness of their data and their precise timing: information is coded into events which are generated with a microsecond precision. This opens doors to a whole new paradigm for machine vision, relying on time instead of using images. We use these sensors to develop applications such as object tracking or recognition and feature extraction. We also used computational neuromorphic platforms to better implement these algorithms which led us to rethink the idea of computation itself. This work proposes new ways of thinking computer vision via event-based sensors and a new paradigm for computation. Time is replacing memory to allow for completely local operations, enabling highly parallel machines in a non-Von Neumann architecture
La, Barbera Selina. "Development of filamentary Memristive devices for synaptic plasticity implementation". Thesis, Lille 1, 2015. http://www.theses.fr/2015LIL10163/document.
Texto completo da fonteReplicating the computational functionalities of the brain remains one of the biggest challenges for the future of information and communication technologies. In this context, neuromorphic engineering appears a very promising direction. In this context memristive devices have been recently proposed for the implementation of synaptic functions, offering the required features and integration potentiality in a single component. In this dissertation, we present how advanced synaptic features can be implemented in memristive nanodevices. By exploiting the physical properties of filamentary switching, we successfully implemented a non-Hebbian plasticity form corresponding to the synaptic adaptation. We demonstrate that complex filament shape, such as dendritic paths of variable density and width, can reproduce short- and long- term processes observed in biological synapses and can be conveniently controlled by achieving a flexible way to program the device memory state and the relative state volatility. Then, we show that filamentary switching can be additionally controlled to reproduce a Hebbian plasticity form that corresponds to an increase of the synaptic weight when time correlation between pre- and post-neuron firing is experienced at the synaptic connection. We interpreted our results in the framework of a phenomenological model developed for biological synapses. Finally, we exploit this model to investigate how spike-based systems can be realized for memory and computing applications. These results pave the way for future engineering of neuromorphic computing systems, where complex behaviors of memristive physics can be exploited
Ly, Denys. "Mémoires résistives et technologies 3D monolithiques pour processeurs neuromorphiques impulsionnels et reconfigurables". Thesis, Université Grenoble Alpes, 2020. http://www.theses.fr/2020GRALT016.
Texto completo da fonteThe human brain is a complex, energy-efficient computational system that excels at cognitive tasks thanks to its natural capability to perform inference. By contrast, conventional computing systems based on the classic Von Neumann architecture require large power budget to execute such assignments. Herein comes the idea to build brain-inspired electronic computing systems, the so-called neuromorphic approach. In this thesis, we explore the use of novel technologies, namely Resistive Memories (RRAMs) and three-dimensional (3D) monolithic technologies, to enable the hardware implementation of compact, low-power reconfigurable Spiking Neural Network (SNN) processors. We first provide a comprehensive study of the impact of RRAM electrical properties on SNNs with RRAM synapses and trained with unsupervised learning (Spike-Timing-Dependent Plasticity (STDP)). In particular, we clarify the role of synaptic variability originating from RRAM resistance variability. Second, we investigate the use of RRAM-based Ternary Content-Addressable Memory (TCAM) arrays as synaptic routing tables in SNN processors to enable on-the-fly reconfigurability of network topology. For this purpose, we present in-depth electrical characterisations of two RRAM-based TCAM circuits: (i) the most common two-transistors/two-RRAMs (2T2R) RRAM-based TCAM, and (ii) a novel one-transistor/two-RRAMs/one-transistor (1T2R1T) RRAM-based TCAM, both featuring the smallest silicon area up-to-date. We compare both structures in terms of performance, reliability, and endurance. Finally, we explore the potential of 3D monolithic technologies to improve area-efficiency. In addition to the conventional monolithic integration of RRAMs in the back-end-of-line of CMOS technology, we examine the vertical stacking of CMOS over CMOS transistors. To this end, we demonstrate the full 3D monolithic integration of two tiers of CMOS transistors with one tier of RRAM devices, and present electrical characterisations performed on the fabricated devices
Bichler, Olivier. "Contribution à la conception d'architecture de calcul auto-adaptative intégrant des nanocomposants neuromorphiques et applications potentielles". Phd thesis, Université Paris Sud - Paris XI, 2012. http://tel.archives-ouvertes.fr/tel-00781811.
Texto completo da fonteHaessig, Germain. "Neuromorphic computation using event-based sensors : from algorithms to hardware implementations". Thesis, Sorbonne université, 2018. http://www.theses.fr/2018SORUS422/document.
Texto completo da fonteThis thesis is about the implementation of neuromorphic algorithms, using, as a first step, data from a silicon retina, mimicking the human eye’s behavior, and then evolve towards all kind of event-based signals. These eventbased signals are coming from a paradigm shift in the data representation, thus allowing a high dynamic range, a precise temporal resolution and a sensor-level data compression. Especially, we will study the development of a high frequency monocular depth map generator, a real-time spike sorting algorithm for intelligent brain-machine interfaces, and an unsupervised learning algorithm for pattern recognition. Some of these algorithms (Optical flow detection, depth map construction from stereovision) will be in the meantime developed on available neuromorphic platforms (SpiNNaker, TrueNorth), thus allowing a fully neuromorphic pipeline, from sensing to computing, with a low power budget
Bennett, Christopher H. "Apprentissage local avec des dispositifs de mémoire hautement analogiques". Thesis, Université Paris-Saclay (ComUE), 2018. http://www.theses.fr/2018SACLS037/document.
Texto completo da fonteIn the next era of distributed computing, brain-based computers that perform operations locally rather than in remote servers would be a major benefit in reducing global energy costs. A new generation of emerging nonvolatile memory devices is a leading candidate for achieving this neuromorphic vision. Using theoretical and experimental work, we have explored critical issues that arise when physically realizing modern artificial neural network (ANN) architectures using emerging memory devices (“memristors”). In our experimental work, we showed organic nanosynapses adapting automatically as logic gates via a companion digital neuron and programmable logic cell (FGPA). In our theoretical work, we also considered multilayer memristive ANNs. We have developed and simulated random projection (NoProp) and backpropagation (Multilayer Perceptron) variants that use two crossbars. These local learning systems showed critical dependencies on the physical constraints of nanodevices. Finally, we examined how feed-forward ANN designs can be modified to exploit temporal effects. We focused in particular on improving bio-inspiration and performance of the NoProp system, for example, we improved the performance with plasticity effects in the first layer. These effects were obtained using a silver ionic nanodevice with intrinsic plasticity transition behavior
Bedecarrats, Thomas. "Etude et intégration d’un circuit analogique, basse consommation et à faible surface d'empreinte, de neurone impulsionnel basé sur l’utilisation du BIMOS en technologie 28 nm FD-SOI". Thesis, Université Grenoble Alpes (ComUE), 2019. http://www.theses.fr/2019GREAT045.
Texto completo da fonteWhile Moore’s law reaches its limits, microelectronics actors are looking for new paradigms to ensure future developments of our information society. Inspired by biologic nervous systems, neuromorphic engineering is providing new perspectives which have already enabled breakthroughs in artificial intelligence. To achieve sufficient performances to allow their spread, neural processors have to integrate neuron circuits as small and as low power(ed) as possible so that artificial neural networks they implement reach a critical size. In this work, we show that it is possible to reduce the number of components necessary to design an analogue spiking neuron circuit thanks to the functionalisation of parasitic generation currents in a BIMOS transistor integrated in 28 nm FD-SOI technology and sized with the minimum dimensions allowed by this technology. After a systematic characterization of the FD-SOI BIMOS currents under several biases through quasi-static measurements at room temperature, a compact model of this component, adapted from the CEA-LETI UTSOI one, is proposed. The BIMOS-based leaky, integrate-and-fire spiking neuron (BB-LIF SN) circuit is described. Influence of the different design and bias parameters on its behaviour observed during measurements performed on a demonstrator fabricated in silicon is explained in detail. A simple analytic model of its operating boundaries is proposed. The coherence between measurement and compact simulation results and predictions coming from the simple analytic model attests to the relevance of the proposed analysis. In its most successful achievement, the BB-LIF SN circuit is 15 µm², consumes around 2 pJ/spike, triggers at a rate between 3 and 75 kHz for 600 pA to 25 nA synaptic currents under a 3 V power supply
Spyrou, Theofilos. "Functional safety and reliability of neuromorphic computing systems". Electronic Thesis or Diss., Sorbonne université, 2023. http://www.theses.fr/2023SORUS118.
Texto completo da fonteThe recent rise of Artificial Intelligence (AI) has found a wide range of applications essentially integrating it gaining more and more ground in almost every field of our lives. With this steep integration of AI, it is reasonable for concerns to arise, which need to be eliminated before the employment of AI in the field, especially in mission- and safety-critical applications like autonomous vehicles. Spiking Neural Networks (SNNs), although biologically inspired, inherit only partially the remarkable fault resilience capabilities of their biological counterparts, being vulnerable to electronic defects and faults occurring at hardware level. Hence, a methodological exploration of the dependability characteristics of AI hardware accelerators and neuromorphic platforms is of utmost importance. This thesis tackles the subjects of testing and fault tolerance in SNNs and their neuromorphic implementations on hardware
Reverter, Valeiras David. "Event-based detection and tracking". Thesis, Paris 6, 2017. http://www.theses.fr/2017PA066566/document.
Texto completo da fonteThe main goal of this thesis is the development of event-based algorithms for visual detection and tracking. This algorithms are specifically designed to work on the output of neuromorphic event-based cameras. This type of cameras are a new type of bioinspired sensors, whose principle of operation is based on the functioning of the retina: every pixel is independent and generates events asynchronously when a sufficient amount of change is detected in the luminance at the corresponding position on the focal plane. This new way of encoding visual information calls for new processing methods. First, a part-based shape tracking is presented, which represents an object as a set of simple shapes linked by springs. The resulting virtual mechanical system is simulated with every incoming event. Next, a line and segment detection algorithm is introduced, which can be employed as an event-based low level feature. Two event-based methods for 3D pose estimation are then presented. The first of these 3D algorithms is based on the assumption that the current estimation is close to the true pose of the object, and it consequently requires a manual initialization step. The second of the 3D methods is designed to overcome this limitation. All the presented methods update the estimated position (2D or 3D) of the tracked object with every incoming event. This results in a series of trackers capable of estimating the position of the tracked object with microsecond resolution. This thesis shows that event-based vision allows to reformulate a broad set of computer vision problems, often resulting in simpler but accurate algorithms
Vodenicarevic, Damir. "Rhythms and oscillations : a vision for nanoelectronics". Thesis, Université Paris-Saclay (ComUE), 2017. http://www.theses.fr/2017SACLS518/document.
Texto completo da fonteWith the advent of "artificial intelligence", computers, mobile devices and other connected objects are being pushed beyond the realm of arithmetic and logic operations, for which they have been optimized over decades, in order to process "cognitive" tasks such as automatic translation and image or voice recognition, for which they are not the ideal substrate. As a result, supercomputers may require megawatts to process tasks for which the human brain only needs 20 watt. This has revived interest into the design of alternative computing schemes inspired by the brain. In particular, neural oscillations that appear to be linked to computational activity in the brain have inspired approaches leveraging the complex physics of networks of coupled oscillators in order to process cognitive tasks efficiently. In the light of recent advances in nano-technology allowing the fabrication of highly integrable nano-oscillators, this thesis proposes and studies novel neuro-inspired oscillator-based pattern classification architectures that could be implemented on chip
Daouzli, Adel Mohamed Renaud Sylvie Saïghi Sylvain. "Systèmes neuromorphiques étude et implantation de fonctions d'apprentissage et de plasticité /". S. l. : Bordeaux 1, 2009. http://ori-oai.u-bordeaux1.fr/pdf/2009/DAOUZLI_ADEL_MOHAMED_2009.pdf.
Texto completo da fonteMarcireau, Alexandre. "Vision par ordinateur évènementielle couleur : cadriciel, prototype et applications". Thesis, Sorbonne université, 2019. http://www.theses.fr/2019SORUS248.
Texto completo da fonteNeuromorphic engineering is a bio-inspired approach to sensors and computers design. It aims to mimic biological systems down to the transistor level, to match their unparalleled robustness and power efficiency. In this context, event-based vision sensors have been developed. Unlike conventional cameras, they feature independent pixels which asynchronously generate an output upon detecting changes in their field of view, with high temporal precision. These properties are not leveraged by conventional computer vision algorithms, thus a new paradigm has been devised. It advocates short calculations performed on each event to mimic the brain, and shows promise both for computer vision and as a model of biological vision. This thesis explores event-based computer vision to improve our understanding of visual perception and identify potential applications. We approach the issue through color, a mostly unexplored aspect of event-based sensors. We introduce a framework supporting color events, as well as two experimental devices leveraging it: a three-chip event-based camera performing absolute color measurements, and a visual psychophysics setup to study the role of precise-timing in the brain. We explore the possibility to apply the color sensor to the genetic engineering Brainbow method, and present a new mathematical model for the latter
Chappet, de Vangel Benoît. "Modèles cellulaires de champs neuronaux dynamiques". Thesis, Université de Lorraine, 2016. http://www.theses.fr/2016LORR0194/document.
Texto completo da fonteIn the constant search for design going beyond the limits of the von Neumann architecture, non conventional computing offers various solutions like neuromorphic engineering and cellular computing. Like von Neumann who roughly reproduced brain structures to design computers architecture, neuromorphic engineering takes its inspiration directly from neurons and synapses using analog substratum. Cellular computing influence comes from natural substratum (chemistry, physic or biology) imposing locality of interactions from which organisation and computation emerge. Research on neural mechanisms was able to demonstrate several emergent properties of the neurons and synapses. One of them is the attractor dynamics described in different frameworks by Amari with the dynamic neural fields (DNF) and Amit and Zhang with the continuous attractor neural networks. These neural fields have various computing properties and are particularly relevant for spatial representations and early stages of visual cortex processing. They were used, for instance, in autonomous robotics, classification and clusterization. Similarly to many neuronal computing models, they are robust to noise and faults and thus are good candidates for noisy hardware computation models which would enable to keep up or surpass the Moore law. Indeed, transistor area reductions is leading to more and more noise and the relaxation of the approx. 0% fault during production and operation of integrated circuits would lead to tremendous savings. Furthermore, progress towards many-cores circuits with more and more cores leads to difficulties due to the centralised computation mode of usual parallel algorithms and their communication bottleneck. Cellular computing is the natural answer to these problems. Based on these different arguments, the goal of this thesis is to enable rich computations and applications of dynamic neural fields on hardware substratum with neuro-cellular models enabling a true locality, decentralization and scalability of the computations. This work is an attempt to go beyond von Neumann architectures by using cellular and neuronal computing principles. However, we will stay in the digital framework by exploring performances of proposed architectures on FPGA. Analog hardware like VLSI would also be very interesting but is not studied here. The main contributions of this work are : 1) Neuromorphic DNF computation ; 2) Local DNF computations with randomly spiking dynamic neural fields (RSDNF model) ; 3) Local and asynchronous DNF computations with cellular arrays of stochastic asynchronous spiking DNFs (CASAS-DNF model)
Roclin, David. "Utilisation des nano-composants électroniques dans les architectures de traitement associées aux imageurs". Thesis, Paris 11, 2014. http://www.theses.fr/2014PA112408/document.
Texto completo da fonteBy using learning mechanisms extracted from recent discoveries in neuroscience, spiking neural networks have demonstrated their ability to efficiently analyze the large amount of data from our environment. The implementation of such circuits on conventional processors does not allow the efficient exploitation of their parallelism. The use of digital memory to implement the synaptic weight does not allow the parallel reading or the parallel programming of the synapses and it is limited by the bandwidth of the connection between the memory and the processing unit. Emergent memristive memory technologies could allow implementing this parallelism directly in the heart of the memory.In this thesis, we consider the development of an embedded spiking neural network based on emerging memory devices. First, we analyze a spiking network to optimize its different components: the neuron, the synapse and the STDP learning mechanism for digital implementation. Then, we consider implementing the synaptic memory with emergent memristive devices. Finally, we present the development of a neuromorphic chip co-integrating CMOS neurons with CBRAM synapses
Daouzli, Adel Mohamed. "Systèmes neuromorphiques : étude et implantation de fonctions d'apprentissage et de plasticité". Thesis, Bordeaux 1, 2009. http://www.theses.fr/2009BOR13806/document.
Texto completo da fonteIn this work, we have investigated the effect of input noise patterns on synaptic plasticity applied to a neural network. The study was realised using a neuromorphic hardware simulation system. We have implemented a neural conductance model based on Hodgkin and Huxley formalism, and a biophysical model for plasticity. The tasks performed during this thesis project included the configuration of the system, the development of software tools, the analysis tools to explore experimental results, and the development of the software modules for the remote access to the system via Internet using PyNN scripts (PyNN is a neural network description language commonly used in computational neurosciences)
Torralba, Barriuso Antonio. "Architectures analogiques pour la vision : réseaux cellulaires et circuits neuromorphiques". Grenoble INPG, 1999. http://www.theses.fr/1999INPG0189.
Texto completo da fonteVision machines based on actual computational methods require the development of simple low-level feature detectors. The low-level feature detectors measure local image properties as scale, orientation, and velocity. Analog VLSI devices that mimic some functionality of biological systems appear to be robust, low power consuming, and fast enough to solve vision problems in real time. In this thesis, it is shown that active resistive diffusion networks with low connectivity offer a common framework for the implementation of the low-level feature detectors commonly used in vision (band-pass, wedge, endstopped, velocity-tuned, etc. ) yielding to a simple and homogeneous architecture. Diffusion networks with four neighbor interactions implement velocity-tuned spatiotemporal filters and oriented spatial filters. Velocity-tuned filters yield to efficient and reliable motion estimation using an analog architecture based on active resistive networks from the photoreceptor level to velocity estimation. Oriented spatial filters using resistive diffusion networks yield to a filter basis able to generate complex filters commonly used in vision. From this basis of filters, we generate more complex filters (e. G. Oriented quadrature band-pass, quadrature wedge filters) that are approximated by a linear combination of that basis. Changing the linear combination of the basis filters allows the tuning of the architecture to different features. The proposed architecture offers a way to implement both spatial and spatiotemporal filters (motion sensors) with a low cost. This approach opens an issue to the problem of implementing large sets of spatial and spatiotemporal filters tuned to different features (edges, junctions, velocity, etc. ) in a single chip
Marquez, Alfonzo Bicky. "Reservoir computing photonique et méthodes non-linéaires de représentation de signaux complexes : Application à la prédiction de séries temporelles". Thesis, Bourgogne Franche-Comté, 2018. http://www.theses.fr/2018UBFCD042/document.
Texto completo da fonteArtificial neural networks are systems prominently used in computation and investigations of biological neural systems. They provide state-of-the-art performance in challenging problems like the prediction of chaotic signals. Yet, the understanding of how neural networks actually solve problems like prediction remains vague; the black-box analogy is often employed. Merging nonlinear dynamical systems theory with machine learning, we develop a new concept which describes neural networks and prediction within the same framework. Taking profit of the obtained insight, we a-priori design a hybrid computer, which extends a neural network by an external memory. Furthermore, we identify mechanisms based on spatio-temporal synchronization with which random recurrent neural networks operated beyond their fixed point could reduce the negative impact of regular spontaneous dynamics on their computational performance. Finally, we build a recurrent delay network in an electro-optical setup inspired by the Ikeda system, which at first is investigated in a nonlinear dynamics framework. We then implement a neuromorphic processor dedicated to a prediction task
Williame, Jérôme. "Oscillateurs nanomagnétiques soumis à une boucle de rétroaction à retard : Bruit, chaos et applications neuromorphiques". Thesis, Université Paris-Saclay (ComUE), 2019. http://www.theses.fr/2019SACLS119.
Texto completo da fonteA delay feedback loop occurs when the output of a system is used to modify the input signal of the system. This phenomenon appears in fields as varied as the physics of amplifiers, the biology of insulin regulation or in social interactions. The effects of a delay feedback loop on an electronic system are well known and have given rise to many applications: phase-locked loops to improve stochastic properties, amplification or regulation loops, and so on. However, these feedback effects remain relatively unexplored in the context of nanomagnetic systems. In this thesis I have studied theoretically the consequences of delayed feedback on the magnetization dynamics of three different nanoscale systems with a separate focus for each system. The first involves spin-torque nano-oscillators whose stochastic properties and the impact of a feedback loop on them have been studied. It is found that significant changes can occur to the spectral linewidth, along with the appearance of secondary frequencies at large delays. The second system involves the macrospin oscillator, where I investigated how delayed feedback can induce chaotic transitions between the in-plane and out-ofplane precession states. These complex dynamics can be used to generate random numbers. The third system represents a proposal for implementing a Mackey-Glass oscillator using a domain wall racetrack-like geometry. By deforming this domain wall with spin polarized currents and with a suitable readout function, I show that this oscillator can be used for a time-delay architecture for reservoir computing. Tests of nonlinear time series prediction are conducted to evaluate the performance of this system
Ambroise, Matthieu. "Hybridation des réseaux de neurones : de la conception du réseau à l’interopérabilité des systèmes neuromorphiques". Thesis, Bordeaux, 2015. http://www.theses.fr/2015BORD0394/document.
Texto completo da fonteHYBRID experiments allow to connect a biological neural network with an artificial one,used in neuroscience research and therapeutic purposes. During these three yearsof PhD, this thesis focused on hybridization in a close-up view (bi-diretionnal direct communication between the artificial and the living) and in a broader view (interoperability ofneuromorphic systems). In the early 2000s, an analog neuromorphic system has been connected to a biological neural network. This work is firstly, about the design of a digital neural network, for hybrid experimentsin two multi-disciplinary projects underway in AS2N team of IMS presented in this document : HYRENE (ANR 2010-Blan-031601), aiming the development of a hybrid system for therestoration of motor activity in the case of a spinal cord lesion,BRAINBOW (European project FP7-ICT-2011-C), aiming the development of innovativeneuro-prostheses that can restore communication around cortical lesions. Having a configurable architecture, a digital neural network was designed for these twoprojects. For the first project, the artificial neural network emulates the activity of CPGs (Central Pattern Generator), causing the locomotion in the animal kingdom. This activity will trigger aseries of stimuli in the injured spinal cord textit in vitro and recreating locomotion previously lost. In the second project, the neural network topology will be determined by the analysis anddecryption of biological signals from groups of neurons grown on electrodes, as well as modeling and simulations performed by our partners. The neural network will be able to repair the injuredneural network. This work show the two different networks design approach and preliminary results obtained in the two projects.Secondly, this work hybridization to extend the interoperability of neuromorphic systems. Through a communication protocol using Ethernet, it is possible to interconnect electronic neuralnetworks, computer and biological. In the near future, it will increase the complexity and size of networks
Belhadj-Mohamed, Bilel. "Systèmes neuromorphiques temps réel : contribution à l’intégration de réseaux de neurones biologiquement réalistes avec fonctions de plasticité". Thesis, Bordeaux 1, 2010. http://www.theses.fr/2010BOR14051/document.
Texto completo da fonteThis work has been supported by the European FACETS project. Within this project, we contribute in developing hardware mixed-signal devices for real-time spiking neural network simulation. These devices may potentially contribute to an improved understanding of learning phenomena in the neo-cortex. Neuron behaviours are reproduced using analog integrated circuits which implement Hodgkin-Huxley based models. In this work, we propose a digital architecture aiming to connect many neuron circuits together, forming a network. The inter-neuron connections are reconfigurable and can be ruled by a plasticity model. The architecture is mapped onto a commercial programmable circuit (FPGA). Many methods are developed to optimize the utilisation of hardware resources as well as to meet real-time constraints. In particular, a token-passing communication protocol has been designed and developed to guarantee real-time aspects of the dialogue between several FPGAs in a multiboard system allowing the integration of a large number of neurons. The global system is able to run neural simulations in biological real-time with high degree of realism, and then can be used by neurobiologists and computer scientists to carry on neural experiments
Ezzadeen, Mona. "Conception d'un circuit dédié au calcul dans la mémoire à base de technologie 3D innovante". Electronic Thesis or Diss., Aix-Marseille, 2022. http://theses.univ-amu.fr.lama.univ-amu.fr/221212_EZZADEEN_955e754k888gvxorp699jljcho_TH.pdf.
Texto completo da fonteWith the advent of edge devices and artificial intelligence, the data deluge is a reality, making energy-efficient computing systems a must-have. Unfortunately, classical von Neumann architectures suffer from the high cost of data transfers between memories and processing units. At the same time, CMOS scaling seems more and more challenging and costly to afford, limiting the chips' performance due to power consumption issues.In this context, bringing the computation directly inside or near memories (I/NMC) seems an appealing solution. However, data-centric applications require an important amount of non-volatile storage, and modern Flash memories suffer from scaling issues and are not very suited for I/NMC. On the other hand, emerging memory technologies such as ReRAM present very appealing memory performances, good scalability, and interesting I/NMC features. However, they suffer from variability issues and from a degraded density integration if an access transistor per bitcell (1T1R) is used to limit the sneak-path currents. This thesis work aims to overcome these two challenges. First, the variability impact on read and I/NMC operations is assessed and new robust and low-overhead ReRAM-based boolean operations are proposed. In the context of neural networks, new ReRAM-based neuromorphic accelerators are developed and characterized, with an emphasis on good robustness against variability, good parallelism, and high energy efficiency. Second, to resolve the density integration issues, an ultra-dense 3D 1T1R ReRAM-based Cube and its architecture are proposed, which can be used as a 3D NOR memory as well as a low overhead and energy-efficient I/NMC accelerator
Ali, Elsayed Sarah. "Fault Tolerance in Hardware Spiking Neural Networks". Electronic Thesis or Diss., Sorbonne université, 2021. http://www.theses.fr/2021SORUS310.
Texto completo da fonteArtificial Intelligence (AI) and machine learning algorithms are taking up the lion's share of the technology market nowadays, and hardware AI accelerators are foreseen to play an increasing role in numerous applications, many of which are mission-critical and safety-critical. This requires assessing their reliability and developing cost-effective fault tolerance techniques; an issue that remains largely unexplored for neuromorphic chips and Spiking Neural Networks (SNNs). A tacit assumption is often made that reliability and error-resiliency in Artificial Neural Networks (ANNs) are inherently achieved thanks to the high parallelism, structural redundancy, and the resemblance to their biological counterparts. However, prior work in the literature unraveled the falsity of this assumption and exposed the vulnerability of ANNs to faults. This requires assessing their reliability and developing cost-effective fault tolerance techniques; an issue that remains largely unexplored for neuromorphic chips and Spiking Neural Networks (SNNs). In this thesis, we tackle the subject of testing and fault tolerance in hardware SNNs. We start by addressing the issue of post-manufacturing test and behavior-oriented self-test of hardware neurons. Then we move on towards a global solution for the acceleration of testing and resiliency analysis of SNNs against hardware-level faults. We also propose a neuron fault tolerance strategy for SNNs, optimized for low area and power overhead. Finally, we present a hardware case-study which would be used as a platform for demonstrating fault-injection experiments and fault-tolerance capabilities
Falez, Pierre. "Improving spiking neural networks trained with spike timing dependent plasticity for image recognition". Thesis, Lille 1, 2019. http://www.theses.fr/2019LIL1I101.
Texto completo da fonteComputer vision is a strategic field, in consequence of its great number of potential applications which could have a high impact on society. This area has quickly improved over the last decades, especially thanks to the advances of artificial intelligence and more particularly thanks to the accession of deep learning. Nevertheless, these methods present two main drawbacks in contrast with biological brains: they are extremely energy intensive and they need large labeled training sets. Spiking neural networks are alternative models offering an answer to the energy consumption issue. One attribute of these models is that they can be implemented very efficiently on hardware, in order to build ultra low-power architectures. In return, these models impose certain limitations, such as the use of only local memory and computations. It prevents the use of traditional learning methods, for example the gradient back-propagation. STDP is a learning rule, observed in biology, which can be used in spiking neural networks. This rule reinforces the synapses in which local correlations of spike timing are detected. It also weakens the other synapses. The fact that it is local and unsupervised makes it possible to abide by the constraints of neuromorphic architectures, which means it can be implemented efficiently, but it also provides a solution to the data set labeling issue. However, spiking neural networks trained with the STDP rule are affected by lower performances in comparison to those following a deep learning process. The literature about STDP still uses simple data but the behavior of this rule has seldom been used with more complex data, such as sets made of a large variety of real-world images.The aim of this manuscript is to study the behavior of these spiking models, trained through the STDP rule, on image classification tasks. The main goal is to improve the performances of these models, while respecting as much as possible the constraints of neuromorphic architectures. The first contribution focuses on the software simulations of spiking neural networks. Hardware implementation being a long and costly process, using simulation is a good alternative in order to study more quickly the behavior of different models. Then, the contributions focus on the establishment of multi-layered spiking networks; networks made of several layers, such as those in deep learning methods, allow to process more complex data. One of the chapters revolves around the matter of frequency loss seen in several spiking neural networks. This issue prevents the stacking of multiple spiking layers. The center point then switches to a study of STDP behavior on more complex data, especially colored real-world image. Multiple measurements are used, such as the coherence of filters or the sparsity of activations, to better understand the reasons for the performance gap between STDP and the more traditional methods. Lastly, the manuscript describes the making of multi-layered networks. To this end, a new threshold adaptation mechanism is introduced, along with a multi-layer training protocol. It is proven that such networks can improve the state-of-the-art for STDP
Hirtzlin, Tifenn. "Digital Implementation of Neuromorphic systems using Emerging Memory devices". Thesis, université Paris-Saclay, 2020. http://www.theses.fr/2020UPAST071.
Texto completo da fonteWhile electronics has prospered inexorably for several decades, its leading source of progress will stop in the next coming years, due to the fundamental technological limits of transistors. Nevertheless, microelectronics is currently offering a major breakthrough: in recent years, memory technologies have undergone incredible progress, opening the way for multiple research venues in embedded systems. Additionally, a major feature for future years will be the ability to integrate different technologies on the same chip. new emerging memory devices that can be embedded in the core of the CMOS, such as Resistive Random Access Memory (RRAM) or Spin Torque Magnetic Tunnel Junction (STMRAM) based on naturally intelligent inmemory-computing architecture. Three braininspired algorithms are carefully examined: Bayesian reasoning binarized neural networks, and an approach that further exploits the intrinsic behavior of components, population coding of neurons. Each of these approaches explores different aspects of in-memory computing
Grassia, Filippo Giovanni. "Silicon neural networks : implementation of cortical cells to improve the artificial-biological hybrid technique". Thesis, Bordeaux 1, 2013. http://www.theses.fr/2013BOR14748/document.
Texto completo da fonteThis work has been supported by the European FACETS-ITN project. Within the frameworkof this project, we contribute to the simulation of cortical cell types (employingexperimental electrophysiological data of these cells as references), using a specific VLSIneural circuit to simulate, at the single cell level, the models studied as references in theFACETS project. The real-time intrinsic properties of the neuromorphic circuits, whichprecisely compute neuron conductance-based models, will allow a systematic and detailedexploration of the models, while the physical and analog aspect of the simulations, as opposedthe software simulation aspect, will provide inputs for the development of the neuralhardware at the network level. The second goal of this thesis is to contribute to the designof a mixed hardware-software platform (PAX), specifically designed to simulate spikingneural networks. The tasks performed during this thesis project included: 1) the methodsused to obtain the appropriate parameter sets of the cortical neuron models that can beimplemented in our analog neuromimetic chip (the parameter extraction steps was validatedusing a bifurcation analysis that shows that the simplified HH model implementedin our silicon neuron shares the dynamics of the HH model); 2) the fully customizablefitting method, in voltage-clamp mode, to tune our neuromimetic integrated circuits usinga metaheuristic algorithm; 3) the contribution to the development of the PAX systemin terms of software tools and a VHDL driver interface for neuron configuration in theplatform. Finally, it also addresses the issue of synaptic tuning for future SNN simulation