To see the other types of publications on this topic, follow the link: Neubau.
Dissertations / Theses on the topic 'Neubau'
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 'Neubau.'
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
Spitzer, Denis Spitzer Kristina. "Neubau eines Mehrfamilienhauses." Kassel Univ. Bibliothek, 2007. https://kobra.bibliothek.uni-kassel.de/handle/urn:nbn:de:hebis:34-2007101919452.
Full text
2
Hotz, Christine. "Neubau Wärmebad Käferberg Zürich Diplomthema C /." Zürich : ETH, Eidgenössische Technische Hochschule Zürich, Departement Architektur, 2007. http://e-collection.ethbib.ethz.ch/show?type=dipl&nr=296.
Full text
3
Vogel, Michael. "Emil Adolf Rossmässler - Namensgeber für den Tharandter Neubau." Saechsische Landesbibliothek- Staats- und Universitaetsbibliothek Dresden, 2006. http://nbn-resolving.de/urn:nbn:de:swb:14-1160131672462-79925.
Full textAbstract:
Dem Tharandter Neubau Bibliothek/Mensa wurde am 19.10.2005 im Rahmen des Festaktes zur feierlichen Eröffnung (SLUB-Kurier 4/2005) der Gebäudename 'Rossmässler-Bau' durch den Rektor der TU Dresden, Magnifizenz Kokenge, verliehen....
4
Selle, Ricky. "Entwurf eines HBV-Deckensystems für Anwendungen im Neubau." Doctoral thesis, Universitätsbibliothek Leipzig, 2016. http://nbn-resolving.de/urn:nbn:de:bsz:15-qucosa-193755.
Full textAbstract:
Die Dissertationsschrift beschreibt die Entwicklung eines Deckensystems im HBV für Anwendungen im Neubau. Dabei werden sowohl die Mechanik für den eigentlichen Verbund zwischen den Baustoffen als auch der Verbundquerschnitt des Deckenbauteils mit seinen Materialien und Dimensionen entworfen. Die Ausgestaltung des Deckensystems mit den wichtigsten Details wird theoretisch und versuchstechnisch begründet. Die Ergebnisse bauakustischer Untersuchungen an verschiedenen Deckenbauteilen, die Resultate von Scherversuchen mit unterschiedlichen Ausprägungen der Verbinder und die eines Biegeversuchs mit einem Deckenelement in den originalen Dimensionen fließen in die Überlegungen ein. Ein wichtiges Ergebnis der Arbeit ist die Anwendung des Systems in einem Pilotprojekt. Für die Standsicherheitsnachweise dieses Deckensystems auf der normativen Grundlage des Eurocode 5 werden teilweise vom Regelwerk abweichende Ansätze begründet. Außerdem wird dieses System in wirtschaftlicher und ökologischer Hinsicht bewertet.
5
Worm, Wolfram, Clemens Deilmann, and Karin Gruhler. "Umnutzung versus Neubau im Dorf - Energetisch-stoffliche sowie kostenseitige Gegenüberstellung von Sanierung, Umnutzung und Neubau von Wohn- und Gewerbebauten im ländlichen Raum." Saechsische Landesbibliothek- Staats- und Universitaetsbibliothek Dresden, 2008. http://nbn-resolving.de/urn:nbn:de:bsz:14-ds-1212665303077-65126.
Full textAbstract:
Ländliche Bausubstanz bewahren oder besser Neubauen? Dieser kontrovers diskutierten Frage ging ein Forscherteam des Leibniz-Instituts für ökologische Raumentwicklung e.V. genauer nach. Im Auftrag und in Begleitung der Sächsischen Landesanstalt für Landwirtschaft wurden Sanierungs- und Umnutzungsszenarien unter stofflichen, energetischen und finanziellen Aspekten untersucht. Die Ergebnisse wurden denen eines modernen Einfamilienhauses gegenübergestellt. Die entstandene Studie zeigt, dass Umnutzungs- und Sanierungsvorhaben energetisch und finanziell mit vergleichbaren Neubauten konkurrieren können. Außerdem lassen sich durch Umnutzungen in erheblichem Umfang Baustoffe und Energien einsparen.
6
Golsch, Michael. "Zukunft Campusmitte. Die Zweigbibliothek Medizin der SLUB plant ihren Neubau." Saechsische Landesbibliothek- Staats- und Universitaetsbibliothek Dresden, 2009. http://nbn-resolving.de/urn:nbn:de:bsz:14-qucosa-25636.
Full textAbstract:
Zukunftsmodell Bibliothek? Man muss ganz sicher kein Prophet sein, um unserer Wissens- und Informationsgesellschaft auch in den nächsten Jahren eine dynamische Entwicklung vorherzusagen. Zumindest die Quantität des Faktenwissens wird mindestens ebenso rasch wie bisher zunehmen. Gleiches dürfte für die Menge und für die Vielfalt der Informationsangebote gelten, für die der technologische Fortschritt auch künftig neue elektronische Austauschformen stimulieren wird. Die bereits heute festzustellende Enträumlichung von Wissen und Informationen wird sich damit weiter fortsetzen und über veränderte Rezeptions- und Kommunikationsformen unser soziales Zusammenleben auch in Zukunft entscheidend prägen und beeinflussen – so wie wir dies seit dem „Phänomen Internet“ und in der „Digitalen Revolution“bereits täglich erleben.
7
Selle, Ricky [Verfasser], Burkhard [Akademischer Betreuer] Pahl, Werner [Gutachter] Schneider, and Klaus [Gutachter] Holschemacher. "Entwurf eines HBV-Deckensystems für Anwendungen im Neubau : Entwurf eines HBV-Deckensystems für Anwendungen im Neubau / Ricky Selle ; Gutachter: Werner Schneider, Klaus Holschemacher ; Betreuer: Burkhard Pahl." Leipzig : Universitätsbibliothek Leipzig, 2016. http://d-nb.info/124031549X/34.
Full text
8
Kortmann, Konstantin. "Abriss und Neubau oder Kernsanierung? eine empirische Untersuchung zur Nutzungsdauer von Wohngebäuden des 20. Jahrhunderts im Ruhrgebiet." Köln R. Müller, 2007. http://d-nb.info/990659798/04.
Full text
9
Vandrey, Petra. "Neubau des Umweltrechts? : Untersuchung der Vor- und Nachteile einer Kodifizierung des Umweltrechts unter Berücksichtigung entsprechender Erfahrungen im Ausland /." Berlin : Köster, 1995. http://www.gbv.de/dms/spk/sbb/recht/toc/277995213.pdf.
Full text
10
Reuther, Stefan. "Der Um- und Neubau des Schlosses Rochlitz von 1375 bis 1400: Parallelen zum Baugeschehen auf Burg Mildenstein in Leisnig." Sandstein Verlag, 2004. https://slub.qucosa.de/id/qucosa%3A38453.
Full textAbstract:
Im letzten Viertel des 14. Jahrhunderts und insbesondere in der Regierungszeit des Markgrafen Wilhelm I. wurden einige markmeißnische Residenzen tiefgreifend umgebaut und modernisiert. Auch die Burg Rochlitz erhielt mit dem Neubau zweier Westtürme sowie dem Fürsten- und Querhaus-Komplex ihr schlossähnliches Antlitz, welches das Erscheinungsbild bis heute entscheidend prägt.
11
Sewing, Silke. "Neues Zuhause für Musikalien." Saechsische Landesbibliothek- Staats- und Universitaetsbibliothek Dresden, 2010. http://nbn-resolving.de/urn:nbn:de:bsz:14-qucosa-61303.
Full textAbstract:
Im Dezember 2010 zieht das Deutsche Musikarchiv (DMA) von Berlin an den Standort der Deutschen Nationalbibliothek Leipzig. Mit dem DMA wird sich in Leipzig die zentrale nationale Sammlung von Musikalien und Tonträgern und das musikbibliografische Informationszentrum Deutschlands ansiedeln.
12
Jacobs, Stephanie. "Ein ehrwürdiges Haus der Bücher – neu aufgestellt." Saechsische Landesbibliothek- Staats- und Universitaetsbibliothek Dresden, 2009. http://nbn-resolving.de/urn:nbn:de:bsz:14-ds-1243948237000-93490.
Full textAbstract:
Vom Baustaub umwölkt, vom Kreischen der Sägen betäubt und von den Chancen einer Neupositionierung beflügelt feiert das Deutsche Buch- und Schriftmuseum der Deutschen Nationalbibliothek in diesem Jahr seinen 125. Geburtstag. 1884 als Deutsches Buchgewerbemuseum in Leipzig gegründet gehört es zu den weltweit ältesten und nach Umfang und Qualität der Bestände bedeutendsten Museen auf dem Gebiet der Buchkultur.
13
Steck, Sarah [Verfasser]. "Prä-post-Untersuchung des Effektes von Umzug und Neubau des BKH Kempten auf die Stationsatmosphäre in der Wahrnehmung von Patienten und Mitarbeitern / Sarah Steck." Ulm : Universität Ulm, 2020. http://d-nb.info/1236842057/34.
Full text
14
Boechat, Teske Luiza. "Hotel Heinrich Heine." Master's thesis, Hochschule für Technik und Wirtschaft Dresden, 2016. http://nbn-resolving.de/urn:nbn:de:bsz:520-qucosa-213231.
Full textAbstract:
Viele Hotels in Schierke stehen seit Ende des 19./Anfang des 20. Jahrhunderts leer. Um ein attraktives und besucherreiches Hotel zu entwerfen, besteht die Aufgabe darin, das unbewohnte Hotel „Heinrich Heine“ als Vier-Sterne-Hotel zu entwickeln. Für den Entwurf wurden zwei Analysen durchgeführt. In der ersten Analyse war eine Stadt zu wählen, die eine enge Bedeutung und Bezug zur Stadt Schierke hat, damit die touristische Infrastruktur analysiert werden kann. Das Ziel der zweiten Analyse bestand darin, ein Hotel zu finden, welches sich sehr gut in die bestehende Umgebung (entweder Alpen oder Bergort) einpasst. Außerdem wurde ein Hotel gesucht, welches sich an den Prinzipien der Nachhaltigkeit orientiert. Am Ende wurde der Grundgedanke und das Grundkonzept eines Passivhauses favorisiert. Vor Beginn des Entwurfs fand ein Workshop in Schierke statt, bei dem jede Gruppe Ideen für die Gestaltung eines Ortsentwicklungskonzepts (Masterplan) für die Stadt diskutierte und entwickelte. Aus dieser Idee heraus, ist es wichtig, das Hotelprofil und die Zielgruppe miteinander zu verbinden. Auf Grundlage der Analysen und des Masterplans für die Entwicklung der Stadt wurde ein Entwurf für das Hotel Heinrich Heine erarbeitet. Eckpunkte für die Entwicklung des Projekts waren ebenso die Analyse des Grundstücks, historische Aspekte des Hotels, die Funktionalität, Materialität, Fassade und Aspekte, wie Barrierefreiheit und Nachhaltigkeit.
15
Rabe, Roman. "Die neue Dresdner Zentralbibliothek wächst hinter der Fassade des Kulturpalastes." SLUB Dresden, 2016. https://slub.qucosa.de/id/qucosa%3A7784.
Full textAbstract:
Der folgende Bericht soll einen Einblick in den Planungsprozess verschaffen. Der Schwerpunkt liegt bei der Einrichtungsplanung. Für eine Präsentation des Gesamtkonzeptes ist ein BIS-Sonderheft im Mai 2017 vorgesehen.
16
Farah, Kamal. "ROC- and LTF- based estimates of neural- behavioral and neural-neural correlations." Thesis, McGill University, 2014. http://digitool.Library.McGill.CA:80/R/?func=dbin-jump-full&object_id=123145.
Full textAbstract:
Observed correlations between responses in visual cortex and perceptual performance help draw a functional link between neural activity and visually guided behavior. These correlations are commonly derived with ROC-based neural-behavioral covariances (referred to as choice or detect probability) using boxcar analysis windows. Although boxcar windows capture the covariation between neural activity and behavior during steady-state stimulus presentations, they are not optimized to capture these correlations during realistic time-varying visual inputs. In this thesis, we implemented a matched-filter technique, combined with cross-validation, to improve the estimation of ROC-based neural-behavioral covariance under dynamic stimulus conditions. We show that this approach maximizes the area under the ROC curve and converges to the true neural-behavioral covariance using a Poisson spiking model. We also demonstrate that the matched filter, combined with cross-validation, reveals the dynamics of the neural-behavioral covariations of individual MT neurons during the detection of a brief motion stimulus.Temporal correlations among responses in the visual cortex, on the other hand, have a substantial effect on determining the functional connectivity between neurons. In order to study the interactions of correlations in neural processing, we measured the linear transfer function between cortical areas V1 and V2 using data collected with multi-electrode extracellular recordings. Our aim was to study the effect that a single V1 action potential has on V2 neurons using linear systems identification. A linear transfer function (referred to as a kernel) is a useful metric for understanding functional connectivity between two cortical areas because it quantifies the temporal integration by V2 of V1 inputs. We used a multiple-input single-output regression model to estimate the pairwise V1-V2 kernels. Because of the large number of simultaneous V1 inputs used in our model, this multivariate analysis has the advantage of minimizing the bias in the kernels due to correlated activity between V1 neurons.We estimated 25,470 kernels from the blank and stimulus periods. Kernel quality was evaluated based on a signal-to-noise ratio (SNR) of the estimated kernel variance to the shuffled kernel variance. Putative good kernels (high SNR > 1) were extracted from the blank (4,665) and stimulus (2,542) conditions, and both were found to be exponential in shape with 18 and 27 ms time constants, respectively. Thus, V2 neurons tended to integrate V1 inputs over an exponentially decaying window, with a combined average time constant of 21 ms, that was independent of the occurrence of a visual stimulus. Although the dynamics of cortical circuitry likely contribute to our measured kernels, the integrative properties of single neurons appear to be a dominant component of the V1-V2 linear transfer function.Les corrélations entre le comportement du cortex visuel et les habiletés visuelles permettent de comprendre le lien fonctionnel entre l'activité des neurones et le comportement d'un sujet lors du visionnement d'un écran. Ces corrélations sont déduites à travers des courbes ROC représentant les covariances entre les comportements et les neurones (la probabilité de choix et de détection) et ce, en utilisant des filtres avec une fonction de transfert Sinc. Ces filtres capturent efficacement les covariances entre les activités neurologiques et les comportements durant des stimuli constants. Toutefois, il n'est pas optimal d'utiliser ces filtres pour capturer ces corrélations lors de situations réelles et variantes en fonction du temps. Dans cette thèse, nous présentons une technique de filtrage équivalent combinée avec une validation croisée pour améliorer l'estimation de la courbe ROC représentant les covariances entre les comportements et les neurones lors de conditions de stimulus dynamiques. Suite à l'utilisation d'un modèle de déchargement sous forme d'une distribution Poisson, nous démontrons que cette approche maximise l'aire sous la courbe ROC et converge vers la vraie covariance entre les comportements et les neurones. Nous démontrons également que le filtre équivalent combiné à une validation croisée permet de représenter la dynamique des covariances des comportements et des neurones de la région visuelle MT durant la détection de brefs stimuli. Les corrélations temporelles entre les réponses du cortex visuel ont un effet considérable sur la détermination des connections fonctionnelles entre les neurones. Pour être en mesure d'étudier les interactions des corrélations du fonctionnement des neurones, nous avons calculé les valeurs des fonctions de transfert linéaires entre les régions corticales V1 et V2 en utilisant des données collectées à l'aide d'électrodes extracellulaires multiples enregistrées. Le but derrière cette analyse est de comprendre l'effet de l'influx nerveux de V1 sur les neurones V2 en usant de systèmes d'identification linéaires. Une fonction de transfert linéaire (ou fonction « kernel ») est une métrique utile à comprendre la connexion entre deux régions du cortex puisqu'elle quantifie l'intégration temporelle des neurones V1 dans V2. Nous avons employé un modèle de régression à variables multiples pour estimer la combinaison entre les « kernels » V1-V2. Face à l'insertion simultanée d'une multitude de variables V1 dans notre modèle, cette analyse à variables multiple s'est avérée être adéquate; permettant de minimiser le biais dans les « kernels » causés par les activités corrélées entre les neurones V1.Nous avons estimé 25,470 « kernels » à l'état de repos (lorsqu'aucune image n'apparait sur l'écran) et à l'état de stimulus. La qualité a été évaluée dépendamment de la valeur du ratio du signal-bruit de la variance estimée du « kernel » sur la variance du « kernel » brouillé. Les « kernels » considérés de bonne qualité (ayant un ratio signal-bruit supérieur à 1) ont été extraits des données de l'état de repos (4,665) et celles des conditions de stimuli (2,542). Il s'est avéré que ces deux types de données suivent une forme exponentielle avec des constantes de temps égales à 18 et 27 ms respectivement. Donc la convolution des V1 avec le « kernel » est équivalente à V2. V2 à son tour tend à intégrer les variables V1 dans la fonction exponentielle décroissante dont la moyenne combinée est de 21 ms. Nous pouvons noter que la forme du « kernel » est indépendante de la manifestation du stimulus visuel. Bien que les dynamiques du circuit du cortex peuvent contribuer au « kernel » quantifié, les propriétés intégratives des neurones s'avèrent être des éléments importants de la fonction de transfert linéaire de V1-V2.
17
Chen, Prakoon. "The Neural Shell : a neural networks simulator." Connect to resource, 1989. http://rave.ohiolink.edu/etdc/view?acc%5Fnum=osu1228839518.
Full text
18
Sari, Yunita. "Neural and non-neural approaches to authorship attribution." Thesis, University of Sheffield, 2018. http://etheses.whiterose.ac.uk/21415/.
Full text
19
Khliobas. "NEURAL NETWORK." Thesis, Київ 2018, 2018. http://er.nau.edu.ua/handle/NAU/33752.
Full text
20
Distasio, Andrew. "Novel Regulators of Neural Crest and Neural Progenitor Survival." University of Cincinnati / OhioLINK, 2020. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1593170783550813.
Full text
21
Sandström, Paul A. "The neural effects of palytoxinThe neural effects of palytoxin." Thesis, Queensland University of Technology, 2000. https://eprints.qut.edu.au/36765/1/36765_Digitised%20Thesis.pdf.
Full textAbstract:
Palytoxin, the most potent known marine toxin, has been demonstrated to produce acute neurological disturbances in several animal species and, possibly, even in humans. However, the effects of palytoxin on excitable membranes have not been well characterised or explained.
Palytoxin occurs within the same ecosystem in which other seafood toxins such as ciguatoxin, saxitoxin and brevetoxins are found. Also, several reports suggest that, occasionally, palytoxin may contribute to the ciguatera syndrome. As these other toxins are known to affect the sodium ion channel, either by activation or inactivation, evidence that palytoxin may also affect this structure was sought.
The studies performed on the ventral coccygeal nerve of the rat tail, subsequent to an intraperitoneal injection of palytoxin, demonstrated a significant slowing of mixed nerve and motor conduction velocities and a reduction in the amplitude of the mixed nerve and motor action potentials. However, palytoxin exerted no effect on the latency of the shortest F-wave response. In the palytoxin-treated group, the absolute and relative refractory periods were prolonged. Palytoxin also appeared to induce a significant prolongation of the supernormal period of nerve excitability. The response of the palytoxin-treated ventral coccygeal nerve to repetitive stimulation demonstrated no consistent abnormality.
These studies lend indirect support to the proposition that at least one action of palytoxin is that of an alteration in the excitability of neural tissue by inducing persistent sodium ion channel activation.
Lignocaine, administered via the intraperitoneal route, was demonstrated to reverse many of the electrophysiological disturbances and, particularly, the prolonged supernormal period, in palytoxin-treated rats. These results suggest that prior activation of sodium ion channels by palytoxin may be blocked by lignocaine.
Ethanol, administered via the intraperitoneal route, was observed to reverse many of the electrophysiological disturbances recorded in the palytoxin-treated rats, however, the supernormal period remained prolonged, yet not particularly enhanced or diminished in magnitude.
Ouabain, administered via the intraperitoneal route, also reversed many of the electrophysiological disturbances induced in palytoxin-treated rats. However, in these animals, the palytoxin-induced supernormal period remained prolonged, yet not exaggerated or diminished in magnitude. These results suggest that membrane excitability may occur independently of, or with, the (Na+,K+)A TPase mechanism.
The effect of palytoxin on mammalian nerve tissue appeared to be modified at a lower temperature. Mixed nerve and motor conduction, F-wave responses, absolute and relative refractory period and supernormal period studies were performed on palytoxin-treated animals at 25°C. The only significant abnormalities were a prolongation of the relative refractory period and of the supernormal period. This study suggests that, at least, in mammalian nerve tissue, this toxin is less active at a lower ambient temperature.
In terms of central studies, in palytoxin-treated rats the brainstem auditory evoked response was not significantly altered although there was significant prolongation of the corticospinal evoked response. This finding suggests that palytoxin is capable of crossing the blood-brain barrier and, then, exerting an effect on central nerve membranes.
These results are similar to those induced by ciguatoxin in the ventral coccygeal nerve of the rat. Both toxins produce slowing of the mixed nerve and motor conduction velocities, a reduction in the amplitude of the mixed nerve and motor action potentials and a prolongation of the refractory period and supernormal period. Lignocaine abolishes the supernormal period induced by both toxins. At a temperature of 25°C, both toxins produce reduced electrophysiological effects, in comparison with those produced at 37°C.
Ciguatoxin is known to bind to and to activate or open sodium ion channels. These studies suggest that at least one action of palytoxin may be that of a similar activation of sodium ion channels.
With reference to these studies and to the fact that both toxins occur within the same ecosystem, palytoxin could potentially contribute, on occasions, to the ciguatera syndrome.
22
Dong, Yue. "Higher Order Neural Networks and Neural Networks for Stream Learning." Thesis, Université d'Ottawa / University of Ottawa, 2017. http://hdl.handle.net/10393/35731.
Full textAbstract:
The goal of this thesis is to explore some variations of neural networks. The thesis is mainly split into two parts: a variation of the shaping functions in neural networks and a variation of learning rules in neural networks.
In the first part, we mainly investigate polynomial perceptrons - a perceptron with a polynomial shaping function instead of a linear one. We prove the polynomial perceptron convergence theorem and illustrate the notion by showing that a higher order perceptron can learn the XOR function through empirical experiments with implementation. In the second part, we propose three models (SMLP, SA, SA2) for stream learning and anomaly detection in streams. The main technique allowing these models to perform at a level comparable to the state-of-the-art algorithms in stream learning is the learning rule used. We employ mini-batch gradient descent algorithm and stochastic gradient descent algorithm to speed up the models. In addition, the use of parallel processing with multi-threads makes the proposed methods highly efficient in dealing with streaming data. Our analysis shows that all models have linear runtime and constant memory requirement. We also demonstrate empirically that the proposed methods feature high detection rate, low false alarm rate, and fast response.
The paper on the first two models (SMLP, SA) is published in the 29th Canadian AI Conference and won the best paper award. The invited journal paper on the third model (SA2) for Computational Intelligence is under peer review.
23
Golestani, Narly A. "Phonetic learning abilities : behavioral, neural functional, and neural anatomical correlates." Thesis, McGill University, 2001. http://digitool.Library.McGill.CA:80/R/?func=dbin-jump-full&object_id=38196.
Full textAbstract:
The studies included in this thesis had as aim to elucidate how individual differences in phonetic learning abilities might be related to differences in more general, psychoacoustic learning abilities, and in how they might be related to differences in brain function and brain morphology.We tested and trained English speaking volunteers to perceive the Hindi dental-retroflex phonetic contrast. We found evidence suggesting that the ability to accurately perceive "difficult" non-native contrasts is not permanently lost during development. We also tested and trained subjects to perceive the difference between non-linguistic rapidly changing and steady-state tonal sounds, and found evidence supporting the hypothesis that successful phonetic learning is in part a function of a more general psychoacoustic ability to process rapidly changing sounds.
The aim of the second study was to determine how the pattern of brain activity may change as a result of training with non-native speech sounds, and in whether it is possible to differentiate "learners" from "non-learners" on the basis of neural activation patterns. Results of this functional magnetic resonance imaging (fMRI) investigation suggested that successful learning of a non-native contrast results in the recruitment of the same areas that are involved in the processing of native contrasts; but the degree of success in learning is accompanied by more efficient neural processing in classical frontal speech regions, while making greater processing demands in left parieto-temporal speech regions.
In the final study, we correlated phonetic learning measures with brain morphology throughout the whole brain volume. We found evidence for overall larger parietal volumes in the left relative to the right hemisphere, and for more white relative to gray matter in the left hemisphere in the learners and not in the nonlearners. This finding is consistent with findings by other investigators suggesting that left-hemispheric dominance for speech may be in part accounted for by hemispheric differences in white matter connectivity, which may allow faster intra- and inter-hemispheric neural transmission. This latter feature may be critical for the processing of consonant speech sounds, which depends on the ability to process sounds that change on the time scale of 30--50 milliseconds.
24
BRUCE, WILLIAM, and OTTER EDVIN VON. "Artificial Neural Network Autonomous Vehicle : Artificial Neural Network controlled vehicle." Thesis, KTH, Maskinkonstruktion (Inst.), 2016. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-191192.
Full textAbstract:
This thesis aims to explain how a Artificial Neural Network algorithm could be used as means of control for a Autonomous Vehicle. It describes the theory behind the neural network and Autonomous Vehicles, and how a prototype with a camera as its only input can be designed to test and evaluate the algorithms capabilites, and also drive using it. The thesis will show that the Artificial Neural Network can, with a image resolution of 100 × 100 and a training set with 900 images, makes decisions with a 0.78 confidence level.Denna rapport har som mal att beskriva hur en Artificiellt Neuronnatverk al- goritm kan anvandas for att kontrollera en bil. Det beskriver teorin bakom neu- ronnatverk och autonoma farkoster samt hur en prototyp, som endast anvander en kamera som indata, kan designas for att testa och utvardera algoritmens formagor. Rapporten kommer visa att ett neuronnatverk kan, med bildupplos- ningen 100 × 100 och traningsdata innehallande 900 bilder, ta beslut med en 0.78 sakerhet.
25
Bader, Sebastian. "Neural-Symbolic Integration." Doctoral thesis, Saechsische Landesbibliothek- Staats- und Universitaetsbibliothek Dresden, 2009. http://nbn-resolving.de/urn:nbn:de:bsz:14-qucosa-25468.
Full textAbstract:
In this thesis, we discuss different techniques to bridge the gap between two different approaches to artificial intelligence: the symbolic and the connectionist paradigm. Both approaches have quite contrasting advantages and disadvantages. Research in the area of neural-symbolic integration aims at bridging the gap between them.
Starting from a human readable logic program, we construct connectionist systems, which behave equivalently. Afterwards, those systems can be trained, and later the refined knowledge be extracted.
26
Allen, T. J. "Optoelectronic neural networks." Thesis, University of Nottingham, 1996. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.362900.
Full text
27
Somers, Harriet. "A neural computer." Thesis, University of York, 1995. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.362021.
Full text
28
SANTOS, RAFAEL DE OLIVAES VALLE DOS. "NEURAL EXPERT WEIGHTING." PONTIFÍCIA UNIVERSIDADE CATÓLICA DO RIO DE JANEIRO, 2012. http://www.maxwell.vrac.puc-rio.br/Busca_etds.php?strSecao=resultado&nrSeq=20153@1.
Full textAbstract:
PONTIFÍCIA UNIVERSIDADE CATÓLICA DO RIO DE JANEIRODiversos resultados empíricos na área de séries temporais indicam que combinar previsores (experts) é, em média, melhor que tentar selecionar um único modelo de previsão. Na medida em que se decide por um esquema de combinação linear, há vários métodos disponíveis para determinar o quanto cada previsor deve contribuir para a resposta consensual, ou em outras palavras, quais devem ser os pesos dos previsores envolvidos. Em um primeiro momento, este trabalho explora o uso prático de diversos métodos tradicionais de ponderação para combinação linear de previsores. Em seguida, propõe um novo sistema para geração de pesos, especialmente projetado para a melhoria do desempenho nas previsões múltiplos passos a frente. O sistema, batizado de Ponderação Neural de Experts (NEW – Neural Expert Weighting), gera modelos de ponderação dinâmica baseados em redes neurais. As redes neurais oferecem a robustez necessária para a simulação de funções de ponderação de alto desempenho, derivadas de um ou mais métodos tradicionais de geração de pesos. O sistema NEW foi avaliado em diversos experimentos comparativos, contemplando 13 séries temporais divididas em dois estudos de casos – derivados do petróleo e competição NN3, uma competição entre metodologias de previsão baseadas em inteligência computacional. Os resultados obtidos foram considerados promissores.
Several empirical results on the time series field indicate that combining forecasting models (experts) is, on average, better than selecting a single forecasting model. Once the linear combination framework is chosen, there are many ways to define the amount of contribution of each combining model to the consensual response; in other words, there are many possible weighting methods. At first, the present work explores the usage of traditional weight generation schemes for the linear combination of forecasters. Afterwards, it proposes a new weight generation framework, specially designed to improve multistep ahead forecasting. The framework, called Neural Experts Weighting (NEW), generates dynamic weighting models based on neural networks. The neural networks provide the desired robustness for the simulation of high performance weighting functions, derived from one or more traditional weighting methods. The NEW framework was assessed with several comparative experiments, encompassing 13 time series divided into two case studies – downstream and NN3 competition, a forecasting competition for computational intelligence methodologies. Results were considered to be promising.
29
Sloan, Cooper Stokes. "Neural bus networks." Thesis, Massachusetts Institute of Technology, 2018. http://hdl.handle.net/1721.1/119711.
Full textAbstract:
Thesis: M. Eng., Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, 2018.This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections.
Cataloged from student-submitted PDF version of thesis.
Includes bibliographical references (pages 65-68).
Bus schedules are unreliable, leaving passengers waiting and increasing commute times. This problem can be solved by modeling the traffic network, and delivering predicted arrival times to passengers. Research attempts to model traffic networks use historical, statistical and learning based models, with learning based models achieving the best results. This research compares several neural network architectures trained on historical data from Boston buses. Three models are trained: multilayer perceptron, convolutional neural network and recurrent neural network. Recurrent neural networks show the best performance when compared to feed forward models. This indicates that neural time series models are effective at modeling bus networks. The large amount of data available for training bus network models and the effectiveness of large neural networks at modeling this data show that great progress can be made in improving commutes for passengers.
by Cooper Stokes Sloan.
M. Eng.
30
Minev, Ivan Rusev. "Soft neural interfaces." Thesis, University of Cambridge, 2012. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.610257.
Full text
31
De, Jongh Albert. "Neural network ensembles." Thesis, Stellenbosch : Stellenbosch University, 2004. http://hdl.handle.net/10019.1/50035.
Full textAbstract:
Thesis (MSc)--Stellenbosch University, 2004.ENGLISH ABSTRACT: It is possible to improve on the accuracy of a single neural network by using an ensemble of diverse and accurate networks. This thesis explores diversity in ensembles and looks at the underlying theory and mechanisms employed to generate and combine ensemble members. Bagging and boosting are studied in detail and I explain their success in terms of well-known theoretical instruments. An empirical evaluation of their performance is conducted and I compare them to a single classifier and to each other in terms of accuracy and diversity.
AFRIKAANSE OPSOMMING: Dit is moontlik om op die akkuraatheid van 'n enkele neurale netwerk te verbeter deur 'n ensemble van diverse en akkurate netwerke te gebruik. Hierdie tesis ondersoek diversiteit in ensembles, asook die meganismes waardeur lede van 'n ensemble geskep en gekombineer kan word. Die algoritmes "bagging" en "boosting" word in diepte bestudeer en hulle sukses word aan die hand van bekende teoretiese instrumente verduidelik. Die prestasie van hierdie twee algoritmes word eksperimenteel gemeet en hulle akkuraatheid en diversiteit word met 'n enkele netwerk vergelyk.
32
Simmen, Martin Walter. "Neural network optimization." Thesis, University of Edinburgh, 1992. http://hdl.handle.net/1842/12942.
Full textAbstract:
Combinatorial optimization problems arise throughout science, industry, and commerce. The demonstration that analogue neural networks could, in principle, rapidly find near-optimal solutions to such problems - many of which appear computationally intractable - was important both for the novelty of the approach and because these networks are potentially implementable in parallel hardware. However, subsequent research, conducted largely on the travelling salesman problem, revealed problems regarding the original network's parameter sensitivity and tendency to give invalid states. Although this has led to improvements and new network designs which at least partly overcome the above problems, many issues concerning the performance of optimization networks remain unresolved. This thesis explores how to optimize the performance of two neural networks current in the literature: the elastic net, and the mean field Potts network, both of which are designed for the travelling salesman problem. Analytical methods elucidate issues of parameter sensitivty and enable parameter values to be chosen in a rational manner. Systematic numerical experiments on realistic size problems complement and support the theoretical analyses throughout. An existing analysis of how the elastic net algorithm may generate invalid solutions is reviewed and extended. A new analysis locates the parameter regime in which the net may converge to a second type of invalid solution. Combining the two analyses yields a prescription for setting the value of a key parameter optimally with respect to avoiding invalid solutions. The elastic net operates by minimizing a computational energy function. Several new forms of dynamics using locally adaptive step-sizes are developed, and shown to increase greatly the efficiency of the minimization process. Analytical work constraining the range of safe adaptation rates is presented. A new form of dynamics, with a user defined step-size, is introduced for the mean field Potts network. An analysis of the network's critical temperature under these dynamics is given, by generalizing a previous analysis valid for a special case of the dynamics.
33
Lewis, Sarah Noami. "The neural marketplace." Thesis, Imperial College London, 2016. http://hdl.handle.net/10044/1/43539.
Full textAbstract:
The 'retroaxonal hypothesis' (Harris, 2008) posits a role for slow retrograde signalling in learning. It is based on the intuition that cells with strong output synapses tend to be those that encode useful information; and that cells which encode useful information should not modify their input synapses too readily. The hypothesis has two parts: rst, that the stronger a cell's output synapses, the less likely it is to change its input synapses; and second, that a cell is more likely to revert changes to its input synapses when the changes are followed by weakening of its output synapses. It is motivated in part by analogy between a neural network and a market economy, viewing neurons as 'entrepreneurs' who 'sell' spike trains to each other. In this view, the slow retrograde signals which tell a neuron that it has strong output synapses are 'money' and imply that what it produces is useful. This thesis constructs a mathematical model of learning, which validates the intuition of the retroaxonal hypothesis. In this model, we show that neurons can estimate their usefulness, or 'worth', from the magnitude of their output weights. We also show that by making each cell's input synapses more or less plastic according to its worth, the performance of a network can be improved.
34
Смаль, Богдан Віталійович. "Artificial Neural Network." Thesis, Київський національний університет технологій та дизайну, 2017. https://er.knutd.edu.ua/handle/123456789/7384.
Full text
35
Boychenko, I. V., and G. I. Litvinenko. "Artificial neural networks." Thesis, Вид-во СумДУ, 2009. http://essuir.sumdu.edu.ua/handle/123456789/17044.
Full text
36
Landry, Kenneth D. "Evolutionary neural networks." Thesis, Virginia Polytechnic Institute and State University, 1988. http://hdl.handle.net/10919/51904.
Full textAbstract:
To create neural networks that work, one needs to specify a structure and the interconnection weights between each pair of connected computing elements. The structure of a network can be selected by the designer depending on the application, although the selection of interconnection weights is a much larger problem. Algorithms have been developed to alter the weights slightly in order to produce the desired results. Learning algorithms such as Hebb's rule, the Delta rule and error propagation have been used, with success, to learn the appropriate weights. The major objection to this class of algorithms is that one cannot specify what is not desired in the network in addition to what is desired. An alternate method to learning the correct interconnection weights is to evolve a network in an environment that rewards "good” behavior and punishes "bad" behavior, This technique allows interesting networks to appear which otherwise may not be discovered by other methods of learning. In order to teach a network the correct weights, this approach simply needs a direction where an acceptable solution can be obtained rather than a complete answer to the problem.Master of Science
37
Merrill, Bradford Chester. "The neural dogma." abstract and full text PDF (UNR users only), 2008. http://0-gateway.proquest.com.innopac.library.unr.edu/openurl?url_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:dissertation&res_dat=xri:pqdiss&rft_dat=xri:pqdiss:1460809.
Full text
38
Lunn, Julian Alexander. "Canine Neural Angiostrongyliasis." Thesis, The University of Sydney, 2006. http://hdl.handle.net/2123/2077.
Full textAbstract:
Summary Canine Neural Angiostrongyliasis (CNA) is caused by the obligatory neural migration of Angiostrongylus cantonensis larvae in dogs. Characteristically, cases are juvenile dogs with progressive CNS dysfunction characterised by hyperaesthesia and often associated with eosinophilic pleocytosis of the CSF. In Australia, most cases occur between March and June. The rat lungworm, A cantonensis was first described by Chen in 1935 in Canton, China. While initially called Pulmonema cantonensis the parasite was later reclassified as A cantonensis. A disease diagnosed as eosinophilic meningoencephalitis was first described in 1944 in Taiwan. The same disease was reported in 1948 in the East Caroline Islands but it was not until 1961 that A cantonensis was confirmed as the aetiological agent when a patient in a Hawaiian mental institution, who had died of eosinophilic meningoencephalitis, had A cantonensis larvae recovered from the brain and spinal cord. The first reports of animals infected with A cantonensis were made by Mason in 1976 when he described a syndrome occurring in puppies in the Brisbane area, characterised by urinary incontinence, hind limb paresis and hyperaesthesia, often associated with eosinophilic pleocytosis of the CSF. Reports of infection in other species followed including macropods, bats, horses, primates and birds. Twenty-two cases of suspected CNA were collected prospectively to compare with those previously described, including 37 cases published by Mason in 1983, and to examine the accuracy of an ELISA used to diagnose human neural angiostrongyliasis in Australia. Samples were collected from two control populations in an attempt to validate the ELISA results. In the prospective series of cases, there was a significantly older subpopulation of dogs in addition to “classical” young dogs, suggesting that this syndrome can occur at any age and should be considered a differential in any dog with progressive neurological disease. The mortality rate in the prospective group was lower than in the published group, which is a reflection of the severity of the disease in younger animals as is the case with human patients. Definitive diagnosis of neural angiostrongyliasis in human patients has been achieved by identifying A cantonensis larvae within the CSF or aqueous humour. In dogs, the only definitive way to diagnose CNA has been via necropsy. While many cases of CNA are characteristic and presumptive diagnosis can be made based on typical history, signalment, clinical signs, CSF analysis and response to glucocorticoids, there appear to be an increasing number of cases occurring in older dogs, that displaying focal, atypical clinical signs or that develop permanent sequelae. Serology has been a useful tool in diagnosing neural angiostrongyliasis in humans. In its current form the ELISA is not sensitive or specific enough to allow a definitive diagnosis of CNA to be made using serum but is useful when applied to CSF specimens. Further refinement of the antigen or using monoclonal rather than polyclonal antibodies may improve the accuracy of the serology. Alternatively, methods such as Western Blot, Immuno-PCR or dot-blot ELISA, which have been successfully used to diagnoses angiostrongyliasis in humans, may be worthy of investigation The major differential diagnosis for CNA is neosporosis. Other differential diagnoses include idiopathic eosinophilic meningoencephalitis, parasitic infections including Toxoplasma gondii, Taenia solium, Gnathostoma spinigerum, visceral larval migrans (Toxocara canis) and schistosomiasis, fungal, bacterial, viral and rickettsial infections as well as neoplasia, trauma, drug reactions and toxicities. Treatment of CNA has been limited to glucocorticoids, however there may be adjunct therapies including anthelmintices, cyclosporine, and matrix metalloproteinase inhibitors. In Mason’s series of cases the use of anthelmintics significantly worsened the clinical outcome for patients. It does not appear, however, that the use of these agents in species other than the dog exacerbates clinical signs. Acquired immunity is short lived in rats and mice, which would suggest the same is true in dogs. Routine heartworm and intestinal parasite prophylaxis appears to have no influence on the occurrence of CNA.
39
Park, Sean. "Neural malware detection." Thesis, Federation University Australia, 2019. http://researchonline.federation.edu.au/vital/access/HandleResolver/1959.17/173759.
Full textAbstract:
At the heart of today’s malware problem lies theoretically infinite diversity created by metamorphism. The majority of conventional machine learning techniques tackle the problem with the assumptions that a sufficiently large number of training samples exist and that the training set is independent and identically distributed. However, the lack of semantic features combined with the models under these wrong assumptions result largely in overfitting with many false positives against real world samples, resulting in systems being left vulnerable to various adversarial attacks. A key observation is that modern malware authors write a script that automatically generates an arbitrarily large number of diverse samples that share similar characteristics in program logic, which is a very cost-effective way to evade detection with minimum effort. Given that many malware campaigns follow this paradigm of economic malware manufacturing model, the samples within a campaign are likely to share coherent semantic characteristics. This opens up a possibility of one-to-many detection. Therefore, it is crucial to capture this non-linear metamorphic pattern unique to the campaign in order to detect these seemingly diverse but identically rooted variants. To address these issues, this dissertation proposes novel deep learning models, including generative static malware outbreak detection model, generative dynamic malware detection model using spatio-temporal isomorphic dynamic features, and instruction cognitive malware detection. A comparative study on metamorphic threats is also conducted as part of the thesis. Generative adversarial autoencoder (AAE) over convolutional network with global average pooling is introduced as a fundamental deep learning framework for malware detection, which captures highly complex non-linear metamorphism through translation invariancy and local variation insensitivity. Generative Adversarial Network (GAN) used as a part of the framework enables oneshot training where semantically isomorphic malware campaigns are identified by a single malware instance sampled from the very initial outbreak. This is a major innovation because, to the best of our knowledge, no approach has been found to this challenging training objective against the malware distribution that consists of a large number of very sparse groups artificially driven by arms race between attackers and defenders. In addition, we propose a novel method that extracts instruction cognitive representation from uninterpreted raw binary executables, which can be used for oneto- many malware detection via one-shot training against frequency spectrum of the Transformer’s encoded latent representation. The method works regardless of the presence of diverse malware variations while remaining resilient to adversarial attacks that mostly use random perturbation against raw binaries. Comprehensive performance analyses including mathematical formulations and experimental evaluations are provided, with the proposed deep learning framework for malware detection exhibiting a superior performance over conventional machine learning methods. The methods proposed in this thesis are applicable to a variety of threat environments here artificially formed sparse distributions arise at the cyber battle fronts.Doctor of Philosophy
40
Coelho, Regina Célia. "Síntese,modelagem e simulação de estruturas neurais morfologicamente realísticas." Universidade de São Paulo, 1998. http://www.teses.usp.br/teses/disponiveis/76/76132/tde-15052009-091429/.
Full textAbstract:
Os aspectos morfológicos dos neurônios e estruturas neurais, embora potencialmente importantes, têm recebido relativamente pouca atenção na literatura em neurociência. Este trabalho consiste numa substancial parte de um projeto em desenvolvimento no Grupo de Pesquisa em Visão Cibernética voltado para o estudo da relação formal/função neural. Mais especificamente, o presente trabalho dedica particular atenção para a síntese, modelagem e simulação de estruturas neurais morfologicamente realísticas. A tese se inicia com revisões bibliográficas sobre visão biológica e neurociência, direcionadas aos assuntos a serem aqui abordados. Começamos a descrição dos desenvolvimentos com um levantamento, avaliação e proposta de medidas neuromorfométricas adequadas para exprimir as propriedades mais representativas para nosso trabalho, tais como cobertura espacial, complexidade e decaimento eletrônico. Incluímos nessa parte a metodologia utilizada para a geração de neurônios artificiais bidimensionais estatisticamente semelhantes aos naturais. Apresenta-se também a extensão desta metodologia para o caso tridimensional, validada pela análise neuroinorfométrica dos neurônios gerados. Na seqüência, descrevemos o processo de geração de estruturas neurais compostas de neurônios. Considerando modelos com uma camada neural para a codificação de especificidade de orientação, mas sem levar em conta a forma neural, vários casos são simulados, utilizando gradientes na distribuição dos pesos sinápticos e distribuições regulares ou aleatórias (uniformes) dos neurônios na estrutura. A extensão dessas simulações utilizando estruturas que consideram mais detalhadamente a forma neural, usando agora neurônios artificiais gerados pelo método descrito nesta monografia, é apresentada na seqüência. Entre outros efeitos, mostramos que a extensão da arborização dendrítica é um fator determinante da taxa de convergência e seletividade nos modelos, e que gradientes na extensão das arborizações sinápticas são essenciais para a adequada codificação de orientações em módulos cêntricos contendo somatas aleatoriamente distribuídos.The morphological aspects of neurons and neural structures, although potentially important, have received relatively little attention in the literature in neuroscience. This work consists in a substantial part of a project in development at the Cybernetic Vision Research Group, directed to the study of the form/function relationship. More specifically, the present work dedicates particular attention to the synthesis, modeling, and simulation of morphologically realistic neural structures. The thesis begins with a bibliographic review about biological vision and neuroscience, focusing on the subjects to be here considered. We start the description of the developments with the revision; evaluation and proposal of neuromorphometric measures adequate express the properties more representative to the work, such as spatial cover, complexity and electrotonic decay. We include in this part the methodology used for the generation of bidimensional artificial neurons statistically similar to natural ones. The extension of these developments to the tridimensional case, including their respective validation (performed in terms of neuromorphometric analysis of the generated neurons) is also presented. Next, we describe the generation process of neural structures composed of neurons. Using one-layer neural models for orientation specificity encoding, but without considering the neural shape, several cases are simulated, using gradients in the distribution of the synaptic weights and regular or random (uniform) distributions of the neurons in the structures. The extension of these simulations using structures that consider the neural form in more detail, composed of artificial neurons generated by the described method in this monograph is presented in the sequence. We show that the extension of the dendritic arborization is a determinant factor on the convergence rate and selectivity in the models, and that gradients in the extension of the synaptic arborizations are essentials for the adequate codification of orientations in centric models containing distributed random somata.
41
Hajj, Hassan Mohamad. "Hybrid NeuroMEMS simultaneous recording of neural electro- potentials and neural biomarkers /." Thesis, McGill University, 2010. http://digitool.Library.McGill.CA:8881/R/?func=dbin-jump-full&object_id=92391.
Full text
42
Kobak, Eva-Maria. "Neural representation of movements : insights from neural activity and from behaviour." Thesis, Imperial College London, 2013. http://hdl.handle.net/10044/1/39132.
Full textAbstract:
It is long known that the motor cortex is involved in the execution of movements. However, how exactly movements are represented in the brain remains largely unknown. In this thesis, I adopt two different approaches to study natural control of movements. In the first approach, neural signals in the motor cortex are analysed and brought in relation to concurrent movements. In the second approach the behaviour of subjects is observed to deduce characteristics of the underlying neural representations. For the analysis of neural signals, I had access to multi-electrode recordings from motor cortex of behaving monkeys. In the phase of local field potentials, a spatiotemporal pattern resembling a propagating wave can be observed. On a single trial basis, I could show that this pattern is present over a very broad frequency range (0.3???90 Hz) throughout which the wavelength stays relatively constant, causing a linear increase in propagation speed with frequency. Two models, a travelling wavepacket and a dipole model, will be discussed as possible sources of the observed pattern. Additionally, I investigated the information content of the phases of LFP signals with regards to movement direction, which could potentially have implications for the application in a brain machine interface. The behaviour of human subjects was studied in a motor-psychophysical experiment in which a reaching movement was performed under a visuomotor rotation. The two parts of the movement, reaching forward to the target and returning to the starting location, were analysed separately and subjects were prevented from learning the rotation directly in the latter. Thereby, I could estimate the influence of proprioception on backward motor control, and could demonstrate that there is learning transfer between forward and backward movements.
43
Dunn, Nathan A. "A Novel Neural Network Analysis Method Applied to Biological Neural Networks." Thesis, view abstract or download file of text, 2006. http://proquest.umi.com/pqdweb?did=1251892251&sid=2&Fmt=2&clientId=11238&RQT=309&VName=PQD.
Full textAbstract:
Thesis (Ph. D.)--University of Oregon, 2006.Typescript. Includes vita and abstract. Includes bibliographical references (leaves 122- 131). Also available for download via the World Wide Web; free to University of Oregon users.
44
Di, Gangi Mattia Antonino. "Neural Speech Translation: From Neural Machine Translation to Direct Speech Translation." Doctoral thesis, Università degli studi di Trento, 2020. http://hdl.handle.net/11572/259137.
Full textAbstract:
Sequence-to-sequence learning led to significant improvements to machine translation (MT) and automatic speech recognition (ASR) systems. These advancements were first reflected in spoken language translation (SLT) when using a cascade of (at least) ASR and MT with the new "neural" models, then by using sequence-to-sequence learning to directly translate the input audio speech into text in the target language. In this thesis we cover both approaches to the SLT task. First, we show the limits of NMT in terms of robustness to input errors when compared to the previous phrase-based state of the art. We then focus on the NMT component to achieve better translation quality with higher computational efficiency by using a network based on weakly-recurrent units. Our last work involving a cascade explores the effects on the NMT robustness when adding automatic transcripts to the training data. In order to move to the direct speech-to-text approach, we introduce MuST-C, the largest multilingual SLT corpus for training direct translation systems. MuST-C increases significantly the size of publicly available data for this task as well as their language coverage. With such availability of data, we adapted the Transformer architecture to the SLT task for its computational efficiency . Our adaptation, which we call S-Transformer, is meant to better model the audio input, and with it we set a new state of the art for MuST-C. Building on these positive results, we finally use S-Transformer with different data applications: i) one-to-many multilingual translation by training it on MuST-C; ii participation to the IWSLT 19 shared task with data augmentation; and iii) instance-based adaptation for using the training data at test time. The results in this thesis show a steady quality improvement in direct SLT. Our hope is that the presented resources and technological solutions will increase its adoption in the near future, so to make multilingual information access easier in a globalized world.
45
Nistor, Paul Andrei. "TWIST1 : a subtle modulator of neural differentiation and neural tube formation." Thesis, University of Edinburgh, 2013. http://hdl.handle.net/1842/11823.
Full textAbstract:
The central nervous system is formed from epiblast precursor cells through Neurulation. Neural induction can be studied in its main aspects in vitro. However, the process is poorly understood, especially in regard to when and how a cell becomes specified, and then committed, to be a neural cell. It is, on the other hand, well established that neural formation requires absence or, inhibition of the BMP signalling both in vivo and in vitro. ID1 is a direct target of BMP signalling with major influence on in vitro neural differentiation. A cDNA library screen, looking for transcription factors negatively regulated by ID1, reported TWIST1, along with only two other proteins. Twist1 expression is upregulated during in vitro neural differentiation. Furthermore, targeted deletion of Twist1 has dramatic consequences on anterior neural development. Twist1 knock-out mice fail to form the closed neural tube in the prospective brain, followed by exencephaly and, early embryonic death. In this thesis I investigate the influence on in vitro neural differentiation of a TWIST1 constitutively active form, insensitive to ID1 inhibition. I report that this transcriptionally active TWIST1 accelerates neural differentiation, in vitro and, biases it, towards dorsal phenotypes. I provide, for the first time, evidence for Twist1 expression in the neural tissue, observed weakly in a restricted domain, temporally and spatially, in the dorsal part of the neural tube. I propose a new model for TWIST1 influence at this level. I also investigate how TWIST1 actions depend on levels of expression and dimer choice. I found that, TWIST1 can exert its neural modulating actions only at low levels, as high levels divert a cell fate towards non-neural lineages.
46
Goerendt, Ines Katharina. "Reward processing in Parkinson's disease : adaptation in neural circuitries and neural systems." Thesis, Imperial College London, 2002. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.396317.
Full text
47
Bolt, George Ravuama. "Fault tolerance in artificial neural networks : are neural networks inherently fault tolerant?" Thesis, University of York, 1992. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.317683.
Full text
48
Cooper, J. E. "Is neural crest cell delamination required for normal cranial neural tube closure?" Thesis, University College London (University of London), 2012. http://discovery.ucl.ac.uk/1343905/.
Full textAbstract:
Numerous correlations are described in the literature between cranial neural tube defects (NTDs) and neurocristopathies indicating that cranial neural tube closure and neural crest cell (NCC) development may be linked by more than just spatial and temporal contiguity. Detailed analysis of the morphology of the cranial neural plate identified the midbrain as a region in which several features combine that are likely to result in resistance to the apposition and subsequent closure of the neural folds. The relationship between elevation, bending and closure of the midbrain neural folds and the specification and delamination of NCC indicates that NCC may act in conjunction with other permissive processes to facilitate a) elevation of the neural folds by contributing to expansion of cranial mesenchyme and b) formation of dorsolateral hinge points (DLHP) by reducing cell density and thus enhancing flexibility of the dorsal neural folds. These two processes are requirements for the subsequent closure of the midbrain. To address the hypothesis that NCC delamination is required for the elevation of the midbrain neural folds and their bending at the DLHP, mouse models known to harbour mutations resulting in both NTDs and neurocristopathies were studied to assess the relationship between the two defects. In support of the idea that NCC delamination facilitates midbrain elevation and DLHP formation, failure of cranial NCC delamination associates with reduced cranial elevation, absence of DLHPs and midbrain exencephaly in the Kumba mutant mouse model. This is in contrast to the dissociation between the trunk NCC phenotype and hindbrain exencephaly observed in the Splotch model. The hypothesis was tested experimentally by chemically inhibiting the delamination of NCC in cultured embryos. This adversely affected elevation of the neural folds and DLHP formation, and in some instances resulted in failure of midbrain closure. A transgenic model was developed which was predicted to provide an in vivo model of inhibition of delamination of NCC. The shRNA mediated knockdown of FoxD3 expression in NCC did not, however, affect the early specification or delamination of NCC. Instead it resulted in a failure of maintenance of NCC progenitors during their migration in the cranial mesenchyme. This model displayed no incidence of midbrain exencephaly. Failure of proper NCC derived mesenchymal ‘scaffolding’ surrounding the cranial neural tube did, however, lead to a reopening of the forebrain in some instances. Based on the evidence described above, I propose a model in which the development of NCC exerts complex multilevel mechanical regulation on the formation and maintenance of the neural tube. NCC delamination facilitates DLHP formation, while NCC migration and proliferation in the mesenchyme contributes to elevation of the cranial neural folds and also ‘scaffolds’ the neural folds to maintain closure.
49
Vieira, Inês. "Neural circuits of the mouse olfactory cortex : linking neural connectivity to behavior." Thesis, Paris 6, 2017. http://www.theses.fr/2017PA066543/document.
Full textAbstract:
Comment les odeurs contrôlent-elles le comportement animal ? Dans ma thèse, j'ai utilisé des manipulations optogénétiques et chimiogénétiques in vivo de l'activité neurale combinées à des analyses comportementales pour explorer l'organisation de circuits cérébraux impliqués dans des comportements olfactifs chez la souris. J'ai mis au point un test de conditionnement aversif olfactif indépendant de l'intensité des odeurs. J'ai démontré que les souris pouvaient généraliser une réponse aversive en présentant différentes concentrations d'odeurs. J’ai ensuite testé si les souris pouvaient apprendre cette tâche en inactivant les interneurones exprimant la parvalbumine dans le cortex olfactif (piriforme). J'ai trouvé que l’inactivation des cellules PV, n'était pas suffisante pour abolir l'aversion aux odeurs acquise, ce qui suggère que des composants de circuits neuronaux supplémentaires contribuent à la perception de l'odeur indépendamment de sa concentration. Ensuite, j'ai tenté de comprendre la constitution relative des différentes voies neurales du piriforme dans ce comportement d’aversion apprise. À l'aide d'outils génétiques et viraux, j'ai ciblé des sous-populations distinctes de neurones piriformes, et j'ai constaté que l'activité neurale induite par la lumière dans les cellules du piriforme projetant vers le bulbe olfactif et vers le cortex préfrontal, mais pas dans les cellules du piriforme projetant vers l’amygdale corticale et vers le cortex entorhinal latéral était suffisante pour supporter le conditionnement aversif. Ces résultats contribuent à mieux comprendre les propriétés fonctionnelles des circuits neuronaux corticaux pour l'olfactionHow do odors control animal behavior ? In my thesis, I have used in vivo optogenetic and chemogenetic manipulations of neural activity combined with behavioral analyses to explore the organization of brain circuits involved in olfactory behaviors in mice. In the first part of the thesis, I established an odor intensity-independent olfactory conditioning task. I demonstrated that mice were able to generalize a learned escape behavior across a range of different odor concentrations. I then tested if by silencing Parvalbumin-expressing interneurons in the olfactory (piriform) cortex, a candidate cell population for mediating odor concentration invariance, mice would fail to learn the task. I found that silencing PV cells was not sufficient to abolish learned aversion, suggesting that additional neural circuit components contribute to concentration-invariant odor perception. Next, I asked whether different piriform neural output pathways differed in their ability to support learned aversion. Using viral-genetic tools, I targeted distinct subpopulations of piriform neurons and I found that light-induced neural activity in only piriform principle cells could drive a behavioral response. Furthermore, I tested the sufficiency of subpopulations of piriform projection neurons to drive learned aversion. I found that photostimulation of olfactory bulb- and prefrontal cortex-projecting piriform neurons was sufficient to support aversive conditioning, but not the photostimulation of cortical amygdala- and lateral entorhinal cortex-projecting piriform neurons. Together, these results provide new insights into the functional properties of cortical neural circuits for olfaction
50
Borra, Davide <1992>. "Interpretable Convolutional Neural Networks for Decoding and Analyzing Neural Time Series Data." Doctoral thesis, Alma Mater Studiorum - Università di Bologna, 2022. http://amsdottorato.unibo.it/10345/1/phdthesis_dborra.pdf.
Full textAbstract:
Machine learning is widely adopted to decode multi-variate neural time series, including electroencephalographic (EEG) and single-cell recordings. Recent solutions based on deep learning (DL) outperformed traditional decoders by automatically extracting relevant discriminative features from raw or minimally pre-processed signals.
Convolutional Neural Networks (CNNs) have been successfully applied to EEG and are the most common DL-based EEG decoders in the state-of-the-art (SOA). However, the current research is affected by some limitations. SOA CNNs for EEG decoding usually exploit deep and heavy structures with the risk of overfitting small datasets, and architectures are often defined empirically. Furthermore, CNNs are mainly validated by designing within-subject decoders. Crucially, the automatically learned features mainly remain unexplored; conversely, interpreting these features may be of great value to use decoders also as analysis tools, highlighting neural signatures underlying the different decoded brain or behavioral states in a data-driven way. Lastly, SOA DL-based algorithms used to decode single-cell recordings rely on more complex, slower to train and less interpretable networks than CNNs, and the use of CNNs with these signals has not been investigated.
This PhD research addresses the previous limitations, with reference to P300 and motor decoding from EEG, and motor decoding from single-neuron activity. CNNs were designed light, compact, and interpretable. Moreover, multiple training strategies were adopted, including transfer learning, which could reduce training times promoting the application of CNNs in practice. Furthermore, CNN-based EEG analyses were proposed to study neural features in the spatial, temporal and frequency domains, and proved to better highlight and enhance relevant neural features related to P300 and motor states than canonical EEG analyses. Remarkably, these analyses could be used, in perspective, to design novel EEG biomarkers for neurological or neurodevelopmental disorders. Lastly, CNNs were developed to decode single-neuron activity, providing a better compromise between performance and model complexity.