Segui questo link per vedere altri tipi di pubblicazioni sul tema: Neural network.
Tesi sul tema "Neural network"
Cita una fonte nei formati APA, MLA, Chicago, Harvard e in molti altri stili
Vedi i top-50 saggi (tesi di laurea o di dottorato) per l'attività di ricerca sul tema "Neural network".
Accanto a ogni fonte nell'elenco di riferimenti c'è un pulsante "Aggiungi alla bibliografia". Premilo e genereremo automaticamente la citazione bibliografica dell'opera scelta nello stile citazionale di cui hai bisogno: APA, MLA, Harvard, Chicago, Vancouver ecc.
Puoi anche scaricare il testo completo della pubblicazione scientifica nel formato .pdf e leggere online l'abstract (il sommario) dell'opera se è presente nei metadati.
Vedi le tesi di molte aree scientifiche e compila una bibliografia corretta.
1
Xu, Shuxiang, University of Western Sydney, and of Informatics Science and Technology Faculty. "Neuron-adaptive neural network models and applications." THESIS_FIST_XXX_Xu_S.xml, 1999. http://handle.uws.edu.au:8081/1959.7/275.
Testo completoAbstract (sommario):
Artificial Neural Networks have been widely probed by worldwide researchers to cope with the problems such as function approximation and data simulation. This thesis deals with Feed-forward Neural Networks (FNN's) with a new neuron activation function called Neuron-adaptive Activation Function (NAF), and Feed-forward Higher Order Neural Networks (HONN's) with this new neuron activation function. We have designed a new neural network model, the Neuron-Adaptive Neural Network (NANN), and mathematically proved that one NANN can approximate any piecewise continuous function to any desired accuracy. In the neural network literature only Zhang proved the universal approximation ability of FNN Group to any piecewise continuous function. Next, we have developed the approximation properties of Neuron Adaptive Higher Order Neural Networks (NAHONN's), a combination of HONN's and NAF, to any continuous function, functional and operator. Finally, we have created a software program called MASFinance which runs on the Solaris system for the approximation of continuous or discontinuous functions, and for the simulation of any continuous or discontinuous data (especially financial data). Our work distinguishes itself from previous work in the following ways: we use a new neuron-adaptive activation function, while the neuron activation functions in most existing work are all fixed and can't be tuned to adapt to different approximation problems; we only use on NANN to approximate any piecewise continuous function, while a neural network group must be utilised in previous research; we combine HONN's with NAF and investigate its approximation properties to any continuous function, functional, and operator; we present a new software program, MASFinance, for function approximation and data simulation. Experiments running MASFinance indicate that the proposed NANN's present several advantages over traditional neuron-fixed networks (such as greatly reduced network size, faster learning, and lessened simulation errors), and that the suggested NANN's can effectively approximate piecewise continuous functions better than neural networks groups. Experiments also indicate that NANN's are especially suitable for data simulation<br>Doctor of Philosophy (PhD)
2
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.
Testo completo
3
Patterson, Raymond A. "Hybrid Neural networks and network design." Connect to resource, 1995. http://rave.ohiolink.edu/etdc/view.cgi?acc%5Fnum=osu1262707683.
Testo completo
4
Khliobas. "NEURAL NETWORK." Thesis, Київ 2018, 2018. http://er.nau.edu.ua/handle/NAU/33752.
Testo completo
5
Rastogi, Preeti. "Assessing Wireless Network Dependability Using Neural Networks." Ohio University / OhioLINK, 2005. http://rave.ohiolink.edu/etdc/view?acc_num=ohiou1129134364.
Testo completo
6
Chambers, Mark Andrew. "Queuing network construction using artificial neural networks /." The Ohio State University, 2000. http://rave.ohiolink.edu/etdc/view?acc_num=osu1488193665234291.
Testo completo
7
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.
Testo completoAbstract (sommario):
Thesis (Ph. D.)--University of Oregon, 2006.<br>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.
8
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.
Testo completoAbstract (sommario):
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.<br>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.
9
De, Jongh Albert. "Neural network ensembles." Thesis, Stellenbosch : Stellenbosch University, 2004. http://hdl.handle.net/10019.1/50035.
Testo completoAbstract (sommario):
Thesis (MSc)--Stellenbosch University, 2004.<br>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.<br>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.
10
Simmen, Martin Walter. "Neural network optimization." Thesis, University of Edinburgh, 1992. http://hdl.handle.net/1842/12942.
Testo completoAbstract (sommario):
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.
11
Смаль, Богдан Віталійович. "Artificial Neural Network." Thesis, Київський національний університет технологій та дизайну, 2017. https://er.knutd.edu.ua/handle/123456789/7384.
Testo completo
12
Post, David L. "Network Management: Assessing Internet Network-Element Fault Status Using Neural Networks." Ohio : Ohio University, 2008. http://www.ohiolink.edu/etd/view.cgi?ohiou1220632155.
Testo completo
13
Xu, Shuxiang. "Neuron-adaptive neural network models and applications." Thesis, [Campbelltown, N.S.W. : The Author], 1999. http://handle.uws.edu.au:8081/1959.7/275.
Testo completoAbstract (sommario):
Artificial Neural Networks have been widely probed by worldwide researchers to cope with the problems such as function approximation and data simulation. This thesis deals with Feed-forward Neural Networks (FNN's) with a new neuron activation function called Neuron-adaptive Activation Function (NAF), and Feed-forward Higher Order Neural Networks (HONN's) with this new neuron activation function. We have designed a new neural network model, the Neuron-Adaptive Neural Network (NANN), and mathematically proved that one NANN can approximate any piecewise continuous function to any desired accuracy. In the neural network literature only Zhang proved the universal approximation ability of FNN Group to any piecewise continuous function. Next, we have developed the approximation properties of Neuron Adaptive Higher Order Neural Networks (NAHONN's), a combination of HONN's and NAF, to any continuous function, functional and operator. Finally, we have created a software program called MASFinance which runs on the Solaris system for the approximation of continuous or discontinuous functions, and for the simulation of any continuous or discontinuous data (especially financial data). Our work distinguishes itself from previous work in the following ways: we use a new neuron-adaptive activation function, while the neuron activation functions in most existing work are all fixed and can't be tuned to adapt to different approximation problems; we only use on NANN to approximate any piecewise continuous function, while a neural network group must be utilised in previous research; we combine HONN's with NAF and investigate its approximation properties to any continuous function, functional, and operator; we present a new software program, MASFinance, for function approximation and data simulation. Experiments running MASFinance indicate that the proposed NANN's present several advantages over traditional neuron-fixed networks (such as greatly reduced network size, faster learning, and lessened simulation errors), and that the suggested NANN's can effectively approximate piecewise continuous functions better than neural networks groups. Experiments also indicate that NANN's are especially suitable for data simulation
14
Xu, Shuxiang. "Neuron-adaptive neural network models and applications /." [Campbelltown, N.S.W. : The Author], 1999. http://library.uws.edu.au/adt-NUWS/public/adt-NUWS20030702.085320/index.html.
Testo completo
15
Brande, Julia K. Jr. "Computer Network Routing with a Fuzzy Neural Network." Diss., Virginia Tech, 1997. http://hdl.handle.net/10919/29685.
Testo completoAbstract (sommario):
The growing usage of computer networks is requiring improvements in network technologies and management techniques so users will receive high quality service. As more individuals transmit data through a computer network, the quality of service received by the users begins to degrade. A major aspect of computer networks that is vital to quality of service is data routing. A more effective method for routing data through a computer network can assist with the new problems being encountered with today's growing networks.
Effective routing algorithms use various techniques to determine the most appropriate route for transmitting data. Determining the best route through a wide area network (WAN), requires the routing algorithm to obtain information concerning all of the nodes, links, and devices present on the network. The most relevant routing information involves various measures that are often obtained in an imprecise or inaccurate manner, thus suggesting that fuzzy reasoning is a natural method to employ in an improved routing scheme. The neural network is deemed as a suitable accompaniment because it maintains the ability to learn in dynamic situations. Once the neural network is initially designed, any alterations in the computer routing environment can easily be learned by this adaptive artificial intelligence method. The capability to learn and adapt is essential in today's rapidly growing and changing computer networks. These techniques, fuzzy reasoning and neural networks, when combined together provide a very effective routing algorithm for computer networks.
Computer simulation is employed to prove the new fuzzy routing algorithm outperforms the Shortest Path First (SPF) algorithm in most computer network situations. The benefits increase as the computer network migrates from a stable network to a more variable one. The advantages of applying this fuzzy routing algorithm are apparent when considering the dynamic nature of modern computer networks.<br>Ph. D.
16
Ayoub, Issa. "Multimodal Affective Computing Using Temporal Convolutional Neural Network and Deep Convolutional Neural Networks." Thesis, Université d'Ottawa / University of Ottawa, 2019. http://hdl.handle.net/10393/39337.
Testo completoAbstract (sommario):
Affective computing has gained significant attention from researchers in the last decade due to the wide variety of applications that can benefit from this technology. Often, researchers describe affect using emotional dimensions such as arousal and valence. Valence refers to the spectrum of negative to positive emotions while arousal determines the level of excitement. Describing emotions through continuous dimensions (e.g. valence and arousal) allows us to encode subtle and complex affects as opposed to discrete emotions, such as the basic six emotions: happy, anger, fear, disgust, sad and neutral.
Recognizing spontaneous and subtle emotions remains a challenging problem for computers. In our work, we employ two modalities of information: video and audio. Hence, we extract visual and audio features using deep neural network models. Given that emotions are time-dependent, we apply the Temporal Convolutional Neural Network (TCN) to model the variations in emotions. Additionally, we investigate an alternative model that combines a Convolutional Neural Network (CNN) and a Recurrent Neural Network (RNN). Given our inability to fit the latter deep model into the main memory, we divide the RNN into smaller segments and propose a scheme to back-propagate gradients across all segments. We configure the hyperparameters of all models using Gaussian processes to obtain a fair comparison between the proposed models. Our results show that TCN outperforms RNN for the recognition of the arousal and valence emotional dimensions. Therefore, we propose the adoption of TCN for emotion detection problems as a baseline method for future work. Our experimental results show that TCN outperforms all RNN based models yielding a concordance correlation coefficient of 0.7895 (vs. 0.7544) on valence and 0.8207 (vs. 0.7357) on arousal on the validation dataset of SEWA dataset for emotion prediction.
17
Roadknight, C. M. "Transparent neural network modelling." Thesis, Nottingham Trent University, 2000. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.314107.
Testo completo
18
Kulakov, Anton. "Multiprocessing neural network simulator." Thesis, University of Southampton, 2013. https://eprints.soton.ac.uk/348420/.
Testo completoAbstract (sommario):
Over the last few years tremendous progress has been made in neuroscience by employing simulation tools for investigating neural network behaviour. Many simulators have been created during last few decades, and their number and set of features continually grows due to persistent interest from groups of researchers and engineers. A simulation software that is able to simulate a large-scale neural network has been developed and presented in this work. Based on a highly abstract integrate-and-fire neuron model a clock-driven sequential simulator has been developed in C++. The created program is able to associate the input patterns with the output patterns. The novel biologically plausible learning mechanism uses Long Term Potentiation and Long Term Depression to change the strength of the connections between the neurons based on a global binary feedback. Later, the sequentially executed model has been extended to a multi-processor system, which executes the described learning algorithm using the event-driven technique on a parallel distributed framework, simulating a neural network asynchronously. This allows the simulation to manage larger scale neural networks being immune to processor failure and communication problems. The multi-processor neural network simulator has been created, the main benefit of which is the possibility to simulate large scale neural networks using high-parallel distributed computing. For that reason the design of the simulator has been implemented considering an efficient weight-adjusting algorithm and an efficient way for asynchronous local communication between processors.
19
Åström, Fredrik. "Neural Network on Compute Shader : Running and Training a Neural Network using GPGPU." Thesis, Blekinge Tekniska Högskola, Sektionen för datavetenskap och kommunikation, 2011. http://urn.kb.se/resolve?urn=urn:nbn:se:bth-2036.
Testo completoAbstract (sommario):
In this thesis I look into how one can train and run an artificial neural network using Compute Shader and what kind of performance can be expected. An artificial neural network is a computational model that is inspired by biological neural networks, e.g. a brain. Finding what kind of performance can be expected was done by creating an implementation that uses Compute Shader and then compare it to the FANN library, i.e. a fast artificial neural network library written in C. The conclusion is that you can improve performance by training an artificial neural network on the compute shader as long as you are using non-trivial datasets and neural network configurations.
20
Turk, Fethi. "Improvements To Neural Network Based Restoration In Optical Networks." Master's thesis, METU, 2008. http://etd.lib.metu.edu.tr/upload/12609491/index.pdf.
Testo completoAbstract (sommario):
Performance of neural network based restoration of optical networks is evaluated and a few possible improvements are proposed. Neural network based restoration is simulated with optical link capacities assigned by a new method. Two new improvement methods are developed to reduce the neural network size and the restoration time of severed optical connections. Cycle based restoration is suggested, which reduces the neural network structure by restoring the severed connections for each optical node, iteratively. Additionally, to reduce the restoration time, the necessary waiting time before the neuron outputs fire for the propagation delay over the network is computed and embedded in the control structure of the neural network. The improvement methods are evaluated by simulations in terms of
restorability, restoration time, network redundancy and average length of restoration paths for different failure cases and different security requirements.
21
Onyiagha, Chyke Godfrey. "Intelligent neural network access control of ATM network." Thesis, King's College London (University of London), 2002. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.271321.
Testo completo
22
Leija, Carlos Ivan. "An artificial neural network with reconfigurable interconnection network." To access this resource online via ProQuest Dissertations and Theses @ UTEP, 2008. http://0-proquest.umi.com.lib.utep.edu/login?COPT=REJTPTU0YmImSU5UPTAmVkVSPTI=&clientId=2515.
Testo completo
23
Emery, Robin. "Neural network-on-chip : A connection-centric reconfigurable platform for biological neural network models." Thesis, University of Newcastle Upon Tyne, 2010. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.531749.
Testo completo
24
McMichael, Lonny D. (Lonny Dean). "A Neural Network Configuration Compiler Based on the Adaptrode Neuronal Model." Thesis, University of North Texas, 1992. https://digital.library.unt.edu/ark:/67531/metadc501018/.
Testo completoAbstract (sommario):
A useful compiler has been designed that takes a high level neural network specification and constructs a low level configuration file explicitly specifying all network parameters and connections. The neural network model for which this compiler was designed is the adaptrode neuronal model, and the configuration file created can be used by the Adnet simulation engine to perform network experiments. The specification language is very flexible and provides a general framework from which almost any network wiring configuration may be created. While the compiler was created for the specialized adaptrode model, the wiring specification algorithms could also be used to specify the connections in other types of networks.
25
Pasca, Isabela Mona. "Neural network digital hardware implementation." Thesis, University of Ottawa (Canada), 2007. http://hdl.handle.net/10393/27902.
Testo completoAbstract (sommario):
This thesis presents a digital hardware implementation of an artificial neuron with learning ability using the QuartusII 5.1sp1 web edition software on Altera's University Program Development Board (UP2). The learning method implemented is neither backpropagation nor conjugate gradient, but the weight simultaneous perturbation. By combining this method with a pulse density system and using a Field Programmable Gate Array, an interesting artificial neuron hardware architecture is obtained. Finally, two applications of the neuron implementation are presented: an analog function and a digital function.
26
Siqueira, Ines. "Neural network-based cost estimating." Thesis, National Library of Canada = Bibliothèque nationale du Canada, 2000. http://www.collectionscanada.ca/obj/s4/f2/dsk1/tape3/PQDD_0019/MQ47816.pdf.
Testo completo
27
Bahr, Casey S. "Anne : another neural network emulator /." Full text open access at:, 1988. http://content.ohsu.edu/u?/etd,173.
Testo completo
28
Guerin, Olivier Cedrick. ""Neural network" based process monitoring." Thesis, Georgia Institute of Technology, 2002. http://hdl.handle.net/1853/16804.
Testo completo
29
Littlewort, G. C. "Neural network analysis and simulation." Thesis, University of Oxford, 1991. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.292677.
Testo completo
30
Messom, Christopher H. "Engineering reliable neural network systems." Thesis, Loughborough University, 1992. https://dspace.lboro.ac.uk/2134/14137.
Testo completoAbstract (sommario):
This thesis presents a study of neural network representation and behaviour. The study places neural networks in the context of designing reliable systems. Several new results on network size and topology are presented. Knowledge based training of neural networks is examined. This is essential for designing reliable neural systems in which the subsymbolic reasoning processes are well defined. Sandwich nodes are introduced and studied as atomic knowledge elements in neural networks. Two new network architectures are introduced, the Loughborough Net and the Loughborough Control Net. These make use of the parallelism inherent in sandwich node representations. The interpretation of neural network representations as logical transformations and rule systems are presented. An equivalence of the rule systems and neural network representation is proposed and discussed. This equivalence is required in order that the total behaviour of the neural network can be understood. A new methodology for designing reliable neural network systems making use of knowledge based training is proposed. This is used to present a general design methodology for the construction of. reliable neural network control systems using the Loughborough Control Net architecture. A case study is discussed where the methodology was applied to the design of an adhesive dispensing controller.
31
Ghafoori, Elyar. "Wavelet transform and neural network." Thesis, California State University, Long Beach, 2014. http://pqdtopen.proquest.com/#viewpdf?dispub=1527935.
Testo completoAbstract (sommario):
<p> Automatic and accurate detection of Atrial Fibrillation (AF) from the noninvasive ECG signal is imperative in Electrocardiography. AF is mainly reflected in the ECG signal with the absence of P wave and/or irregular RR intervals. Signal processing tools can assess such detailed changes in the ECG, leading to an accurate diagnosis of AF. The proposed method relies on proper noise filtering, Stationary Wavelet Transform, and signal Power Spectrum Estimation. A feature extraction technique and a Neural Network classifier have been employed to determine the presence and absence of the AF episodes. Implementation of the proposed method with 5-fold cross validation on more than 230 hours of ECG data from MIT-BIH arterial fibrillation annotated database demonstrated an accuracy of 93% in classification of the AF and normal ECG signals.</p>
32
Brierley, Matthew Joseph. "Neural network underlying snail feeding." Thesis, University of Sussex, 1993. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.239132.
Testo completo
33
Braun, Harald. "A neural network linking process." Thesis, City University London, 2003. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.410159.
Testo completo
34
Lam, David Tai-Yuen. "Optically implementable neural network algorithms." Thesis, University of Cambridge, 1990. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.385426.
Testo completo
35
Nag, Ronjon. "Neural network applications for finance." Thesis, Massachusetts Institute of Technology, 1991. http://hdl.handle.net/1721.1/13819.
Testo completoAbstract (sommario):
Thesis (M.S.)--Massachusetts Institute of Technology, Sloan School of Management, 1991.<br>Title as it appears in the M.I.T. Graduate List, Feb. 1991: Neural network applications in finance.<br>Includes bibliographical references (leaves 39-42).<br>by Ronjon Nag.<br>M.S.
36
Hadjiprocopis, Andreas. "Feed forward neural network entities." Thesis, City University London, 2000. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.340374.
Testo completo
37
Brown, Gavin. "Diversity in neural network ensembles." Thesis, University of Birmingham, 2004. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.410855.
Testo completo
38
Renals, Stephen John. "Speech and neural network dynamics." Thesis, University of Edinburgh, 1990. http://hdl.handle.net/1842/14271.
Testo completoAbstract (sommario):
This thesis is concerned with two principal issues. Firstly the radial basis functions (RBF) network is introduced and its properties related to other statistical and neural network classifiers. Results from a series of speech recognition experiments, using this network architecture, are reported. These experiments included a continuous speech recognition task with a 571 word lexicon. Secondly, a study of the dynamics of a simple recurrent network model is presented. This study was performed numerically, via a survey of network power spectra and a detailed investigation of the dynamics displayed by a particular network. Word and sentence recognition errors are reported for a continuous speech recognition system using RBF network phoneme modelling with Viterbi smoothing, using either a restricted grammar or no grammar whatsoever. In a cytopathology task domain the best RBF/Viterbi system produced first choice word errors of 6% and sentence errors of 14%, using a grammar of perplexity 6. This compares with word errors of 4% and sentence errors of 8% using the best CSTR hidden Markov model configuration. RBF networks were also used for a static vowel labelling task using hand-segmented vowels excised from continuous speech. Results were not worse than those obtained using statistical classifiers. The second part of this thesis is a computational study of the dynamics of a recurrent neural network model. Two investigations were undertaken. Firstly, a survey of network power spectra was used to map out the temporal activity of this network model (within a four dimensional parameter space) via summary statistics of the network power spectra. Secondly, the dynamics of a particular network were investigated. The dynamics were analysed using bifurcation diagrams, power spectra, the computation of Liapunov exponents and fractal dimensions and the plotting of 2-dimensional attractor projections. Complex dynamical behaviour was observed including Hopf bifurcations, the Ruell-Takens-Newhouse route to chaos with mode-locking at rational winding numbers, the period-doubling route to chaos and the presence of multiple coexisting attractors.
39
Armstrong, James R. "Boolean weightless neural network architectures." Thesis, University of Central Lancashire, 2011. http://clok.uclan.ac.uk/5325/.
Testo completoAbstract (sommario):
A collection of hardware weightless Boolean elements has been developed. These form fundamental building blocks which have particular pertinence to the field of weightless neural networks. They have also been shown to have merit in their own right for the design of robust architectures. A major element of this is a collection of weightless Boolean sum and threshold techniques. These are fundamental building blocks which can be used in weightless architectures particularly within the field of weightless neural networks. Included in these is the implementation of L-max also known as N point thresholding. These elements have been applied to design a Boolean weightless hardware version of Austin’s ADAM neural network. ADAM is further enhanced by the addition of a new learning paradigm, that of non-Hebbian Learning. This new method concentrates on the association of ‘dis-similarity’, believing this is as important as areas of similarity. Image processing using hardware weightless neural networks is investigated through simulation of digital filters using a Type 1 Neuroram neuro-filter. Simulations have been performed using MATLAB to compare the results to a conventional median filter. Type 1 Neuroram has been tested on an extended collection of noise types. The importance of the threshold has been examined and the effect of cascading both types of filters was examined. This research has led to the development of several novel weightless hardware elements that can be applied to image processing. These patented elements include a weightless thermocoder and two weightless median filters. These novel robust high speed weightless filters have been compared with conventional median filters. The robustness of these architectures has been investigated when subjected to accelerated ground based generated neutron radiation simulating the atmospheric radiation spectrum experienced at commercial avionic altitudes. A trial investigating the resilience of weightless hardware Boolean elements in comparison to standard weighted arithmetic logic is detailed, examining the effects on the operation of the function when implemented on hardware experiencing high energy neutron bombardment induced single event effects. Further weightless Boolean elements are detailed which contribute to the development of a weightless implementation of the traditionally weighted self ordered map.
40
Khazanova, Yekaterina. "Experiments with Neural Network Libraries." University of Cincinnati / OhioLINK, 2013. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1527607591612278.
Testo completo
41
Mosher, Stephen Glenn. "Neural Network Applications in Seismology." Thesis, Université d'Ottawa / University of Ottawa, 2021. http://hdl.handle.net/10393/42329.
Testo completoAbstract (sommario):
Neural networks are extremely versatile tools, as evidenced by their widespread adoption into many fields in the sciences and beyond, including the geosciences. In seismology neural networks have been primarily used to automatically detect and discriminate seismic signals within time-series data, as well as provide location estimates for their sources. However, as neural network research has significantly progressed over the past three decades, so too have its applications in seismology. Such applications now include earthquake early warning systems based on smartphone data collected from large numbers of users, the prediction of peak ground acceleration from earthquake source parameters, the efficient computation of synthetic seismograms, providing probabilistic estimates of solutions to geophysical inverse problems, and many others. This thesis contains three components, each of which explore novel uses of neural networks in seismology. In the first component, a previously established earthquake detection and location method is supplemented with a neural network in order to automate the detection process. The detection procedure is then applied to a large volume of seismic data. In addition to automating the detection process, the neural network removes the need for several user-defined thresholds, subjective criteria otherwise necessary for the method. In the second component, a novel approach is developed for inverting seafloor compliance data recorded by ocean-bottom seismometers for the shallow shear-wave velocity structure of oceanic tectonic plates. The approach makes use of mixture density networks, a type of neural network designed to provide probabilistic estimates of solutions to inverse problems, something that standard neural networks are incapable of. In the final component of this thesis, the mixture density network approach to compliance inversion is applied to a group of ocean-bottom seismometers deployed along the continental shelf of the Cascadia Subduction Zone in order to investigate shelf sediment properties.
42
Rimer, Michael Edwin. "Improving Neural Network Classification Training." Diss., CLICK HERE for online access, 2007. http://contentdm.lib.byu.edu/ETD/image/etd2094.pdf.
Testo completo
43
Kabore, Raogo. "Hybrid deep neural network anomaly detection system for SCADA networks." Thesis, Ecole nationale supérieure Mines-Télécom Atlantique Bretagne Pays de la Loire, 2020. http://www.theses.fr/2020IMTA0190.
Testo completoAbstract (sommario):
Les systèmes SCADA sont de plus en plus ciblés par les cyberattaques en raison de nombreuses vulnérabilités dans le matériel, les logiciels, les protocoles et la pile de communication. Ces systèmes utilisent aujourd'hui du matériel, des logiciels, des systèmes d'exploitation et des protocoles standard. De plus, les systèmes SCADA qui étaient auparavant isolés sont désormais interconnectés aux réseaux d'entreprise et à Internet, élargissant ainsi la surface d'attaque. Dans cette thèse, nous utilisons une approche deep learning pour proposer un réseau de neurones profonds hybride efficace pour la détection d'anomalies dans les systèmes SCADA. Les principales caractéristiques des données SCADA sont apprises de manière automatique et non supervisée, puis transmises à un classificateur supervisé afin de déterminer si ces données sont normales ou anormales, c'est-à-dire s'il y a une cyber-attaque ou non. Par la suite, en réponse au défi dû au temps d’entraînement élevé des modèles deep learning, nous avons proposé une approche distribuée de notre système de détection d'anomalies afin de réduire le temps d’entraînement de notre modèle<br>SCADA systems are more and more targeted by cyber-attacks because of many vulnerabilities inhardware, software, protocols and the communication stack. Those systems nowadays use standard hardware, software, operating systems and protocols. Furthermore, SCADA systems which used to be air-gaped are now interconnected to corporate networks and to the Internet, widening the attack surface.In this thesis, we are using a deep learning approach to propose an efficient hybrid deep neural network for anomaly detection in SCADA systems. The salient features of SCADA data are automatically and unsupervisingly learnt, and then fed to a supervised classifier in order to dertermine if those data are normal or abnormal, i.e if there is a cyber-attack or not. Afterwards, as a response to the challenge caused by high training time of deep learning models, we proposed a distributed approach of our anomaly detection system in order lo lessen the training time of our model
44
Friji, Hamdi. "Graph neural network-based intrusion detection for secure edge networks." Electronic Thesis or Diss., Institut polytechnique de Paris, 2024. http://www.theses.fr/2024IPPAS030.
Testo completoAbstract (sommario):
Face à l'escalade de la complexité et à la fréquence des cyberattaques, cette thèse propose des approches innovantes pour la détection d'intrusion dans les réseaux, en exploitant les capacités avancées des réseaux de neurones en graphe (Graph Neural Networks, GNNs) et de nouvelles représentations sous forme de graphes. Nous commençons par une analyse critique des jeux de données et des représentations de réseaux actuels, en abordant des questions clés sur leur efficacité. Nous introduisons une nouvelle représentation des flux de communication sous forme de graphes, offrant une plus grande robustesse face aux manipulations de type attaques adversariales (adversarial attacks).Nous présentons ensuite l'un des premiers systèmes de détection d'intrusion utilisant notre représentation en graphe basée sur les GNNs. Ce système permet d'évaluer les comportements malveillants en capturant des motifs complexes souvent ignorés par les méthodes traditionnelles. Les résultats montrent que notre approche surpasse largement les solutions existantes basées sur l'apprentissage automatique et les GNNs en termes de précision et de robustesse.Pour relever les défis de la scalabilité et de l'efficacité dans les environnements à grande échelle, nous introduisons G-DEMIS (« Graph-based DEcentralized Multi-agent Intrusion detection System »), un système multi-agent décentralisé exploitant les GNNs pour une détection rapide des activités malveillantes. En agrégeant les informations locales à travers le réseau, G-DEMIS améliore la détection en temps réel tout en réduisant la consommation d'énergie de 58 % et le temps de réponse de 17,13 % par rapport aux approches centralisées. Enfin, nous proposons un algorithme innovant pour tracer les chemins de propagation des attaques, aidant à identifier les machines compromises dans le contexte des menaces persistantes avancées. Les travaux de cette thèse permettent non seulement de faire progresser l'état de l'art en matière de détection d'intrusions, mais ouvrent également la voie à de nouvelles avancées en cybersécurité<br>In light of the escalating complexity and frequency of cyberattacks, this thesis presents innovative approachs to network intrusion detection that leverages the advanced capabilities of Graph Neural Networks (GNNs) and novel graph-based representations. To lay the foundation for our research, we first conduct a critical review of existing intrusion detection datasets and network representations, focusing on their effectiveness in addressing key research challenges. This thesis presents our insights and analysis of two widely used datasets: ToN IoT and CICIDS 2017, highlighting their strengths and limitations. Our approach introduces a new flow-based graph representation of communication flows, which enhances existing solutions by increasing robustness against adversarial attacks.First, we present one of the pioneering GNN-based intrusion detection systems, which utilizes our graph representation and GNN algorithms to compute maliciousness scores. This system captures complex relational patterns that traditional methods often overlook. Our findings demonstrate that this framework significantly outperforms the current state-of-the-art machine learning and GNN-based solutions in terms of both accuracy and robustness. Additionally, we propose a three-stage intrusion detection system inspired by the Lockheed Martin cyber kill chain, designed to detect advanced multi-step attacks. This system achieved an average F1-score of 94% on the ToN IoT dataset, surpassing traditional random forest models and demonstrating its effectiveness for real-world applications.To address scalability and efficiency challenges in large-scale environments, we introduce G-DEMIS, a Graph-based DEcentralized Multi-agent Intrusion detection System that enhances the use of GNNs for a fast detection of malicious activities. G-DEMIS employs a collaborative approach in which multiple agents monitor different network segments, aggregating local graph information to form a comprehensive view of the network. This framework not only enhances real-time detection capabilities but also reduces energy consumption by 58.08% and detection time by 17.13% compared to centralized models.Finally, we tackle the challenge of defending against Advanced Persistent Threats (APTs) by proposing a novel algorithm for reconstructing attack propagation paths. This algorithm assists engineers in identifying compromised machines following an APT attack by detecting and analyzing anomalous behaviors in the network, tracing the progression of the attack, and providing a detailed understanding of the attack paths.This thesis not only advances the current state of intrusion detection but also lays the foundation for future innovations in cybersecurity
45
Keisala, Simon. "Designing an Artificial Neural Network for state evaluation in Arimaa : Using a Convolutional Neural Network." Thesis, Linköpings universitet, Artificiell intelligens och integrerade datorsystem, 2017. http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-143188.
Testo completoAbstract (sommario):
Agents being able to play board games such as Tic Tac Toe, Chess, Go and Arimaa has been, and still is, a major difficulty in Artificial Intelligence. For the mentioned board games, there is a certain amount of legal moves a player can do in a specific board state. Tic Tac Toe have in average around 4-5 legal moves, with a total amount of 255168 possible games. Both Chess, Go and Arimaa have an increased amount of possible legal moves to do, and an almost infinite amount of possible games, making it impossible to have complete knowledge of the outcome. This thesis work have created various Neural Networks, with the purpose of evaluating the likelihood of winning a game given a certain board state. An improved evaluation function would compensate for the inability of doing a deeper tree search in Arimaa, and the anticipation is to compete on equal skills against another well-performing agent (meijin) having one less search depth. The results shows great potential. From a mere one hundred games against meijin, the network manages to separate good from bad positions, and after another one hundred games able to beat meijin with equal search depth. It seems promising that by improving the training and by testing different sizes for the neural network that a neural network could win even with one less search depth. The huge branching factor of Arimaa makes such an improvement of the evaluation beneficial, even if the evaluation would be 10 000 times more slow.
46
Teng, Peiyuan. "Tensor network and neural network methods in physical systems." The Ohio State University, 2018. http://rave.ohiolink.edu/etdc/view?acc_num=osu1524836522115804.
Testo completo
47
Brouwer, Roelof K. "Pattern recognition using a generalised discrete Hopfield network." Thesis, University of Warwick, 1995. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.307974.
Testo completo
48
Sazli, Murat Husnu Işık Can. "Neural network applications to turbo decoding." Related Electronic Resource: Current Research at SU : database of SU dissertations, recent titles available full text, 2003. http://wwwlib.umi.com/cr/syr/main.
Testo completo
49
Li, Keh-Tsong, and 李克聰. "Neural Network combined with Genetic Algorithm-Evolutionary Neural Network." Thesis, 1999. http://ndltd.ncl.edu.tw/handle/37236508646662658444.
Testo completoAbstract (sommario):
碩士<br>國立交通大學<br>電機與控制工程系<br>87<br>This thesis presents a Real-Coded Rank-Based Genetic Algorithm (RCRBGA), which is represented by a chromosome containing parameters in floating-point. The use of rank-based fitness increases the population diversity. The offspring are generated by the rank-based reproduction, real parametric crossover and mutation in the evolving process. Besides, an Evolutionary Neural Network (ENN) which combines RCRBGA and Back-Propagation (BP) is introduced. ENN applies the learning concept to the evolution process, like the behavior of human beings. It not only improves the disadvantage of easily slumping in to local minima of BP but also overcomes the defect of genetic algorithm, which can't efficiently converge to minima. Finally, the search ability of RCRBGA is demonstrated by an example, linear state-feedback controller via pole-assignment method. In addition, ENN applies to a classifying problem of the modified XOR to show its advantage.
50
"Restoration network design and neural network." Chinese University of Hong Kong, 1992. http://library.cuhk.edu.hk/record=b5887053.
Testo completoAbstract (sommario):
by Leung Lee.<br>Thesis (M.Sc.)--Chinese University of Hong Kong, 1992.<br>Includes bibliographical references.<br>Chapter SECTION 1. --- Introduction --- p.1<br>Chapter SECTION 2. --- Formulation of Problem --- p.2<br>Chapter 2.1 --- Problem Identification --- p.2<br>Chapter 2.2 --- Network Planning Parameters and Assumptions --- p.3<br>Chapter 2.3 --- Neural Network Model Transformation --- p.5<br>Chapter 2.4 --- Algorithm and Implementation --- p.12<br>Chapter SECTION 3. --- Simulation Results --- p.15<br>Chapter 3.1 --- All Link Costs Are Same or Nearly the Same --- p.17<br>Chapter 3.2 --- Fluctuated Cost in One or Two Fibre Paths --- p.18<br>Chapter 3.3 --- Sudden Traffic Demand Change in Last Season --- p.19<br>Chapter SECTION 4. --- Discussion --- p.20<br>Chapter SECTION 5. --- Conclusion --- p.26<br>GLOSSARY OF TERMS --- p.27<br>BIBLIOGRAPHY --- p.29<br>APPENDIX --- p.A1<br>Chapter A --- Simulation Results --- p.A1<br>Chapter B --- ANN Traffic Routing Example --- p.B1