This 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.

La recherche dans le domaine de l’informatique neuro-inspirée suscite beaucoup d'intérêt depuis quelques années. Avec des applications potentielles dans des domaines tels que le traitement de données à grande échelle, la robotique ou encore les systèmes autonomes intelligents pour ne citer qu'eux, des paradigmes de calcul bio-inspirés sont étudies pour la prochaine génération solutions informatiques (post-Moore, non-Von Neumann) ultra-basse consommation. Dans ce travail, nous discutons les rôles que les différentes technologies de mémoire résistive non-volatiles émergentes (RRAM), notamment (i) Phase Change Memory (PCM), (ii) Conductive-Bridge Memory (CBRAM) et de la mémoire basée sur une structure Metal-Oxide (OXRAM) peuvent jouer dans des dispositifs neuromorphiques dédies. Nous nous concentrons sur l'émulation des effets de plasticité synaptique comme la potentialisation à long terme (Long Term Potentiation, LTP), la dépression à long terme (Long Term Depression, LTD) et la théorie STDP (Spike-Timing Dependent Plasticity) avec des synapses RRAM. Nous avons développé à la fois de nouvelles architectures de faiblement énergivore, des méthodologies de programmation ainsi que des règles d’apprentissages simplifiées inspirées de la théorie STDP spécifiquement optimisées pour certaines technologies RRAM. Nous montrons l’implémentation de systèmes neuromorphiques a grande échelle et efficace énergétiquement selon deux approches différentes: (i) des synapses multi-niveaux déterministes et (ii) des synapses stochastiques binaires. Des prototypes d'applications telles que l’extraction de schéma visuel et auditif complexe sont également montres en utilisant des réseaux de neurones impulsionnels (Feed-forward Spiking Neural Network, SNN). Nous introduisons également une nouvelle méthodologie pour concevoir des neurones stochastiques très compacts qui exploitent les caractéristiques physiques intrinsèques des appareils CBRAM. / Research in the field of neuromorphic- and cognitive- computing has generated a lot of interest in recent years. With potential application in fields such as large-scale data driven computing, robotics, intelligent autonomous systems to name a few, bio-inspired computing paradigms are being investigated as the next generation (post-Moore, non-Von Neumann) ultra-low power computing solutions. In this work we discuss the role that different emerging non-volatile resistive memory technologies (RRAM), specifically (i) Phase Change Memory (PCM), (ii) Conductive-Bridge Memory (CBRAM) and Metal-Oxide based Memory (OXRAM) can play in dedicated neuromorphic hardware. We focus on the emulation of synaptic plasticity effects such as long-term potentiation (LTP), long term depression (LTD) and spike-timing dependent plasticity (STDP) with RRAM synapses. We developed novel low-power architectures, programming methodologies, and simplified STDP-like learning rules, optimized specifically for some RRAM technologies. We show the implementation of large-scale energy efficient neuromorphic systems with two different approaches (i) deterministic multi-level synapses and (ii) stochastic-binary synapses. Prototype applications such as complex visual- and auditory- pattern extraction are also shown using feed-forward spiking neural networks (SNN). We also introduce a novel methodology to design low-area efficient stochastic neurons that exploit intrinsic physical effects of CBRAM devices.

Die sequenzielle Aktivität von Ortszellen im Hippocampus entspricht vielfach früheren Erlebnissen, was auf eine Rolle in Gedächtnisprozessen hinweist. Jüngere experimentelle Befunde zeigen allerdings, dass Zielorte in sequenzieller Aktivität überrepräsentiert sind. Dies legt eine Rolle dieser Aktivitätsmuster in der Verhaltensplanung nahe, wobei ein detailliertes Verständnis sowohl des Ursprungs als auch der Funktion von Ortszellsequenzen im Hippocampus bislang fehlt. Insbesondere ist nicht bekannt, welcher Mechanismus solche Sequenzen auf adaptive und konstruktive Weise generiert, wodurch effizientes Planen ermöglicht würde. Um der Beantwortung dieser Fragen näher zu kommen, stelle ich ein neu entwickeltes pulscodiertes Netzwerkmodell vor, in dem räumliches Lernen und die Generierung von Sequenzen untrennbar voneinander abhängig sind. Anhand von Simulationen zeige ich, dass dieses Modell die Erzeugung von noch nicht erlebten Sequenztrajektorien in bekannten Umgebungen erklärt, was deren Nutzen für flexible Pfadplanung hervorhebt. Zusätzlich stelle ich die Ergebnisse eines detaillierten Vergleichs zwischen simulierten neuronalen Pulsfolgen und experimentellen Daten auf der Ebene der Populationsdynamik vor. Diese Resultate zeigen, wie sequenzielle räumliche Repräsentationen durch die Interaktion zwischen lokaler oszillatorischer Dynamik und externen Einflüssen geprägt werden.:1. Introduction 2. Neurobiological and theoretical accounts of hippocampal function 3. A computational model of place-cell sequences for goal-finding 4. A statistical note on step size decoding in place-cell sequences 5. Summary and Discussion Bibliography / Hippocampal place-cell sequences observed during awake immobility often represent previous experience, suggesting a role in memory processes. However, recent reports of goals being overrepresented in sequential activity suggest a role in short-term planning, although a detailed understanding of the origins of hippocampal sequential activity and of its functional role is still lacking. In particular, it is unknown which mechanism could support efficient planning by generating place-cell sequences biased toward known goal locations, in an adaptive and constructive fashion. To address these questions, I propose a spiking network model of spatial learning and sequence generation as interdependent processes. Simulations show that this model explains the generation of never-experienced sequence trajectories in familiar environments and highlights their utility in flexible route planning. In addition, I report the results of a detailed comparison between simulated spike trains and experimental data, at the level of network dynamics. These results demonstrate how sequential spatial representations are shaped by the interaction between local oscillatory dynamics and external inputs.:1. Introduction 2. Neurobiological and theoretical accounts of hippocampal function 3. A computational model of place-cell sequences for goal-finding 4. A statistical note on step size decoding in place-cell sequences 5. Summary and Discussion Bibliography

Many modern neural simulators focus on the simulation of networks of spiking neurons on parallel hardware. Another important framework in computational neuroscience, rate-coded neural networks, is mostly difficult or impossible to implement using these simulators. We present here the ANNarchy (Artificial Neural Networks architect) neural simulator, which allows to easily define and simulate rate-coded and spiking networks, as well as combinations of both. The interface in Python has been designed to be close to the PyNN interface, while the definition of neuron and synapse models can be specified using an equation-oriented mathematical description similar to the Brian neural simulator. This information is used to generate C++ code that will efficiently perform the simulation on the chosen parallel hardware (multi-core system or graphical processing unit). Several numerical methods are available to transform ordinary differential equations into an efficient C++code. We compare the parallel performance of the simulator to existing solutions.

In dieser Arbeit werden mithilfe der Fokker-Planck-Gleichung die Statistiken, vor allem die Leistungsspektren, von Punktprozessen berechnet, die von mehrdimensionalen Integratorneuronen [Engl. integrate-and-fire (IF) neuron], Netzwerken von IF Neuronen und Entscheidungsfindungsmodellen erzeugt werden. Im Gehirn werden Informationen durch Pulszüge von Aktionspotentialen kodiert. IF Neurone mit radikal vereinfachter Erzeugung von Aktionspotentialen haben sich in Studien die auf Pulszeiten fokussiert sind als Standardmodelle etabliert. Eindimensionale IF Modelle können jedoch beobachtetes Pulsverhalten oft nicht beschreiben und müssen dazu erweitert werden. Im erste Teil dieser Arbeit wird eine Theorie zur Berechnung der Pulszugleistungsspektren von stochastischen, multidimensionalen IF Neuronen entwickelt. Ausgehend von der zugehörigen Fokker-Planck-Gleichung werden partiellen Differentialgleichung abgeleitet, deren Lösung sowohl die stationäre Wahrscheinlichkeitsverteilung und Feuerrate, als auch das Pulszugleistungsspektrum beschreibt. Im zweiten Teil wird eine Theorie für große, spärlich verbundene und homogene Netzwerke aus IF Neuronen entwickelt, in der berücksichtigt wird, dass die zeitlichen Korrelationen von Pulszügen selbstkonsistent sind. Neuronale Eingangströme werden durch farbiges Gaußsches Rauschen modelliert, das von einem mehrdimensionalen Ornstein-Uhlenbeck Prozess (OUP) erzeugt wird. Die Koeffizienten des OUP sind vorerst unbekannt und sind als Lösung der Theorie definiert. Um heterogene Netzwerke zu untersuchen, wird eine iterative Methode erweitert. Im dritten Teil wird die Fokker-Planck-Gleichung auf Binärentscheidungen von Diffusionsentscheidungsmodellen [Engl. diffusion-decision models (DDM)] angewendet. Explizite Gleichungen für die Entscheidungszugstatistiken werden für den einfachsten und analytisch lösbaren Fall von der Fokker-Planck-Gleichung hergeleitet. Für nichtliniear Modelle wird die Schwellwertintegrationsmethode erweitert. / This thesis is concerned with the calculation of statistics, in particular the power spectra, of point processes generated by stochastic multidimensional integrate-and-fire (IF) neurons, networks of IF neurons and decision-making models from the corresponding Fokker-Planck equations. In the brain, information is encoded by sequences of action potentials. In studies that focus on spike timing, IF neurons that drastically simplify the spike generation have become the standard model. One-dimensional IF neurons do not suffice to accurately model neural dynamics, however, the extension towards multiple dimensions yields realistic behavior at the price of growing complexity. The first part of this work develops a theory of spike-train power spectra for stochastic, multidimensional IF neurons. From the Fokker-Planck equation, a set of partial differential equations is derived that describes the stationary probability density, the firing rate and the spike-train power spectrum. In the second part of this work, a mean-field theory of large and sparsely connected homogeneous networks of spiking neurons is developed that takes into account the self-consistent temporal correlations of spike trains. Neural input is approximated by colored Gaussian noise generated by a multidimensional Ornstein-Uhlenbeck process of which the coefficients are initially unknown but determined by the self-consistency condition and define the solution of the theory. To explore heterogeneous networks, an iterative scheme is extended to determine the distribution of spectra. In the third part, the Fokker-Planck equation is applied to calculate the statistics of sequences of binary decisions from diffusion-decision models (DDM). For the analytically tractable DDM, the statistics are calculated from the corresponding Fokker-Planck equation. To determine the statistics for nonlinear models, the threshold-integration method is generalized.

yes / This Workshop on Membrane Computing, at the Conference of Unconventional Computation and Natural Computation (UCNC), 12th July 2016, Manchester, UK, is the second event of this type after the Workshop at UCNC 2015 in Auckland, New Zealand*. Following the tradition of the 2015 Workshop the Proceedings are published as technical report. The Workshop consisted of one invited talk and six contributed presentations (three full papers and three extended abstracts) covering a broad spectrum of topics in Membrane Computing, from computational and complexity theory to formal verification, simulation and applications in robotics. All these papers – see below, but the last extended abstract, are included in this volume. The invited talk given by Rudolf Freund, “P SystemsWorking in Set Modes”, presented a general overview on basic topics in the theory of Membrane Computing as well as new developments and future research directions in this area. Radu Nicolescu in “Distributed and Parallel Dynamic Programming Algorithms Modelled on cP Systems” presented an interesting dynamic programming algorithm in a distributed and parallel setting based on P systems enriched with adequate data structure and programming concepts representation. Omar Belingheri, Antonio E. Porreca and Claudio Zandron showed in “P Systems with Hybrid Sets” that P systems with negative multiplicities of objects are less powerful than Turing machines. Artiom Alhazov, Rudolf Freund and Sergiu Ivanov presented in “Extended Spiking Neural P Systems with States” new results regading the newly introduced topic of spiking neural P systems where states are considered. “Selection Criteria for Statistical Model Checker”, by Mehmet E. Bakir and Mike Stannett, presented some early experiments in selecting adequate statistical model checkers for biological systems modelled with P systems. In “Towards Agent-Based Simulation of Kernel P Systems using FLAME and FLAME GPU”, Raluca Lefticaru, Luis F. Macías-Ramos, Ionuţ M. Niculescu, Laurenţiu Mierlă presented some of the advatages of implementing kernel P systems simulations in FLAME. Andrei G. Florea and Cătălin Buiu, in “An Efficient Implementation and Integration of a P Colony Simulator for Swarm Robotics Applications" presented an interesting and efficient implementation based on P colonies for swarms of Kilobot robots. *http://ucnc15.wordpress.fos.auckland.ac.nz/workshop-on-membrane-computingwmc- at-the-conference-on-unconventional-computation-natural-computation/

Spiking neural networks using Leaky-Integrate-and-Fire (LIF) neurons and Spike-timing-depend Plasticity (STDP) learning, are commonly used as more biological possible networks. Compare to DNNs and RNNs, the LIF-STDP networks are models which are closer to the biological cortex. LIF-STDP neurons use spikes to communicate with each other, and they learn through the correlation among these pre- and post-synaptic spikes. Simulation of such networks usually requires high-performance supercomputers which are almost all based on von Neumann architecture that separates storage and computation. In von Neumann architecture solutions, memory access is the bottleneck even for highly optimized Application-Specific Integrated Circuits (ASICs). In this thesis, we propose an optimization method that can reduce the memory access cost by avoiding a dual-access pattern. In LIF-STDP networks, the weights usually are stored in the form of a two-dimensional matrix. Pre- and post-synaptic spikes trigger row and column access correspondingly. But this dual-access pattern is very costly for DRAM. We eliminate the column access by introducing a post-synaptic buffer and an approximation function. The post-synaptic spikes are recorded in the buffer and are processed at pre-synaptic spikes together with the row updates. This column update elimination method will introduce errors due to the limited buffer size. In our error analysis, the experiments show that the probability of introducing intolerable errors can be bounded to a very small number with proper buffer size and approximation function. We also present a performance analysis of the Column Update Elimination (CUE) optimization. The error analysis of the column updates elimination method is the main contribution of our work. / Spikande neurala nätverk som använder LIF-neuroner och STDP-inlärning, används vanligtvis som ett mer biologiskt möjligt nätverk. Jämfört med DNN och RNN är LIF-STDP-nätverken modeller närmare den biologiska cortex. LIFSTDP-neuroner använder spikar för att kommunicera med varandra, och de lär sig genom korrelationen mellan dessa pre- och postsynaptiska spikar. Simulering av sådana nätverk kräver vanligtvis högpresterande superdatorer som nästan alla är baserade på von Neumann-arkitektur som separerar lagring och beräkning. I von Neumanns arkitekturlösningar är minnesåtkomst flaskhalsen även för högt optimerade Application-Specific Integrated Circuits (ASIC). I denna avhandling föreslår vi en optimeringsmetod som kan minska kostnaden för minnesåtkomst genom att undvika ett dubbelåtkomstmönster. I LIF-STDPnätverk lagras vikterna vanligtvis i form av en tvådimensionell matris. Preoch postsynaptiska toppar kommer att utlösa rad- och kolumnåtkomst på motsvarande sätt. Men detta mönster med dubbel åtkomst är mycket dyrt i DRAM. Vi eliminerar kolumnåtkomsten genom att införa en postsynaptisk buffert och en approximationsfunktion. De postsynaptiska topparna registreras i bufferten och bearbetas vid presynaptiska toppar tillsammans med raduppdateringarna. Denna metod för eliminering av kolumnuppdatering kommer att introducera fel på grund av den begränsade buffertstorleken. I vår felanalys visar experimenten att sannolikheten för att införa oacceptabla fel kan begränsas till ett mycket litet antal med korrekt buffertstorlek och approximationsfunktion. Vi presenterar också en prestandaanalys av CUE-optimeringen. Felanalysen av elimineringsmetoden för kolumnuppdateringar är det huvudsakliga bidraget från vårt arbete

Neural networks have become increasingly adopted in society over the last few years. As neural networks consume a lot of energy to train, reducing the energy consumption of these networks is desirable from an environmental perspective. Spiking neural network is a type of neural network inspired by the human brain which is significantly more energy efficient than traditional neural networks. However, there is little research about how the hyper parameters of these networks affect the relationship between accuracy and energy. The aim of this report is therefore to analyse this relationship. To do this, we measure the energy usage of training several different spiking network models. The results of this study shows that the choice of hyper-parameters in a neural network does affect the efficiency of the network. While correlation between any individual factors and energy consumption is inconclusive, this work could be used as a springboard for further research in this area. / Under de senaste åren har neuronnät blivit allt vanligare i samhället. Eftersom neuronnät förbrukar mycket energi för att träna dem är det önskvärt ur miljösynpunkt att minska energiförbrukningen för dessa nätverk. Spikande neuronnät är en typ av neuronnät inspirerade av den mänskliga hjärnan som är betydligt mer energieffektivt än traditionella neuronnät. Det finns dock lite forskning om hur hyperparametrarna i dessa nätverk påverkar sambandet mellan noggrannhet och energi. Syftet med denna rapport är därför att analysera detta samband. För att göra detta mäter vi energiförbrukningen vid träning av flera olika modeller av spikande neuronnät-modeller. Resultaten av denna studie visar att valet av hyperparametrar i ett neuronnät påverkar nätverkets effektivitet. Även om korrelationen mellan enskilda faktorer och energiförbrukning inte är entydig kan detta arbete användas som en startpunkt för ytterligare forskning inom detta område.

京都大学 / 新制・課程博士 / 博士(情報学) / 甲第25444号 / 情博第882号 / 新制||情||148(附属図書館) / 京都大学大学院情報学研究科通信情報システム専攻 / (主査)教授 佐藤 高史, 教授 橋本 昌宜, 教授 新津 葵一 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM

No description available.