To see the other types of publications on this topic, follow the link: Neuron generation.
Dissertations / Theses on the topic 'Neuron generation'
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 'Neuron generation.'
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.
1
Nangla, Jyoti. "The study of neurogenesis in the rodent telencephalon." Thesis, University of Oxford, 1999. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.325881.
Full text
2
Stifani, Nicolas. "Generation of motor neuron diversity in the cervical spinal cord." Thesis, McGill University, 2012. http://digitool.Library.McGill.CA:80/R/?func=dbin-jump-full&object_id=106433.
Full textAbstract:
Motor neurons are important neuronal cell types in charge of controlling the contraction of effector muscles. In order to be able to generate complex movements specific groups of motor neurons (termed pools) must form precise connections with the muscles they innervate. Spinal cord motor neuron identities are specified by the coordination of extrinsic signals and intrinsic sets of transcription factors expressed by undifferentiated progenitors and maturing neurons. In an effort to elucidate the intrinsic mechanisms that control the formation of defined motor neuron pools, the detailed expression pattern of the runt-related transcription factor 1 (Runx1) was characterized. Runx1 is expressed in restricted populations of spinal motor neurons in the cervical segments during their post-mitotic development. Therefore, the expression of Runx1 enlightens the identity of specific motor neuron populations during their development. Loss of Runx1 function does not affect the survival of these motor neurons but results in a loss of expression of motor neuron-specific genes and a concomitant activation of interneuron-specific genes. Conversely, ectopic expression of Runx1 in the spinal cord of developing chick embryos suppresses interneuron gene expression and promotes motor-neuron differentiation programs. These results suggest that Runx1 is both necessary and sufficient to suppress interneuron-specific developmental programs and promote maintenance of motor neuron characteristics. These findings identify Runx1 as an important factor during the consolidation of selected spinal motor-neuron identities. As a result of these findings, this thesis provides an accurate description of spinal motor neuron identities during the period of their formation. Furthermore, these results suggest that maturing motor neurons must ensure the persistence of their motor identity throughout embryonic life by dynamically maintaining motor neuron gene expression and persistently suppressing interneuron developmental program.Les motoneurones sont des cellules nerveuses qui ont un rôle primordial : le contrôle de la contraction des muscles. Afin de réaliser des mouvements complexes, les motoneurones doivent conserver l'identité des muscles qu'ils innervent. L'identité des motoneurones de la moelle épinière s'acquière par l'action conjointe et coordonnée de signaux extrinsèques et de facteurs de transcription intracellulaires. Dans ce contexte, l'expression du facteur de transcription, runt-related transcription factor 1 (Runx1), a été étudié. Runx1 est exprimé de façon transitoire durant le développement post-mitotique embryonnaire de certaines populations spécifiques de motoneurones limités aux segments cervicaux de la moelle épinière. L'inactivation de la fonction de Runx1 n'affecte pas la survie de ces motoneurones mais résulte en une diminution de l'expression de certains gènes impliqués spécifiquement dans le développement des motoneurones ainsi qu'à une activation concomitante de l'expression de gènes impliqués exclusivement dans le programme de développement des inter-neurones. À l'inverse l'expression ectopique de Runx1 dans la moelle épinière d'embryons de poulet réprime l'expression de gènes spécifiques aux inter-neurones et stimule le programme de différentiation des motoneurones. L'ensemble de ces résultats suggère que Runx1 est non seulement nécessaire mais également suffisant pour supprimer le programme de différentiation des inter-neurones et promouvoir la maintenance de caractéristiques propres aux motoneurones. Cette thèse fournit une description précise de l'identité des motoneurones durant leur développement. Ces résultats présentent Runx1 comme un acteur important dans la consolidation de l'identité des motoneurones de la moelle épinière et suggèrent que les motoneurones en développement doivent maintenir la répression de l'expression de gènes impliqués dans développement des inter-neurones afin de conserver l'intégrité de leur identité.
3
Lyon, Alison Nicole. "Generation and Analysis of Motor Neuron Disease Models in Zebrafish." The Ohio State University, 2012. http://rave.ohiolink.edu/etdc/view?acc_num=osu1337276861.
Full text
4
Christou, Yiota Apostolou. "Generation of motor neurons from embryonic stem cells : application in studies of the motor neuron disease mechanism." Thesis, University of Sheffield, 2009. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.505426.
Full textAbstract:
Embryonic stem cells are pluripotent cells with the potential to differentiate into any cell type in the presence of appropriate stimulatory factors and environmental cues. Their broad developmental potential has led to the proposal that in the future, the use of human embryonic stem cells or their differentiated progeny may be beneficial in regenerative medicine. In particular, a current goal in the field of clinical neurology is to use stem cells in cell-based therapies for motor neuron disease (MND) or amyotrophic lateral ~clerosis. MND is a progressive neurodegenerative disease that specifically affects upper and lower motor neurons and leads ultimately to death from respiratory failure. Stem cellderived motor neurons could conceivably be used to replace the degenerated cells, to provide authentic substrates for drug development and screening and for furthering our understanding of disease mechanisms. However, to reliably and accurately culture motor neurons, the complex pathways by which differentiation occurs in vivo must be understood and reiterated in vitro to direct embryonic stem cells towards motor neurons. This thesis presents the work I have performed on the directed differentiation of embryonic stem cells towards motor neuron fates. I describe the various experimental approaches I took in attempts to produce motor neurons in vitro. My studies reveal that it is possible to deploy the signals used during normal development to direct the differentiation of both human and mouse embryonic stem cells into neural and neuronal cells, including motor neurons. Two major limitations precluded my analysis of pure motor neuron cultures: first, the high concentrations of the ventralising morphogen, SHH, apparently required to direct embryonic stem cells towards motor neuron fates, and second, the difficulties encountered in culturing purified motor neurons. However, using a mixed culture, I obtained evidence that motor neurons and their progenitors fail to survive in medium conditioned by mutant SOD1-G93A astrocytes.
5
Yoo, Raphael J. "Generation of fibronectin gradients on a patterned surface for neuron growth studies." Access to citation, abstract and download form provided by ProQuest Information and Learning Company; downloadable PDF file, p, 2005. http://proquest.umi.com/pqdweb?did=974436301&sid=6&Fmt=2&clientId=8331&RQT=309&VName=PQD.
Full text
6
Yang, Yujie. "Analysis of developmental and regenerative spinal motor neuron generation in zebrafish larvae." Thesis, University of Edinburgh, 2017. http://hdl.handle.net/1842/23591.
Full textAbstract:
In contrast to mammals, adult zebrafish are able to regenerate motor neurons and regain swimming ability within 6 weeks after a spinal cord injury. During this regenerative process, a range of developmental signals such as dopamine and serotonin are found to be re-deployed. This makes the research of embryonic signals become essential for the promotion of regeneration in the future. In my research, I am interested in identifying genes that are important for motor neuron development and motor axon differentiation. I also aimed to study the ability of zebrafish larvae to regenerate spinal motor neurons, and whether they can be used to study the essential developmental cues and the mechanisms underlying successful functional recovery. Motor axons grow out of the spinal cord in a motor neuron subtype specific manner and innervate different muscle groups to facilitate locomotor movements. To find genes and important pathways involved in motor neuron generation and axon development in zebrafish, we conducted an ENU-induced mutagenesis screen in islet-1:GFP transgenic zebrafish, in which a subset of dorsally projecting motor neurons are labelled. We have discovered 6 mutants displaying delayed or inhibited appearance of secondary motor neurons and/or motor axon deficits among 111 F2 families screened. Through subsequent mutant phenotypical analysis, I focused my study in two mutant lines manifesting a lack of islet-1:GFP motor neurons, and an absence of islet-1:GFP motor axons. I used various molecular markers to characterise the mutant phenotypes and observed several additional anatomical defects. I also initiated the study of causative mutation analysis based on the candidate gene list generated from Next Generation Sequencing (NGS). To gain an insight of the genes’ role in motor neuron development and axonal differentiation, I started functional analyses in order to confirm genes that are responsible for the observed motor neuron/axon phenotypes, and I have achieved some promising preliminary results. Motor neurons are generated from the motor neuron progenitor domain (pMN). This neurogenesis process sharply declines at 48 hours post-fertilisation (hpf), while pMN progenitor cells continue to proliferate to produce oligodendrocytes. By inflicting a mechanical lesion in the spinal cord of zebrafish larvae, we demonstrated that they are capable of regenerate new motor neurons and achieve full functional recovery within 48 hours following the injury, sharing similar mechanisms to that of the adult zebrafish. I further studied oligodendrocyte generation and found that pMN domain is able to switch from oligodendrogenesis to motor neuron generation after a spinal lesion. This demonstrates the high plasticity of the pMN domain. Interestingly, the generation of dorsal Pax2-positive interneurons was not altered after the lesion, suggesting that the regenerative potential differs in different progenitor domains. This study showed that the motor neuron regenerative process in zebrafish larvae is robust and they can be used for studying motor neuron regeneration. Taken together, the discovery of the genes from our screen will provide insights to the developmental cues that are involved in motor neuron generation and axon growth. Furthermore, spinal cord lesion in larval zebrafish larvae is established as a regenerative model that can be utilized to dissect the roles and mechanisms of these signals and pathways in the promotion of motor neuron regeneration.
7
Šagát, Martin. "Návrh generativní kompetitivní neuronové sítě pro generování umělých EKG záznamů." Master's thesis, Vysoké učení technické v Brně. Fakulta elektrotechniky a komunikačních technologií, 2020. http://www.nusl.cz/ntk/nusl-413114.
Full textAbstract:
The work deals with the generation of ECG signals using generative adversarial networks (GAN). It examines in detail the basics of artificial neural networks and the principles of their operation. It theoretically describes the use and operation and the most common types of failures of generative adversarial networks. In this work, a general procedure of signal preprocessing suitable for GAN training was derived, which was used to compile a database. In this work, a total of 3 different GAN models were designed and implemented. The results of the models were visually displayed and analyzed in detail. Finally, the work comments on the achieved results and suggests further research direction of methods dealing with the generation of ECG signals.
8
Michalikova, Martina. "Mechanisms of spikelet generation in cortical pyramidal neurons." Doctoral thesis, Humboldt-Universität zu Berlin, Lebenswissenschaftliche Fakultät, 2017. http://dx.doi.org/10.18452/17753.
Full textAbstract:
Unter Spikelets versteht man kleine Depolarisationen mit einer Spike-ähnlichen Wellenform, die man in intrazellulären Ableitungen von verschiedenen Neuronentypen messen kann. In kortikalen Pyramidenzellen wurde ausgeprägte Spikelet-Aktivität nachgewiesen, die erheblich das Membranpotential beeinflussen kann (Crochet et al., 2004; Epsztein et al., 2010; Chorev and Brecht, 2012). Nichtsdestotrotz bleibt der Ursprung von Spikelets in diesen Neuronen unbekannt. In der vorgelegten Arbeit nutzte ich theoretische Modellierung um die Mechanismen von Spikelet-Erzeugung in Pyramidenzellen zu untersuchen. Zuerst sah ich die verschiedenen Hypothesen über den Ursprung von Spikelets durch. In der Literatur entdeckte ich zwei verschiedene Typen von Spikelets. Diese Arbeit konzentriert sich auf den häufiger vorkommenden Typ von Spikelets, welcher durch relativ große Amplituden gekennzeichnet ist. Die Eigenschaften dieser Spikelets passen am besten zu einem axonal Erzeugungsmechanismus. Im zweiten Kapitel widmete ich mich der Hypothese, dass somatische Spikelets axonalen Ursprungs mit somato-dendritischen Inputs hervorgerufen werden können. Ich identifizierte Bedingungen, die es erlauben ein Aktionspotential (AP) am Initialsegment vom Axon (AIS) zu initiieren, welches sich entlang des Axons ausbreitet, aber kein AP im Soma auslöst. Schließlich simulierte ich extrazelluläre Wellenformen von APs und Spikelets und verglich sie mit experimentellen Daten (Chorev and Brecht, 2012). Dieser Vergleich zeigte auf, dass die extrazellulären Wellenformen von Spikelets, die innerhalb einer Zellen am AIS erzeugt werden, gut zu den Daten passen. Zusammenfassend unterstützen meine Ergebnisse die Hypothese, dass Spikelets in Pyramidenzellen am AIS entstehen. Dieser Mechanismus könnte ein Mittel zum Energiesparen bei der Erzeugung von Output-APs sein. Außerdem könnte dadurch die dendritische Plastizität, die auf der Rückwärtspropagierung von APs beruht, reguliert werden.Spikelets are transient spike-like depolarizations of small amplitudes that can be measured in somatic intracellular recordings of many neuron types. Pronounced spikelet activity has been demonstrated in cortical pyramidal neurons in vivo (Crochet et al., 2004; Epsztein et al., 2010; Chorev and Brecht, 2012), influencing membrane voltage dynamics including action potential initiation. Nevertheless, the origin of spikelets in these neurons remains elusive. In thi thesis, I used computational modeling to examine the mechanisms of spikelet generation in pyramidal neurons. First, I reviewed the hypotheses previously suggested to explain spikelet origin. I discovered two qualitatively different spikelet types described in the experimental literature. This thesis focuses on the more commonly reported spikelet type, characterized by relatively large amplitudes of up to 20 mV. I found that the properties of these spikelets fit best to an axonal generation mechanism. Second, I explored the hypothesis that somatic spikelets of axonal origin can be evoked with somato-dendritic inputs. I identified the conditions allowing these orthodromic inputs to trigger an action potential at the axon initial segment, which propagates along the axon to the postsynaptic targets, but fails to elicit an action potential in the soma and the dendrites. Third, I simulated extracellular waveforms of action potentials and spikelets and compared them to experimental data (Chorev and Brecht, 2012). This comparison demonstrated that the extracellular waveforms of single-cell spikelets of axonal origin are consistent with the data. Together, my results suggest that spikelets in pyramidal neurons might originate at the axon initial segment within a single cell. Such a mechanism might be a way of reducing the energetic costs associated with the generation of output action potentials. Moreover, it might allow to control the dendritic plasticity by backpropagating action potentials.
9
Shao, Jie. "Putative Role of Connectivity in the Generation of Spontaneous Bursting Activity in an Excitatory Neuron Population." Thesis, Georgia Institute of Technology, 2004. http://hdl.handle.net/1853/5086.
Full textAbstract:
Population-wide synchronized rhythmic bursts of electrical activity are present in a
variety of neural circuits. The proposed general mechanisms for
rhythmogenesis are often attributed to intrinsic and synaptic properties. For example,
the recurrent excitation through excitatory synaptic connections determines
burst initiation, and the slower kinetics of ionic currents or synaptic depression
results in burst termination. In such theories, a slow recovery process is essential
for the slow dynamics associated with bursting.
This thesis presents a new hypothesis that depends on
the connectivity pattern among neurons rather than a slow kinetic process to achieve
the network-wide bursting. The thesis
begins with an introduction of bursts of electrical activity in a purely excitatory
neural network and existing theories explaining this phenomenon. It then covers
the small-world approach, which is applied to modify the network structure in the simulation,
and the Morris-Lecar (ML) neuron model, which is used as the component cells in the network.
Simulation results of the dependence of bursting activity on network connectivity,
as well as the inherent network properties explaining this dependence are described.
This work shows that the network-wide bursting activity emerges in the small-world network
regime but not in the regular or random networks, and this small-world bursting primarily results
from the uniform random distribution of long-range connections in the network, as well as
the unique dynamics in the ML model. Both attributes foster progressive synchronization in
firing activity throughout the network during a burst, and this synchronization may terminate a burst in the absence of an obvious slow recovery process. The thesis concludes with possible future work.
10
Wagner, Justin. "Whole Exome Sequencing to Identify Disease-Causing Mutations in Lower Motor Neuron Disease and Peripheral Neuropathy." Thesis, Université d'Ottawa / University of Ottawa, 2016. http://hdl.handle.net/10393/34124.
Full textAbstract:
Lower motor neuron diseases and peripheral neuropathies are two groups of diseases that include multiple rare disorders where many causes are unknown and definitive treatments are unavailable. Understanding the molecular etiology of these genetic diseases provides an opportunity for rapid diagnosis, preconception genetic counseling and, in a subset, direction for the development of future treatment options. The recent introduction of whole exome sequencing (WES) marks a new era in Mendelian genetic disease research as the majority of the coding region of the genome can be sequenced in a timely and cost-effective manner. In this study, WES was used to investigate the molecular etiology of a cohort of 37 patients presenting with lower motor neuron disease or peripheral neuropathy. A molecular diagnosis was determined for seven patients informing the diagnostic utility of WES. Novel phenotypes were found for three genes originally associated with a different disorder. Finally, the foundation has been laid, through the use of functional studies and large scale data-sharing, to identify novel disease-causing genes for lower motor neuron disease and peripheral neuropathy.
11
Wang, Ling. "Microchannel enhanced neuron-computer interface: design, fabrication, biophysics of signal generation, signal strength optimization, and its applications to ion-channel screening and basic neuroscience research." Doctoral thesis, Universitat Politècnica de Catalunya, 2011. http://hdl.handle.net/10803/52810.
Full textAbstract:
En el presente trabajo, utilizamos técnicas de microfabricación, simulaciones
numéricas, experimentos de electrofisiología para explorar la viabilidad en me-
jorar la interface ordenador-neurona a través de microcanales, y la biofísica para
la generación de señales en los dispositivos con microcanales. También demos-
tramos que los microcanales pueden ser usados como una técnica prometedora
con alto rendimiento en el muestreo automático de canales iónicos a nivel subce-
lular. Finalmente, se ha diseñado, fabricado y probado el micropozo-microcanal
como modificación adicional a los arreglos de multielectrodos, permitiendo una
alta ganancia en la relación señal/ ruido (en inglés Signal to Noise Ratio SNR),
y el registro de múltiples-lugares en poblaciones de baja densidad de redes neu-
ronales del hipocampo in vitro.
Primero, demostramos que son de alto rendimiento los microcanales de bajo
costo con interface neurona-electrodo, para el registro extracelular de la activi-
dad neuronal con baja complexidad, por periodos estables de larga duración y
con alta ganancia SNR.
En seguida, se realiza un estudio mediante experimentos y simulaciones nu-
méricas de la biofísica para la generación de las señales obtenidas de los dispositi-
vos con microcanales. Basados en los resultados, racionalizamos y demostramos
como es que la longitud del canal (siendo 200 μm) y la sección transversal del
microcanal (siendo 12 μm2) canaliza a los potenciales de acción para estar
dentro del rango de milivolts. A pesar del bajo grado de complexidad envuelto
en la fabricación y aplicación, los dispositivos con microcanales otorgan una sola
media de valor SNR de 101 76, lo cual es favorablemente comparable con la
SNR que se obtiene de desarrollos recientes que emplean electrodos curados con
CNT y Si-NWFETs.
Más aún, nosotros demostramos que el microcanal es una técnica promete-
dora para el alto rendimiento del muestro automático de canales iónicos a nivel
subcelular: (1) Información experimental y simulaciones numéricas sugieren que
las señales registradas sólo afectan los parches membranales localizados dentro
del microcanal o alrededor de 100 μm de las entradas del microcanal. (2) La
transferencia de masa de los componentes químicos en los microcanales fue ana-
lizada por experimentos y simulaciones FEM. Los resultados muestran que los
microcanales que contienen glía y tejido neuronal pueden funcionar como barre-
ra de fluido/química. Los componentes químicos pueden ser solamente aplicados
a diferentes compartimentos a nivel subcelular.
Finalmente, basado en simulaciones numéricas y resultados experimentales,
se propone que del micropozo-microcanal, obtenido de la modificación de MEA
(MWMC-MEA), la longitud óptima del canal debe ser 0,3 mm y la posición
1
óptima del electrodo intracanal, hacia la entrada más cercana del microcanal,
debe ser 0,1 mm. Nosotros fabricamos un prototipo de MWMC-MEA, cuyo hoyo
pasante sobre las películas de Polydimethylsiloxane (PDMS) fue microtrabajado
a través de la técnica de grabados reactivos de plasma de iones. La baja densidad
del cultivo (57 neuronas /mm2) en el MWMC-MEAs permitió que las neuronas
vivieran al menos 14 días, con lo que la señal neuronal con la máxima SNR
obtenida fue de 142.
2In this present work, we used microfabrication techniques, numerical simulations, electrophysiological experiments to explore the feasibility of enhancing neuron-computer interfaces with microchannels and the biophysics of the signal generation in microchannel devices. We also demonstrate the microchannel can be used as a promising technique for high-throughput automatic ion-channel screening at subcellular level. Finally, a microwell-microchannel enhanced multielectrode array allowing high signal-to-noise ratio (SNR), multi-site recording from the low-density hippocampal neural network in vitro was designed, fabricated and tested. First, we demonstrate using microchannels as a low-cost neuron-electrode interface to support low-complexity, long-term-stable, high SNR extracellular recording of neural activity, with high-throughput potential. Next, the biophysics of the signal generation of microchannel devices was studied by experiments and numerical simulations. Based on the results, we demonstrate and rationalize how channels with a length of 200 μm and channel cross section of 12 μm2 yielded spike sizes in the millivolt range. Despite the low degree of complexity involved in their fabrication and use, microchannel devices provided a single-unit mean SNR of 101 76, which compares favourably with the SNR obtained from recent developments employing CNT-coated electrodes and Si-NWFETs. Moreover, we further demonstrate that the microchannel is a promising technique for high-throughput automatic ion-channel screening at subcellular level: (1) Experimental data and numerical simulations suggest that the recorded signals are only affected by the membrane patches located inside the microchannel or within 100 μm to the microchannel entrances. (2) The mass transfer of chemical compounds in microchannels was analyzed by experiments and FEM simulations. The results show that the microchannel threaded by glial and neural tissue can function as fluid/chemical barrier. Thus chemical compounds can be applied to different subcellular compartments exclusively. Finally, a microwell-microchannel enhanced MEA (MWMC-MEA), with the optimal channel length of 0.3 mm and the optimal intrachannel electrode position of 0.1 mm to the nearest channel entrance, was proposed based on numerical simulation and experiment results. We fabricated a prototype of the MWMCMEA, whose through-hole feature of Polydimethylsiloxane film (PDMS) was micromachined by reactive-ion etching. The low-density culture (57 neurons/mm2) were survived on the MWMC-MEAs for at least 14 days, from which the neuronal signal with the maximum SNR of 142 was obtained.
12
Almeida, Lirio Onofre Baptista de. "Instrumentação computacional de tempo real integrada para experimentos com o duto óptico da mosca." Universidade de São Paulo, 2013. http://www.teses.usp.br/teses/disponiveis/76/76132/tde-19032013-155627/.
Full textAbstract:
Este trabalho descreve as pesquisas e desenvolvimentos em instrumentação eletrônica computacional, realizados para viabilizar experiências na área de neurobiofísica, tendo como objetivos principais a geração de estímulos visuais para invertebrados e a captação de sinais eletrofisiológicos gerados por sistemas biológicos sensoriais submetidos a estímulos. Trata-se de um conjunto de equipamentos que, operando de maneira integrada, são capazes de fornecer e sincronizar estímulos, realizar a aquisição dos dados de sinais neurais a serem utilizados para controle e análise em experiências in vivo\" nos estudos da visão de invertebrados no Laboratório de Neurobiofísica - DipteraLab do IFSC. A integração desta instrumentação eletrônica visa facilitar a sua utilização durante os experimentos, permitindo o acompanhamento das aquisições de dados neurais, viabilizando a realização de experimentos com alterações dos estímulos através de realimentação em tempo real.This work describes the research and development of computational instrumentation to be used in experimental neurobiophysics. The developed electronic modules operate in an integrated manner and are used to generate visual stimuli for invertebrates and capture electrophysiological signals generated by biological systems subjected to sensory stimuli. They are able to provide synchronized stimuli and perform data acquisition of neural signals events to be used for control and analysis of vision experiments with invertebrates at the Laboratory of Neurobiophysics Dipteralab Laboratory, at the IFSC. The integration of electronic instrumentation facilitate its use during experiments allowing, through its monitoring capabilities of the neural data acquisition, the realization of experiments with real time stimuli changes through feedback. The possibility to perform pre-analyses of neural responses in behavioral closed loop experiments is also implemented.
13
Shmelkov, Konstantin. "Approches pour l'apprentissage incrémental et la génération des images." Thesis, Université Grenoble Alpes (ComUE), 2019. http://www.theses.fr/2019GREAM010/document.
Full textAbstract:
Cette thèse explore deux sujets liés dans le contexte de l'apprentissage profond : l'apprentissage incrémental et la génération des images. L'apprentissage incrémental étudie l'entrainement des modèles dont la fonction objective évolue avec le temps (exemple : Ajout de nouvelles catégories à une tâche de classification). La génération d'images cherche à apprendre une distribution d'images naturelles pour générer de nouvelles images ressemblant aux images de départ.L’apprentissage incrémental est un problème difficile dû au phénomène appelé l'oubli catastrophique : tout changement important de l’objectif au cours de l'entrainement provoque une grave dégradation des connaissances acquises précédemment. Nous présentons un cadre d'apprentissage permettant d'introduire de nouvelles classes dans un réseau de détection d'objets. Il est basé sur l’idée de la distillation du savoir pour lutter les effets de l’oubli catastrophique : une copie fixe du réseau évalue les anciens échantillons et sa sortie est réutilisée dans un objectif auxiliaire pour stabiliser l’apprentissage de nouvelles classes. Notre framework extrait ces échantillons d'anciennes classes à la volée à partir d'images entrantes, contrairement à d'autres solutions qui gardent un sous-ensemble d'échantillons en mémoire.Pour la génération d’images, nous nous appuyons sur le modèle du réseau adverse génératif (en anglais generative adversarial network ou GAN). Récemment, les GANs ont considérablement amélioré la qualité des images générées. Cependant, ils offrent une pauvre couverture de l'ensemble des données : alors que les échantillons individuels sont de grande qualité, certains modes de la distribution d'origine peuvent ne pas être capturés. De plus, contrairement à la mesure de vraisemblance couramment utilisée pour les modèles génératives, les méthodes existantes d'évaluation GAN sont axées sur la qualité de l'image et n'évaluent donc pas la qualité de la couverture du jeu de données. Nous présentons deux approches pour résoudre ces problèmes.La première approche évalue les GANs conditionnels à la classe en utilisant deux mesures complémentaires basées sur la classification d'image - GAN-train et GAN-test, qui approchent respectivement le rappel (diversité) et la précision (qualité d'image) des GANs. Nous évaluons plusieurs approches GANs récentes en fonction de ces deux mesures et démontrons une différence de performance importante. De plus, nous observons que la difficulté croissante du jeu de données, de CIFAR10 à ImageNet, indique une corrélation inverse avec la qualité des GANs, comme le montre clairement nos mesures.Inspirés par notre étude des modèles GANs, la seconde approche applique explicitement la couverture d'un jeux de données pendant la phase d'entrainement de GAN. Nous développons un modèle génératif combinant la qualité d'image GAN et l'architecture VAE dans l'espace latente engendré par un modèle basé sur le flux, Real-NVP. Cela nous permet d’évaluer une vraisemblance correcte et d’assouplir simultanément l’hypothèse d’indépendance dans l’espace RVB qui est courante pour les VAE. Nous obtenons le score Inception et la FID en concurrence avec les GANs à la pointe de la technologie, tout en maintenant une bonne vraisemblance pour cette classe de modèlesThis dissertation explores two related topics in the context of deep learning: incremental learning and image generation. Incremental learning studies training of models with the objective function evolving over time, eg, addition of new categories to a classification task. Image generation seeks to learn a distribution of natural images for generating new images resembling original ones.Incremental learning is a challenging problem due to the phenomenon called catastrophic forgetting: any significant change to the objective during training causes a severe degradation of previously learned knowledge. We present a learning framework to introduce new classes to an object detection network. It is based on the idea of knowledge distillation to counteract catastrophic forgetting effects: fixed copy of the network evaluates old samples and its output is reused in an auxiliary loss to stabilize learning of new classes. Our framework mines these samples of old classes on the fly from incoming images, in contrast to other solutions that keep a subset of samples in memory.On the second topic of image generation, we build on the Generative Adversarial Network (GAN) model. Recently, GANs significantly improved the quality of generated images. However, they suffer from poor coverage of the dataset: while individual samples have great quality, some modes of the original distribution may not be captured. In addition, existing GAN evaluation methods are focused on image quality, and thus do not evaluate how well the dataset is covered, in contrast to the likelihood measure commonly used for generative models. We present two approaches to address these problems.The first method evaluates class-conditional GANs using two complementary measures based on image classification - GAN-train and GAN-test, which approximate recall (diversity) and precision (quality of the image) of GANs respectively. We evaluate several recent GAN approaches based on these two measures, and demonstrate a clear difference in performance. Furthermore, we observe that the increasing difficulty of the dataset, from CIFAR10 over CIFAR100 to ImageNet, shows an inverse correlation with the quality of the GANs, as clearly evident from our measures.Inspired by our study of GAN models, we present a method to explicitly enforce dataset coverage during the GAN training phase. We develop a generative model that combines GAN image quality with VAE architecture in the feature space engendered by a flow-based model Real-NVP. This allows us to evaluate a valid likelihood and simultaneously relax the independence assumption in RGB space which is common for VAEs. We achieve Inception score and FID competitive with state-of-the-art GANs, while maintaining good likelihood for this class of models
14
Garcia, Torres Douglas. "Generation of Synthetic Data with Generative Adversarial Networks." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2018. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-254366.
Full textAbstract:
The aim of synthetic data generation is to provide data that is not real for cases where the use of real data is somehow limited. For example, when there is a need for larger volumes of data, when the data is sensitive to use, or simply when it is hard to get access to the real data. Traditional methods of synthetic data generation use techniques that do not intend to replicate important statistical properties of the original data. Properties such as the distribution, the patterns or the correlation between variables, are often omitted. Moreover, most of the existing tools and approaches require a great deal of user-defined rules and do not make use of advanced techniques like Machine Learning or Deep Learning. While Machine Learning is an innovative area of Artificial Intelligence and Computer Science that uses statistical techniques to give computers the ability to learn from data, Deep Learning is a closely related field based on learning data representations, which may serve useful for the task of synthetic data generation. This thesis focuses on one of the most interesting and promising innovations of the last years in the Machine Learning community: Generative Adversarial Networks. An approach for generating discrete, continuous or text synthetic data with Generative Adversarial Networks is proposed, tested, evaluated and compared with a baseline approach. The results prove the feasibility and show the advantages and disadvantages of using this framework. Despite its high demand for computational resources, a Generative Adversarial Networks framework is capable of generating quality synthetic data that preserves the statistical properties of a given dataset.Syftet med syntetisk datagenerering är att tillhandahålla data som inte är verkliga i fall där användningen av reella data på något sätt är begränsad. Till exempel, när det finns behov av större datamängder, när data är känsliga för användning, eller helt enkelt när det är svårt att få tillgång till den verkliga data. Traditionella metoder för syntetiska datagenererande använder tekniker som inte avser att replikera viktiga statistiska egenskaper hos de ursprungliga data. Egenskaper som fördelningen, mönstren eller korrelationen mellan variabler utelämnas ofta. Dessutom kräver de flesta av de befintliga verktygen och metoderna en hel del användardefinierade regler och använder inte avancerade tekniker som Machine Learning eller Deep Learning. Machine Learning är ett innovativt område för artificiell intelligens och datavetenskap som använder statistiska tekniker för att ge datorer möjlighet att lära av data. Deep Learning ett närbesläktat fält baserat på inlärningsdatapresentationer, vilket kan vara användbart för att generera syntetisk data. Denna avhandling fokuserar på en av de mest intressanta och lovande innovationerna från de senaste åren i Machine Learning-samhället: Generative Adversarial Networks. Generative Adversarial Networks är ett tillvägagångssätt för att generera diskret, kontinuerlig eller textsyntetisk data som föreslås, testas, utvärderas och jämförs med en baslinjemetod. Resultaten visar genomförbarheten och visar fördelarna och nackdelarna med att använda denna metod. Trots dess stora efterfrågan på beräkningsresurser kan ett generativt adversarialnätverk skapa generell syntetisk data som bevarar de statistiska egenskaperna hos ett visst dataset.
15
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 textAbstract:
Des avancées significatives sur les réseaux de neurones profonds ont récemment permis le développement de technologies importantes comme les voitures autonomes et les assistants personnels intelligents basés sur la commande vocale. La plupart des succès en apprentissage profond concernent la prédiction, alors que les percées initiales viennent des modèles génératifs. Actuellement, même s'il existe des outils puissants dans la littérature des modèles génératifs basés sur les réseaux profonds, ces techniques sont essentiellement utilisées pour la prédiction ou pour générer des objets connus (i.e., des images de haute qualité qui appartiennent à des classes connues) : un objet généré qui est à priori inconnu est considéré comme une erreur (Salimans et al., 2016) ou comme un objet fallacieux (Bengio et al., 2013b). En d'autres termes, quand la prédiction est considérée comme le seul objectif possible, la nouveauté est vue comme une erreur - que les chercheurs ont essayé d'éliminer au maximum. Cette thèse défends le point de vue que, plutôt que d'éliminer ces nouveautés, on devrait les étudier et étudier le potentiel génératif des réseaux neuronaux pour créer de la nouveauté utile - particulièrement sachant l'importance économique et sociétale de la création d'objets nouveaux dans les sociétés contemporaines. Cette thèse a pour objectif d'étudier la génération de la nouveauté et sa relation avec les modèles de connaissance produits par les réseaux neurones profonds génératifs. Notre première contribution est la démonstration de l'importance des représentations et leur impact sur le type de nouveautés qui peuvent être générées : une conséquence clé est qu'un agent créatif a besoin de re-représenter les objets connus et utiliser cette représentation pour générer des objets nouveaux. Ensuite, on démontre que les fonctions objectives traditionnelles utilisées dans la théorie de l'apprentissage statistique, comme le maximum de vraisemblance, ne sont pas nécessairement les plus adaptées pour étudier la génération de nouveauté. On propose plusieurs alternatives à un niveau conceptuel. Un deuxième résultat clé est la confirmation que les modèles actuels - qui utilisent les fonctions objectives traditionnelles - peuvent en effet générer des objets inconnus. Cela montre que même si les fonctions objectives comme le maximum de vraisemblance s'efforcent à éliminer la nouveauté, les implémentations en pratique échouent à le faire. A travers une série d'expérimentations, on étudie le comportement de ces modèles ainsi que les objets qu'ils génèrent. En particulier, on propose une nouvelle tâche et des métriques pour la sélection de bons modèles génératifs pour la génération de la nouveauté. Finalement, la thèse conclue avec une série d'expérimentations qui clarifie les caractéristiques des modèles qui génèrent de la nouveauté. Les expériences montrent que la sparsité, le niveaux du niveau de corruption et la restriction de la capacité des modèles tuent la nouveauté et que les modèles qui arrivent à reconnaître des objets nouveaux arrivent généralement aussi à générer de la nouveautéIn 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
16
Daley, Jr John. "Generating Synthetic Schematics with Generative Adversarial Networks." Thesis, Högskolan Kristianstad, Fakulteten för naturvetenskap, 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:hkr:diva-20901.
Full textAbstract:
This study investigates synthetic schematic generation using conditional generative adversarial networks, specifically the Pix2Pix algorithm was implemented for the experimental phase of the study. With the increase in deep neural network’s capabilities and availability, there is a demand for verbose datasets. This in combination with increased privacy concerns, has led to synthetic data generation utilization. Analysis of synthetic images was completed using a survey. Blueprint images were generated and were successful in passing as genuine images with an accuracy of 40%. This study confirms the ability of generative neural networks ability to produce synthetic blueprint images.
17
Machado, Lucas Corrêa Netto. "Método de segmentações geométricas sucessivas para treinamento de redes neurais artificiais." Universidade Federal de Juiz de Fora (UFJF), 2013. https://repositorio.ufjf.br/jspui/handle/ufjf/4164.
Full textAbstract:
Submitted by Renata Lopes (renatasil82@gmail.com) on 2017-04-24T19:36:29Z
No. of bitstreams: 1
lucascorreanettomachado.pdf: 1851458 bytes, checksum: 2a8b67f0adf8343c28d4e1121a757f6d (MD5)Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2017-04-25T15:23:10Z (GMT) No. of bitstreams: 1 lucascorreanettomachado.pdf: 1851458 bytes, checksum: 2a8b67f0adf8343c28d4e1121a757f6d (MD5)
Made available in DSpace on 2017-04-25T15:23:10Z (GMT). No. of bitstreams: 1 lucascorreanettomachado.pdf: 1851458 bytes, checksum: 2a8b67f0adf8343c28d4e1121a757f6d (MD5) Previous issue date: 2013-11-22
CAPES - Coordenação de Aperfeiçoamento de Pessoal de Nível Superior
Este trabalho apresenta uma técnica para treinamento de Redes Neurais Artificiais (RNA), capaz de obter os parâmetros da rede através dos dados disponíveis para treinamento, sem necessidade de estabelecer a arquitetura da rede a priori, denominado Método de Segmentações Geométricas Sucessivas (MSGS). O MSGS agrupa os dados de cada classe em Hipercaixa (HC) onde cada caixa é alinhada de acordo com os eixos de maior distribuição de seu conjunto de pontos. Sendo as caixas linearmente separáveis, um hiperplano de separação é identificado originando um neurônio. Caso não seja possível a separação por um único hiperplano, uma técnica de quebra é aplicada para dividir os dados em classes menores para obter novas HCs. Para cada subdivisão novos neurônios são adicionados à rede. Os resultados dos testes realizados apontam para um método rápido e com alta taxa de sucesso.
This work presents a technique for Artificial Neural Network (ANN) training, able to get the network parameters from the available data for training, without establishing the network architecture a priori, called Successive Geometric Segmentation Method (SGSM). The SGSM groups the data of each class into hyperboxes (HB) aligned in accordance with the largest axis of its points distribution. If the HB are linearly separable, a separating hyperplane may be identified resulting a neuron. If it is not, a segmentation technique is applied to divide the data into smaller classes for new HB. For each subdivision new neurons are added to the network. The tests show a rapid method with high success rate.
18
Hayes, John A. "Phenotypic properties and intrinsic currents of neurons involved in the neural generation of mammalian breathing." W&M ScholarWorks, 2007. https://scholarworks.wm.edu/etd/1539623329.
Full textAbstract:
Breathing is essential for mammalian life. Although there is an emerging consensus that the inspiratory respiratory rhythm is generated in a lower brainstem region known as the preBotzinger Complex (preBotC), the mechanism of rhythmogenesis is still unclear. Additionally, the modulation of intrinsic currents within preBotC neurons has yet to be fully elucidated. This dissertation addresses both of these issues and relies on imaging, electrophysiological, and modeling techniques. The first chapter examines the size and composition of the preBotC. The chapter also decribes the means by which substance P (SP) excites the vast majority of preBotC neurons by illustrating the characteristics of the SP-activated current (/SP) in these neurons. In the subsequent chapter, we characterize a voltage-dependent potassium current that is involved in maintaining stable rhythms during normal fictive breathing. The third chapter presents a mathematical model of heterogeneous and rhythmogenic neurons that initiate network bursts. We show how this behavior relies on feedback synaptic connections within the network that reinforces activity, i.e., recurrent-excitation. We also compare model results to experimental data and make testable predictions. The final chapter elaborates on the discussion of /SP from the first chapter and presents evidence suggesting that a cyclic adenosine monophosphate (cAMP)-modulated non-specific cation channel may account for the depolarizing response in preBotC neurons from several neuromodulators. Altogether, this dissertation advances the field's understanding on several fronts. We have distinguished possible functional roles of neurons from electrophysiological characteristics, estimated the number of neurons necessary for rhythmogenesis, characterized /SP , and clarified the distribution of SP-sensitive receptors among inspiratory neurons. We have identified and characterized a voltage-dependent potassium currrent important for inspiratory activity and analyzed its role. We have also described in detail how rhythmic bursts form from recurrent excitation and how this relates to experimental data. Finally, we have identified and begun characterizing a potentially important and novel mechanism for the modulation of membrane potentials in critical inspiratory neurons.
19
Serene, Stephen Rothrock. "Generative probabilistic models of neuron morphology." Thesis, Massachusetts Institute of Technology, 2013. http://hdl.handle.net/1721.1/85494.
Full textAbstract:
Thesis: M. Eng., Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, 2013.Cataloged from PDF version of thesis.
Includes bibliographical references (page 40).
Thanks to automation in ultrathin sectioning and confocal and electron microscopy, it is now possible to image large populations of neurons at single-cell resolution. This imaging capability promises to create a new field of neural circuit microanatomy. Three goals of such a field would be to trace multi-cell neural networks, to classify neurons into morphological cell types, and to compare patterns and statistics of connectivity in large networks to meaningful null models. However, those goals raise significant computational challenges. In particular, since neural morphology spans six orders of magnitude in length (roughly 1 nm-1 mm), a spatial hierarchy of representations is needed to capture micron-scale morphological features in nanometer resolution images. For this thesis, I have built and characterized a system that learns such a representation as a Multivariate Hidden Markov Model over skeletonized neurons. I have developed and implemented a maximum likelihood method for learning an HMM over a directed, unrooted tree structure of arbitrary degree. In addition, I have developed and implemented a set of object-oriented data structures to support this HMM, and to produce a directed tree given a division of the leaf nodes into inputs and outputs. Furthermore, I have developed a set of features on which to train the HMM based only on information in the skeletonized neuron, and I have tested this system on a dataset consisting of confocal microscope images of 14 fluorescence-labeled mouse retinal ganglion cells. Additionally, I have developed a system to simulate neurons of varying difficulty for the HMM, and analyzed its performance on those neurons. Finally, I have explored whether the HMMs this system learns could successfully detect errors in simulated and, eventually, neural datasets.
by Stephen Rothrock Serene.
M. Eng.
20
Nord, Sofia. "Multivariate Time Series Data Generation using Generative Adversarial Networks : Generating Realistic Sensor Time Series Data of Vehicles with an Abnormal Behaviour using TimeGAN." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-302644.
Full textAbstract:
Large datasets are a crucial requirement to achieve high performance, accuracy, and generalisation for any machine learning task, such as prediction or anomaly detection, However, it is not uncommon for datasets to be small or imbalanced since gathering data can be difficult, time-consuming, and expensive. In the task of collecting vehicle sensor time series data, in particular when the vehicle has an abnormal behaviour, these struggles are present and may hinder the automotive industry in its development. Synthetic data generation has become a growing interest among researchers in several fields to handle the struggles with data gathering. Among the methods explored for generating data, generative adversarial networks (GANs) have become a popular approach due to their wide application domain and successful performance. This thesis focuses on generating multivariate time series data that are similar to vehicle sensor readings from the air pressures in the brake system of vehicles with an abnormal behaviour, meaning there is a leakage somewhere in the system. A novel GAN architecture called TimeGAN was trained to generate such data and was then evaluated using both qualitative and quantitative evaluation metrics. Two versions of this model were tested and compared. The results obtained proved that both models learnt the distribution and the underlying information within the features of the real data. The goal of the thesis was achieved and can become a foundation for future work in this field.När man applicerar en modell för att utföra en maskininlärningsuppgift, till exempel att förutsäga utfall eller upptäcka avvikelser, är det viktigt med stora dataset för att uppnå hög prestanda, noggrannhet och generalisering. Det är dock inte ovanligt att dataset är små eller obalanserade eftersom insamling av data kan vara svårt, tidskrävande och dyrt. När man vill samla tidsserier från sensorer på fordon är dessa problem närvarande och de kan hindra bilindustrin i dess utveckling. Generering av syntetisk data har blivit ett växande intresse bland forskare inom flera områden som ett sätt att hantera problemen med datainsamling. Bland de metoder som undersökts för att generera data har generative adversarial networks (GANs) blivit ett populärt tillvägagångssätt i forskningsvärlden på grund av dess breda applikationsdomän och dess framgångsrika resultat. Denna avhandling fokuserar på att generera flerdimensionell tidsseriedata som liknar fordonssensoravläsningar av lufttryck i bromssystemet av fordon med onormalt beteende, vilket innebär att det finns ett läckage i systemet. En ny GAN modell kallad TimeGAN tränades för att genera sådan data och utvärderades sedan både kvalitativt och kvantitativt. Två versioner av denna modell testades och jämfördes. De erhållna resultaten visade att båda modellerna lärde sig distributionen och den underliggande informationen inom de olika signalerna i den verkliga datan. Målet med denna avhandling uppnåddes och kan lägga grunden för framtida arbete inom detta område.
21
Williamson, Richard. "A new generation neural prosthesis." Thesis, National Library of Canada = Bibliothèque nationale du Canada, 1999. http://www.collectionscanada.ca/obj/s4/f2/dsk1/tape8/PQDD_0021/NQ46945.pdf.
Full text
22
Parodi, Livia. "Identification of genetic modifiers in Hereditary Spastic Paraplegias due to SPAST/SPG4 mutations Spastic paraplegia due to SPAST mutations is modified by the underlying mutation and sex Hereditary spastic paraplegia: More than an upper motor neuron disease." Thesis, Sorbonne université, 2019. http://www.theses.fr/2019SORUS317.
Full textAbstract:
Les Paraplégies Spastiques Héréditaires (PSHs) sont un groupe de maladies neurodégénératives rares qui surviennent suite à la dégénérescence progressive des voies corticospinales, entraînant une spasticité des membres inférieurs, signe distinctif de la pathologie. Elles se caractérisent par une extrême hétérogénéité qui concerne à la fois les facteurs génétiques et cliniques, ainsi que d’autres aspects de la maladie, tels que l’âge d’apparition et la sévérité des signes. Cette variabilité est typiquement observée chez les patients porteurs de mutations pathogènes dans SPAST, le gène le plus fréquemment muté dans les PSHs. Après avoir réuni une cohorte de 842 patients mutés dans SPAST, nous avons utilisé une combinaison de différentes approches de Séquençage de Nouvelle Génération (NGS) afin de mieux comprendre les causes de l’hétérogénéité observée chez les patients, afin d’identifier des facteurs génétiques responsables de variations de l’âge au début de la maladie. Les données résultantes du génotypage de l’ensemble du génome ont ainsi été utilisées pour effectuer des analyses d’association et de liaison qui, combinées aux données de séquençage de l’ARN, ont permis d’identifier différents variantes/gènes candidats, potentiellement impliqués comme facteurs modificateurs de l’âge de début des SPAST-PSHsHereditary Spastic Paraplegias (HSPs) are a group of rare, inherited, neurodegenerative disorders that arise following the progressive degeneration of the corticospinal tracts, leading to lower limbs spasticity, the disorder hallmark. HSPs are characterized by an extreme heterogeneity that encompasses both genetic and clinical features, extending to additional disorder’s features, such as age of onset and severity. This phenotypic variability is typically observed among HSP patients carrying pathogenic mutations in SPAST, the most frequently mutated HSP causative gene. After assembling a cohort of 842 SPAST-HSP patients, a combination of different Next Generation Sequencing approaches was used to dig deeper into the causes of the observed heterogeneity, especially focusing on the identification of age of onset genetic modifiers. Sequencing data resulting from Whole Genome Genotyping were used to perform both association and linkage analysis that, combined with RNA sequencing expression data, allowed to identify different candidate variants/genes, potentially acting as SPAST-HSP age of onset modifiers
23
Rončka, Martin. "Material Artefact Generation." Master's thesis, Vysoké učení technické v Brně. Fakulta informačních technologií, 2019. http://www.nusl.cz/ntk/nusl-399191.
Full textAbstract:
Ne vždy je jednoduché získání dostatečně velké a kvalitní datové sady s obrázky zřetelných artefaktů, ať už kvůli nedostatku ze strany zdroje dat nebo složitosti tvorby anotací. To platí například pro radiologii, nebo také strojírenství. Abychom mohli využít moderní uznávané metody strojového učení které se využívají pro klasifikaci, segmentaci a detekci defektů, je potřeba aby byla datová sada dostatečně velká a vyvážená. Pro malé datové sady čelíme problémům jako je přeučení a slabost dat, které způsobují nesprávnou klasifikaci na úkor málo reprezentovaných tříd. Tato práce se zabývá prozkoumáváním využití generativních sítí pro rozšíření a vyvážení datové sady o nové vygenerované obrázky. Za použití sítí typu Conditional Generative Adversarial Networks (CGAN) a heuristického generátoru anotací jsme schopni generovat velké množství nových snímků součástek s defekty. Pro experimenty s generováním byla použita datová sada závitů. Dále byly použity dvě další datové sady keramiky a snímků z MRI (BraTS). Nad těmito dvěma datovými sadami je provedeno zhodnocení vlivu generovaných dat na učení a zhodnocení přínosu pro zlepšení klasifikace a segmentace.
24
Lai, Ka-Hang. "Neural network approaches to caricature generation." Thesis, Loughborough University, 2007. https://dspace.lboro.ac.uk/2134/34440.
Full textAbstract:
A caricature is defined as a humorous drawing of a human facial figure that makes some of its distinct features appear exaggerated. It is easily observed that the exaggerations made by different artists on facial components are often different and are non-linear. This uniqueness of the exaggerations signifies the drawing style of an artist, but has unfortunately been ignored in the design of existing computer based automatic caricature generation systems. Nevertheless learning the unique drawing style and modelling the non-linear exaggerations distinct to an artist provide the key but a real challenge to the computer based automatic generation of professional caricature. This Thesis proposes a face modelling framework that includes two novel face models, which are capable of representing human faces in caricaturing applications.
25
Vazin, Tandis. "Generation of Dopaminergic Neurons from Human Embryonic Stem Cells." Doctoral thesis, Stockholm : Bioteknologi, Kungliga Tekniska högskolan, 2008. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-9591.
Full text
26
Shiraishi, Atsushi. "Generation of thalamic neurons from mouse embryonic stem cells." Kyoto University, 2018. http://hdl.handle.net/2433/230993.
Full text
27
Guzulaitis, Robertas. "The organisation principles of spinal neural network: temporal integration of somatosensory input and distribution of network activity." Doctoral thesis, Lithuanian Academic Libraries Network (LABT), 2013. http://vddb.laba.lt/obj/LT-eLABa-0001:E.02~2013~D_20130925_093153-76748.
Full textAbstract:
Spinal cord integrates somatosensory information and generates coordinated motor responses. Temporal integration can be used for discrimination of important stimuli from noise. Here it is shown that temporal integration of somatosensory inputs in sub second time scale is possible without changes of intrinsic properties of motoneurons. The activity of premotor neurons increases during temporal integration and can be a mechanism for short term information storage in spinal cord. Suppression of motor activity after painful somatosensory stimulus is called cutaneous silent period. This motor suppression is well described in humans and used for diagnostics. However it is not known if the suppression of motor activity is due to inhibition of motoneurons or reduction of excitatory drive from premotor neurons. Here it is shown that motoneurons are inhibited during cutaneous silent period. Neural networks of spinal cord not only process somatosensory information but generate locomotion and reflexes too. It is accepted that neural networks controlling front and hind limb movements are located in cervical and lumbar enlargements respectfully. Here it is shown that thoracic segments of spinal cord contribute to hind limb movements as well. It means that neural network generating movements is much more widely distributed than previously thought.Nugaros smegenys gauna somatosensorinę informaciją, ją integruoja ir generuoja motorinius atsakus. Disertacijoje parodoma, kad somatosensorinių įėjimų viršsekundinė laikinė integracija nugaros smegenų neuronų tinkle vyksta ne dėl motorinių neuronų vidinių savybių kitimo. Laikinės integracijos metu padidėja priešmotorinių neuronų aktyvumas ir tai gali lemti informacijos apie somatosensorinį įėjimą saugojimą. Somatosensorinis tylos periodas – tai motorinio aktyvumo slopinimas po skausmingo stimulo. Jis plačiai aprašytas žmonėse, bei taikomas diagnostikoje. Nepaisant plataus taikymo, somatosensorinio tylos periodo mechanizmai nėra ištirti – nebuvo žinoma ar šis motorinio aktyvumo slopinimas vyksta slopinant motorinius neuronus, ar eliminuojant motorinių neuronų žadinimą. Disertacijoje parodoma, kad somatosensorinio tylos periodo metu motoriniai neuronai yra slopinami. Be somatosensorinės informacijos apdorojimo nugaros smegenų neuronų tinklai užtikrina judėjimo ir refleksų valdymą. Yra priimta, kad priekines ir užpakalines galūnes valdantys neuronų tinklai išsidėstę atitinkamai nugaros smegenų kaklinės ir strėnų sričių išplatėjimuose. Disertacijoje parodoma, kad ir krūtininiai nugaros smegenų segmentai prisideda prie užpakalinių galūnių motorinio aktyvumo generavimo. Tai leidžia manyti, kad neuronų tinklas generuojantis judesius yra išplitęs labiau, nei manyta iki šiol.
28
Guzulaitis, Robertas. "Nugaros smegenų neuronų tinklo veikimo principai: somatosensorinės informacijos integracija ir aktyvumo išplitimas." Doctoral thesis, Lithuanian Academic Libraries Network (LABT), 2013. http://vddb.laba.lt/obj/LT-eLABa-0001:E.02~2013~D_20130925_093406-59707.
Full textAbstract:
Nugaros smegenys gauna somatosensorinę informaciją, ją integruoja ir generuoja motorinius atsakus. Disertacijoje parodoma, kad somatosensorinių įėjimų viršsekundinė laikinė integracija nugaros smegenų neuronų tinkle vyksta ne dėl motorinių neuronų vidinių savybių kitimo. Laikinės integracijos metu padidėja priešmotorinių neuronų aktyvumas ir tai gali lemti informacijos apie somatosensorinį įėjimą saugojimą. Somatosensorinis tylos periodas – tai motorinio aktyvumo slopinimas po skausmingo stimulo. Jis plačiai aprašytas žmonėse, bei taikomas diagnostikoje. Nepaisant plataus taikymo, somatosensorinio tylos periodo mechanizmai nėra ištirti – nebuvo žinoma ar šis motorinio aktyvumo slopinimas vyksta slopinant motorinius neuronus, ar eliminuojant motorinių neuronų žadinimą. Disertacijoje parodoma, kad somatosensorinio tylos periodo metu motoriniai neuronai yra slopinami. Be somatosensorinės informacijos apdorojimo nugaros smegenų neuronų tinklai užtikrina judėjimo ir refleksų valdymą. Yra priimta, kad priekines ir užpakalines galūnes valdantys neuronų tinklai išsidėstę atitinkamai nugaros smegenų kaklinės ir strėnų sričių išplatėjimuose. Disertacijoje parodoma, kad ir krūtininiai nugaros smegenų segmentai prisideda prie užpakalinių galūnių motorinio aktyvumo generavimo. Tai leidžia manyti, kad neuronų tinklas generuojantis judesius yra išplitęs labiau, nei manyta iki šiol.Spinal cord integrates somatosensory information and generates coordinated motor responses. Temporal integration can be used for discrimination of important stimuli from noise. Here it is shown that temporal integration of somatosensory inputs in sub second time scale is possible without changes of intrinsic properties of motoneurons. The activity of premotor neurons increases during temporal integration and can be a mechanism for short term information storage in spinal cord. Suppression of motor activity after painful somatosensory stimulus is called cutaneous silent period. This motor suppression is well described in humans and used for diagnostics. However it is not known if the suppression of motor activity is due to inhibition of motoneurons or reduction of excitatory drive from premotor neurons. Here it is shown that motoneurons are inhibited during cutaneous silent period. Neural networks of spinal cord not only process somatosensory information but generate locomotion and reflexes too. It is accepted that neural networks controlling front and hind limb movements are located in cervical and lumbar enlargements respectfully. Here it is shown that thoracic segments of spinal cord contribute to hind limb movements as well. It means that neural network generating movements is much more widely distributed than previously thought.
29
Chuong, Amy (Amy S. ). "Development of next-generation optical neural silencers." Thesis, Massachusetts Institute of Technology, 2011. http://hdl.handle.net/1721.1/69521.
Full textAbstract:
Thesis (S.M.)--Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2011.Cataloged from PDF version of thesis.
Includes bibliographical references (p. 65-74).
The ability to rapidly and safely silence the electrical activity of individual neurons or neuron populations is invaluable in the study of brain circuit mapping. The expression of light-driven ion channels and pumps allows these pathways to be observed, mapped and controlled with millisecond timescale resolution. We here show that it is possible to mediate the powerful multiple-color silencing of neural activity through the heterologous expression of light-driven outward proton pumps and inward chloride pumps. We characterized a number of novel opsins through an exploration of ecological and genomic diversity, and further boosted opsin function and trafficking through the appendage of signal sequences. The green-light drivable archaerhodopsin-3 (Arch) from Halorubrum sodomense and the yellow-light drivable archaerhodopsin from Halorubrum strain TP009 (ArchT) are able to mediate complete neuron silencing in the in vivo awake mouse brain, and the blue-light drivable proton pump from Leptosphaeria maculans (Mac) opens up the potential for the multiple-color control of independent neuron populations. Finally, the principles outlined here can be extrapolated to the larger context of synthetic physiology.
by Amy Chuong.
S.M.
30
Noakes, Zoe. "Generation of functional striatal neurons from human pluripotent stem cells." Thesis, Cardiff University, 2016. http://orca.cf.ac.uk/98932/.
Full textAbstract:
The striatal neuronal populations comprise medium spiny projection neurons (MSNs) and GABAergic and cholinergic interneurons. Huntington’s disease (HD) involves massive degeneration of striatal neurons. The derivation of MSNs and interneurons from human pluripotent stem cells (hPSCs) would allow modelling of striatal function and HD in vitro, as well as provide a viable source of tissue for cell replacement therapy. Our lab has previously demonstrated that Activin A can induce MSN fate in hPSCs, and that these cells can survive and differentiate in vitro and in vivo. In this study, it was found that this effect occurs via the Activin receptor, independently of SHH signalling. Furthermore, blockade of BMP signalling accelerated MSN differentiation. Electrophysiological analysis demonstrated their potential to acquire functional membrane properties and synaptic activity in vitro. Wnt inhibition and SHH activation have been shown to pattern hPSCs into medial ganglionic eminence (MGE) progenitors and cortical interneurons. Both cortical and striatal interneurons are born in the MGE. This thesis presents the first account of generating MGE progenitors for the purpose of producing striatal interneurons in vitro. They expressed subtype markers such as parvalbumin, somatostatin, calretinin and choline acetyltransferase. When transplanted into neonatal rat striatum, hPSC-derived MGE progenitors migrated to the septum and hippocampus within 6 weeks. The majority of differentiated neurons became calretinin GABAergic interneurons, and a few in the striatum acquired cholinergic interneuron fate. Patch clamp analysis both in vitro and in vivo revealed functional neuronal characteristics and synaptic connectivity, although a more mature neuronal phenotype was achieved in vivo. In conclusion, functional striatal MSNs and interneurons can be generated using hPSCs, which will be invaluable for research into striatal function and dysfunction in HD and other striatum relevant disorders. They may also serve as a desperately needed therapy for HD, pending further preclinical studies in HD animal models.
31
Arber, Charles. "Generation of disease-relevant neurons from human pluripotent stem cells." Thesis, Imperial College London, 2013. http://hdl.handle.net/10044/1/14670.
Full textAbstract:
This thesis describes investigations into exogenous factors that influence fate-choices during human pluripotent stem cell (PSC) differentiation in vitro. I describe novel growth factor environments and small molecule regimes that provide patterning signals enabling directed differentiation of human PSCs towards biomedically relevant neurons. I provide evidence that the TGFβ growth factor Activin can promote ventral telencephalic differentiation. Small adjustments in Activin administration can produce an over-representation of forebrain derived medium spiny neurons and cortical interneurons, with significance in Huntington’s disease and epilepsy respectively. Additionally, I manipulate the earliest growth factor environments of PSC differentiation to lead to broad changes in rostro-caudal patterning. Inhibiting FGF signalling leads to a midbrain-like phenotype that can be further differentiated towards a ventral midbrain dopaminergic fate, a cell type that degenerates in Parkinson’s disease. Our novel manipulations and differentiation protocols provide insights into early events in human development, which would otherwise be impossible to study. This logic can also be applied to investigate diseased states during human development and ageing via use of disease-specific cell lines. The mature neurons produced may provide a tool that can be applied to large-scale drug screening assays and toxicology testing as well as having the potential for cell based therapies in future years.
32
Patani, Rickie. "Generating motor neuron subtype diversity from human pluripotent stem cells." Thesis, University of Cambridge, 2012. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.610349.
Full text
33
Lewis, John E. "Dynamics of neural networks and respiratory rhythm generation." Thesis, McGill University, 1991. http://digitool.Library.McGill.CA:80/R/?func=dbin-jump-full&object_id=60568.
Full textAbstract:
The phase resetting effects of stimulating the superior laryngeal nerve at different phases of the respiratory cycle in cats were measured in terms of the latency of onset of the cycle following stimulation. Fixed-delay stimulation was also used; for certain combinations of delay, stimulus intensity, and cycles between stimuli, it resulted in (1) a variable, rather than consistent, response, and (2) a transient increase in cycle duration during and after stimulation. Phase resetting and fixed-delay stimulation of a simple three-phase model for neural rhythm generation produce responses that are qualitatively similar to those obtained experimentally.We consider the dynamical properties of a class of theoretical models of neural networks that have the same mathematical formulation as the above three-phase model, but consist of a larger number of randomly connected elements. A simple transformation of these models shows correspondence with previous neural network models and enables a theoretical analysis of steady states and cycles. Complex aperiodic dynamics are found in networks consisting of 6 or more elements.
34
Dowrick, Thomas Martin. "Biologically motivated circuits for third generation neural networks." Thesis, University of Liverpool, 2011. http://livrepository.liverpool.ac.uk/3024754/.
Full text
35
Wen, Tsung-Hsien. "Recurrent neural network language generation for dialogue systems." Thesis, University of Cambridge, 2018. https://www.repository.cam.ac.uk/handle/1810/275648.
Full textAbstract:
Language is the principal medium for ideas, while dialogue is the most natural and effective way for humans to interact with and access information from machines. Natural language generation (NLG) is a critical component of spoken dialogue and it has a significant impact on usability and perceived quality. Many commonly used NLG systems employ rules and heuristics, which tend to generate inflexible and stylised responses without the natural variation of human language. However, the frequent repetition of identical output forms can quickly make dialogue become tedious for most real-world users. Additionally, these rules and heuristics are not scalable and hence not trivially extensible to other domains or languages. A statistical approach to language generation can learn language decisions directly from data without relying on hand-coded rules or heuristics, which brings scalability and flexibility to NLG. Statistical models also provide an opportunity to learn in-domain human colloquialisms and cross-domain model adaptations. A robust and quasi-supervised NLG model is proposed in this thesis. The model leverages a Recurrent Neural Network (RNN)-based surface realiser and a gating mechanism applied to input semantics. The model is motivated by the Long-Short Term Memory (LSTM) network. The RNN-based surface realiser and gating mechanism use a neural network to learn end-to-end language generation decisions from input dialogue act and sentence pairs; it also integrates sentence planning and surface realisation into a single optimisation problem. The single optimisation not only bypasses the costly intermediate linguistic annotations but also generates more natural and human-like responses. Furthermore, a domain adaptation study shows that the proposed model can be readily adapted and extended to new dialogue domains via a proposed recipe. Continuing the success of end-to-end learning, the second part of the thesis speculates on building an end-to-end dialogue system by framing it as a conditional generation problem. The proposed model encapsulates a belief tracker with a minimal state representation and a generator that takes the dialogue context to produce responses. These features suggest comprehension and fast learning. The proposed model is capable of understanding requests and accomplishing tasks after training on only a few hundred human-human dialogues. A complementary Wizard-of-Oz data collection method is also introduced to facilitate the collection of human-human conversations from online workers. The results demonstrate that the proposed model can talk to human judges naturally, without any difficulty, for a sample application domain. In addition, the results also suggest that the introduction of a stochastic latent variable can help the system model intrinsic variation in communicative intention much better.
36
Jasieczek, Christina Bozena. "Investigation of hydrogen bonding and SHG activity of organic salts and co-crystals." Thesis, University of Sussex, 1996. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.318504.
Full text
37
Kalainathan, Diviyan. "Generative Neural Networks to infer Causal Mechanisms : algorithms and applications." Thesis, Université Paris-Saclay (ComUE), 2019. http://www.theses.fr/2019SACLS516.
Full textAbstract:
La découverte de relations causales est primordiale pour la planification, le raisonnement et la décision basée sur des données d'observations ; confondre corrélation et causalité ici peut mener à des conséquences indésirables. La référence pour la découverte de relations causales est d'effectuer des expériences contrôlées. Mais dans la majorité des cas, ces expériences sont coûteuses, immorales ou même impossible à réaliser. Dans ces cas, il est nécessaire d'effectuer la découverte causale seulement sur des données d'observations. Dans ce contexte de causalité observationnelle, retrouver des relations causales introduit traditionellement des hypothèses considérables sur les données et sur le modèle causal sous-jacent. Cette thèse vise à relaxer certaines de ces hypothèses en exploitant à la fois la modularité et l'expressivité des réseaux de neurones pour la causalité, en exploitant à la fois et indépendences conditionnelles et la simplicité des méchanismes causaux, à travers deux algorithmes. Des expériences extensives sur des données simulées et sur des données réelles ainsi qu'une analyse théorique approfondie prouvent la cohérence et bonne performance des approches proposéesCausal 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
38
Bahreininejad, Ardeshir. "Artificial neural networks for parallel finite element computations." Thesis, Heriot-Watt University, 1996. http://hdl.handle.net/10399/707.
Full text
39
Hatami, Maryam [Verfasser], and Thomas [Akademischer Betreuer] Skutella. "Combination of Prox1/NeuroD1 Transcription Factor Overexpression Boosts Generation of Dentate Gyrus Granule Neurons from Pluripotent Stem Cells / Maryam Hatami ; Betreuer: Thomas Skutella." Heidelberg : Universitätsbibliothek Heidelberg, 2018. http://d-nb.info/1177386011/34.
Full text
40
Casas, Manzanares Noé. "Injection of linguistic knowledge into neural text generation models." Doctoral thesis, Universitat Politècnica de Catalunya, 2020. http://hdl.handle.net/10803/671045.
Full textAbstract:
Language is an organic construct. It emanates from the need for communication and changes through time, influenced by multiple factors. The resulting language structures are a mix of regular syntactic and morphological constructions together with divergent irregular elements. Linguistics aims at formalizing these structures, providing a rationalization of the underlying phenomena. However, linguistic information alone is not enough to fully characterize the structures in language, as they are intrinsically tied to meaning, which constrains and modulates the applicability of the linguistic phenomena and also to context and domain.
Classical machine translation approaches, like rule-based systems, relied completely on the linguistic formalisms. Hundreds of morphological and grammatical rules were wired together to analyze input text and translate it into the target language, trying to take into account the semantic load carried by it. While this kind of processing can satisfactorily address most of the low-level language structures, many of the meaning-dependent structures failed to be analyzed correctly.
On the other hand, the dominant neural language processing systems are trained from raw textual data, handling it as a sequence of discrete tokens. These discrete tokens are normally defined looking for reusable word pieces identified statistically from data. In the whole training process, there is no explicit notion of linguistic knowledge: no morphemes, no morphological information, no relationships among words, or hierarchical groupings.This thesis aims at bridging the gap between the neural systems and linguistics-based systems, devising systems that have the flexibility and good results of the former with a base on the linguistic formalisms, with the purposes of improving quality where data alone cannot and forcing human-understandable working dynamics into the otherwise black-box neural systems. For this, we propose techniques to fuse statistical subwords with word-level linguistic information, to remove subwords altogether and rely solely on lemmas and morphological traits of the words, and to drive the text generation process on the ordering defined by syntactic dependencie. The main results of the proposed methods are the improvements in translation quality that can be obtained by injecting morphological information into NMT systems when testing on out-of-domain data for morphologically-rich languages, and the control over the generated text that can be gained by means of linking the generation order to the syntactic structure.El lenguaje es una construcción orgánica que surge de la necesidad de comunicación, y que cambia a lo largo del tiempo, influenciado por múltiples factores, resultando en estructuras del lenguaje donde se mezclan construcciones morfológicas y sintácticas regulares con otros elementos irregulares. La lingüística tiene como objetivo el formalizar estas estructuras, proponiendo interpretaciones de los fenómenos subyacentes. Sin embargo, la lingüística no es suficiente para caracterizar de manera completa las estructuras del lenguaje, ya que éstas se encuentran intrínsicamente ligadas tanto al significado -al restringir y modular éste la aplicabilidad de los fenómenos lingüísticos- como al contexto y al dominio. Las técnicas de traducción automática clásicas empleadas por los sistemas basados en reglas, se basan en formalismos lingüísticos, haciendo uso de miles de reglas morfológicas y gramaticales para analizar texto del idioma de origen y traducirlo al idioma de destino, intentando mantener la carga semántica original. Aunque este tipo de traducción procesa adecuadamente la estructuras de bajo nivel del lenguaje, muchas estructuras dependientes del significado no son analizadas correctamente. Los sistemas de procesado del lenguaje natural dominantes, en cambio, se entrenan usando texto como datos de entrada. Dicho texto se procesa como una secuencia de elementos discretos, normalmente definidos como trozos de palabras o sub-palabras, que se agrupan en una estructura de diccionario que es confecccionado estadísticamente de modo que se maximice el reuso de sus sub-palabras al codificar el texto de entrenamiento. En todo este proceso, no hay ninguna noción explícita de conocimiento lingüístico, ni morfemas, ni información morfológica, ni relaciones sintácticas entre palabras o grupos jerárquicos. El objetivo de esta tesis es hibridizar los sistemas neuronales y los sistemas basados en reglas lingüísticas, de manera que el resultado pueda mostrar la flexibilidad y buenos resultados de los primeros, pero teniendo una base lingüística que le permita tanto mejorar la calidad del texto generado en los casos en los que simplemente más datos no lo consiguen, como establer unas dinámicas de funcionamiento internas que sean entendibles por humanos, a diferencia de la naturaleza de "caja negra" de los sistemas neuronales normales. Para ello, se proponen técnicas para enriqueces las sub-palabras con información lingüística de nivel de palabra, ténicas para prescindir de las sub-palabras y basarse únicamente en el lema y los rasgos lingüísticos de las palabras, y técnicas para dirigir el orden de generación de texto mediante dependencias sintácticas. Los principales resultados de los métodos propuestos son la mejora en la calidad de traducción en sistemas neuronales a los que les inyectamos información lingüística, especialmente en escenarios de lenguas morfológicamente ricas con texto de distinto dominio, y el control directo del proceso de generación al ligarlo a las estructuras sintácticas del texto.
41
Ellerbrock, Thomas M. "Multilayer neural networks learnability, network generation, and network simplification /." [S.l. : s.n.], 1999. http://deposit.ddb.de/cgi-bin/dokserv?idn=958467897.
Full text
42
Vougiouklis, Pavlos. "Neural generation of textual summaries from knowledge base triples." Thesis, University of Southampton, 2019. https://eprints.soton.ac.uk/428045/.
Full textAbstract:
Most people need textual or visual interfaces in order to make sense of Semantic Web data. In this thesis, we investigate the problem of generating natural language summaries for structured data encoded as triples using neural networks. We propose an end-to-end trainable architecture that encodes the information from a set of triples into a vector of fixed dimensionality and generates a textual summary by conditioning the output on this encoded vector. In order to both train and evaluate the performance of our approach, we explore different methodologies for building the required data-to-text corpora. We initially focus our attention on the generation of biographies. Using methods for both automatic and human evaluation, we demonstrated that our technique is capable of scaling to domains with challenging vocabulary sizes of over 400k words. Given the promising results of our approach in biographies, we explore its applicability in the generation of open-domain Wikipedia summaries in two under-resourced languages, Arabic and Esperanto. We propose an adaptation of our original encoder-decoder architecture that outperforms a set of strong baselines of different nature. Furthermore, we conducted a set of community studies in order to measure the usability of the generated content by Wikipedia readers and editors. The targeted communities ranked our generated text close to the expected standards of Wikipedia. In addition, we found that the editors are likely to reuse a large portion of the generated summaries, thus, emphasizing the usefulness of our approach to the involved communities. Finally, we extend the original model with a pointer mechanism that enables it to jointly learn to verbalise in a different number of ways the content from the triples while retaining the ability to generate regular words from a fixed target vocabulary. We evaluate performance with a dataset encompassing the entirety of English Wikipedia. Results from both automatic and human evaluation highlight the superiority of the latter approach compared to our original encoder-decoder architecture and a set of competitive baselines.
43
Zhang, Xingxing. "Natural language generation as neural sequence learning and beyond." Thesis, University of Edinburgh, 2017. http://hdl.handle.net/1842/28930.
Full textAbstract:
Natural Language Generation (NLG) is the task of generating natural language (e.g., English sentences) from machine readable input. In the past few years, deep neural networks have received great attention from the natural language processing community due to impressive performance across different tasks. This thesis addresses NLG problems with deep neural networks from two different modeling views. Under the first view, natural language sentences are modelled as sequences of words, which greatly simplifies their representation and allows us to apply classic sequence modelling neural networks (i.e., recurrent neural networks) to various NLG tasks. Under the second view, natural language sentences are modelled as dependency trees, which are more expressive and allow to capture linguistic generalisations leading to neural models which operate on tree structures. Specifically, this thesis develops several novel neural models for natural language generation. Contrary to many existing models which aim to generate a single sentence, we propose a novel hierarchical recurrent neural network architecture to represent and generate multiple sentences. Beyond the hierarchical recurrent structure, we also propose a means to model context dynamically during generation. We apply this model to the task of Chinese poetry generation and show that it outperforms competitive poetry generation systems. Neural based natural language generation models usually work well when there is a lot of training data. When the training data is not sufficient, prior knowledge for the task at hand becomes very important. To this end, we propose a deep reinforcement learning framework to inject prior knowledge into neural based NLG models and apply it to sentence simplification. Experimental results show promising performance using our reinforcement learning framework. Both poetry generation and sentence simplification are tackled with models following the sequence learning view, where sentences are treated as word sequences. In this thesis, we also explore how to generate natural language sentences as tree structures. We propose a neural model, which combines the advantages of syntactic structure and recurrent neural networks. More concretely, our model defines the probability of a sentence by estimating the generation probability of its dependency tree. At each time step, a node is generated based on the representation of the generated subtree. We show experimentally that this model achieves good performance in language modeling and can also generate dependency trees.
44
Khan, Muhammad Jazib. "Programmable Address Generation Unit for Deep Neural Network Accelerators." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-271884.
Full textAbstract:
The Convolutional Neural Networks are getting more and more popular due to their applications in revolutionary technologies like Autonomous Driving, Biomedical Imaging, and Natural Language Processing. With this increase in adoption, the complexity of underlying algorithms is also increasing. This trend entails implications for the computation platforms as well, i.e. GPUs, FPGA, or ASIC based accelerators, especially for the Address Generation Unit (AGU), which is responsible for the memory access. Existing accelerators typically have Parametrizable Datapath AGUs, which have minimal adaptability towards evolution in algorithms. Hence new hardware is required for new algorithms, which is a very inefficient approach in terms of time, resources, and reusability. In this research, six algorithms with different implications for hardware are evaluated for address generation, and a fully Programmable AGU (PAGU) is presented, which can adapt to these algorithms. These algorithms are Standard, Strided, Dilated, Upsampled and Padded convolution, and MaxPooling. The proposed AGU architecture is a Very Long Instruction Word based Application Specific Instruction Processor which has specialized components like hardware counters and zero-overhead loops and a powerful Instruction Set Architecture (ISA), which can model static and dynamic constraints and affine and non-affine Address Equations. The target has been to minimize the flexibility vs. area, power, and performance trade-off. For a working test network of Semantic Segmentation, results have shown that PAGU shows close to the ideal performance, one cycle per address, for all the algorithms under consideration excepts Upsampled Convolution for which it is 1.7 cycles per address. The area of PAGU is approx. 4.6 times larger than the Parametrizable Datapath approach, which is still reasonable considering the high flexibility benefits. The potential of PAGU is not just limited to neural network applications but also in more general digital signal processing areas, which can be explored in the future.Convolutional Neural Networks blir mer och mer populära på grund av deras applikationer inom revolutionerande tekniker som autonom körning, biomedicinsk bildbehandling och naturligt språkbearbetning. Med denna ökning av antagandet ökar också komplexiteten hos underliggande algoritmer. Detta medför implikationer för beräkningsplattformarna såväl som GPU: er, FPGAeller ASIC-baserade acceleratorer, särskilt för Adressgenerationsenheten (AGU) som är ansvarig för minnesåtkomst. Befintliga acceleratorer har normalt Parametrizable Datapath AGU: er som har mycket begränsad anpassningsförmåga till utveckling i algoritmer. Därför krävs ny hårdvara för nya algoritmer, vilket är en mycket ineffektiv metod när det gäller tid, resurser och återanvändbarhet. I denna forskning utvärderas sex algoritmer med olika implikationer för hårdvara för adressgenerering och en helt programmerbar AGU (PAGU) presenteras som kan anpassa sig till dessa algoritmer. Dessa algoritmer är Standard, Strided, Dilated, Upsampled och Padded convolution och MaxPooling. Den föreslagna AGU-arkitekturen är en Very Long Instruction Word-baserad applikationsspecifik instruktionsprocessor som har specialiserade komponenter som hårdvara räknare och noll-overhead-slingor och en kraftfull Instruktionsuppsättning Arkitektur (ISA) som kan modellera statiska och dynamiska begränsningar och affinera och icke-affinerad adress ekvationer. Målet har varit att minimera flexibiliteten kontra avvägning av område, kraft och prestanda. För ett fungerande testnätverk av semantisk segmentering har resultaten visat att PAGU visar nära den perfekta prestanda, 1 cykel per adress, för alla algoritmer som beaktas undantar Upsampled Convolution för vilken det är 1,7 cykler per adress. Området för PAGU är ungefär 4,6 gånger större än Parametrizable Datapath-metoden, vilket fortfarande är rimligt med tanke på de stora flexibilitetsfördelarna. Potentialen för PAGU är inte bara begränsad till neurala nätverksapplikationer utan också i mer allmänna digitala signalbehandlingsområden som kan utforskas i framtiden.
45
Friel, Ruairi Donal. "The generation of a herpes simplex virus vector to target motor neurons." Thesis, University of Glasgow, 2001. http://theses.gla.ac.uk/3960/.
Full textAbstract:
Herpes simplex virus (HSV) is a neurovirulent virus that in the course of natural infection of man predominantly infects sensory neurons. The aim of this project was to develop a safe, nonvirulent HSV, capable of expressing exogenous genes which altered the binding characteristics of the virus so that tropism was directed predominantly to motor nerves. It was envisaged that these viruses could then act as prototypes for gene therapy vectors targeted to the treatment of motor nerve diseases. To achieve this, two mutant viruses were created, RFa and RFb. These contained deletions of the main HSV glycoprotein involved in cellular binding (glycoprotein C). Gene fusions were created of truncated portions of gC (amino acids 377-511(RFa) and amino acids 477-511 (RFb)) to E. coli heat-labile enterotoxin B-subunit (LTB). The gene fusions were inserted in the RL1 gene thereby abolishing expression of the virulence factor ICP34.5. LTB is a ligand which binds to several gangliosides, including GM1 and GM2 which are motor neuron markers. It was hoped that by deletion of the main viral protein involved in adsorption to cells and replacing it with an LTB-containing fusion protein, the tropism of the mutant viruses could be altered to promote an increase in motor neuron infection. RFb was constructed. RFa constructed but could not be purified to homogeneity. This was thought to be due to poor adsorption/penetration or cell-to-cell spread, brought about by expression of the LTB fusion protein. RFb was analysed to determine the effect of expression of the novel LTB fusion protein within the context of the HSV genome. Western blot analysis using antibodies directed against LTB failed to detect expression of the LTB-gC fusion protein. In vitro replication studies showed that the RFb was non-virulent as demonstrated by its inability to replicate in growth arrested 3T6 cells, a phenotype characteristic of HSV which fails to produce ICP34.5. However no marked difference in virus replication kinetics was seen between RFb and wild type HSV (17+) on two motor neuron-like cell lines (NSC-19 and NSC-34).
46
Brown, Sarah. "Adult generation of dopaminergic neurons in a genetic model of Parkinson's disease." Thesis, University of Sheffield, 2018. http://etheses.whiterose.ac.uk/21269/.
Full text
47
Pace, Ryland Weed. "Mechanisms underlying inspiratory burst generation in preBotzinger complex neurons of neonatal mice." W&M ScholarWorks, 2008. https://scholarworks.wm.edu/etd/1539616802.
Full textAbstract:
Understanding how molecular and cellular events integrate into a physiological behavior is a major question in neuroscience. Breathing can be easily studied using rhythmically active in vitro models that provide experimental access to perform cellular- and synapse-level experiments. While it is widely accepted that breathing depends on a specific region of the brainstem dubbed the preBotzinger complex (preBotC), the mechanisms responsible for rhythm generation remain unclear. In Chapter 1, we examine the pacemaker hypothesis, which posits that pacemaker properties and/or the persistent sodium current (/NaP) are obligatory for rhythm generation. We found that neither pacemaker properties nor /NaP are essential for respiratory rhythm generation in preBotC neurons. Next, we began testing the validity of the group pacemaker hypothesis, which posits that the respiratory rhythm is an emergent network property that depends on recurrent excitation coupled to intrinsic membrane properties in all preBotC neurons. During the inspiration in vitro, all preBotC neurons exhibit 300-500 ms bursts of electrical activity characterized by action potentials superimposed on a 10-30 mV envelope of depolarization, dubbed the inspiratory drive potential. Chapters 2 and 3 examine how synaptic and intrinsic membrane properties integrate to form inspiratory drive potentials. In Chapter 2, we found that the calcium-activated non-specific cationic current (/CAN) is responsible for ∼70% of the inspiratory drive potential. /CAN activation depends on Ca2+ influx from inositol 1,4,5-trisphosphate (IP3Rs)-mediated intracellular Ca2+ release coupled to group I metabotropic glutamate receptors (mGluRs), voltage-gated Ca2+ channels (VGCCs) and possibly to a smaller extent NMDA receptors. Chapter 3 examines how AMPARs trigger inspiratory burst potential generation. We found that AMPAR-mediated depolarizations open VGCCs, which activate /CAN directly. Moreover, Ca2+ influx from VGCC was required to trigger IP3R-mediated intracellular Ca2+ release. In Chapter 4, we interpret respiratory frequency modulation within the context of the group pacemaker hypothesis. We show that blocking low-frequency AMPAR-mediated excitatory postsynaptic potentials (EPSPs) causes rhythm cessation, which suggests that low-frequency EPSPs are important for kindling the initial phase of recurrent excitation. Through a meta-analysis of previously published work, we argue that frequency modulation depends on the temporal summation of EPSPs and is largely independent of changes in interburst spiking. In conclusion, our findings suggest that respiratory rhythm generation and frequency modulation depends on the coupling of synaptic and intrinsic membrane properties, which is most consistent with the group pacemaker hypothesis.
48
Soofi, Wafa Ahmed. "Regulation of rhythmic activity in the stomatogastric ganglion of decapod crustaceans." Diss., Georgia Institute of Technology, 2014. http://hdl.handle.net/1853/53440.
Full textAbstract:
Neuronal networks produce reliable functional output throughout the lifespan of an animal despite ceaseless molecular turnover and a constantly changing environment. The cellular and molecular mechanisms underlying the ability of these networks to maintain functional stability remain poorly understood. Central pattern generating circuits produce a stable, predictable rhythm, making them ideal candidates for studying mechanisms of activity maintenance. By identifying and characterizing the regulators of activity in small neuronal circuits, we not only obtain a clearer understanding of how neural activity is generated, but also arm ourselves with knowledge that may eventually be used to improve medical care for patients whose normal nervous system activity has been disrupted through trauma or disease. We utilize the pattern-generating pyloric circuit in the crustacean stomatogastric nervous system to investigate the general scientific question: How are specific aspects of rhythmic activity regulated in a small neuronal network?
The first aim of this thesis poses this question in the context of a single neuron. We used a single-compartment model neuron database to investigate whether co-regulation of ionic conductances supports the maintenance of spike phase in rhythmically bursting “pacemaker” neurons. The second aim of the project extends the question to a network context. Through a combination of computational and electrophysiology studies, we investigated how the intrinsic membrane conductances of the pacemaker neuron influence its response to synaptic input within the framework of the Phase Resetting Curve (PRC). The third aim of the project further extends the question to a systems-level context. We examined how ambient temperatures affect the stability of the pyloric rhythm in the intact, behaving animal. The results of this work have furthered our understanding of the principles underlying the long-term stability of neuronal network function.
49
Stigeborn, Patrik. "Generating 3D-objects using neural networks." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2018. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-230668.
Full textAbstract:
Enabling a 2D- to 3D-reconstruction is an interesting future service for Mutate AB, where this thesis is conducted. Convolutional neural networks (CNNs) is examined in different aspects, in order to give a realistic perception of what this technology is capable of. The task conducted, is the creation of a CNN that can be used to predict how an object from a 2D image would look in 3D. The main areas that this CNN is optimized for are Quality, Speed, and Simplicity. Where Quality is the output resolution of the 3D object, Speed is measured by the number of seconds it takes to complete a reconstruction, and Simplicity is achieved by using machine learning (ML). Enabling this could potentially ease the creation of 3D games and make the development faster. The chosen solution is to use two CNNs. The first CNN is using convolution to extract features from an input image. The second CNN is using transpose convolution to create a prediction of how the object would look in 3D, from the features extracted by the first neural network. This thesis is using an empirical development approach to reach an optimal solution for the CNN structure and its hyperparameters. The 3D-reconstruction is inspired by a sculpting process, meaning that the reconstruction starts with a low resolution and improves it iteratively. The result shows that the quality gained from each iteration grows exponentially whilst the increased time grows a lot less. Thereof, the conclusion is that the trade-off between speed and quality is in our favor. However, when looking at commercializing this technology or deploy it in a professional environment, it is still too slow to generate high resolution output. Also, in this case, the CNN is fragile when there are a lot of unrecognized shapes in the input image.
50
Liu, Mengxin. "Generative Neural Network for Portfolio Optimization." Thesis, Mälardalens högskola, Akademin för utbildning, kultur och kommunikation, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:mdh:diva-53027.
Full textAbstract:
This thesis aims to overcome the drawbacks of traditional portfolio optimization by employing Generative Deep Neural Networks on real stock data. The proposed framework is capable of generating return data that have similar statistical characteristics as the original stock data. The result is acquired using Monte Carlo simulation method and presented in terms of individual risk. This method is tested on real Swedish stock market data. A practical example demonstrates how to optimize a portfolio based on the output of the proposed Generative Adversarial Networks.