To see the other types of publications on this topic, follow the link: Neural time series.
Dissertations / Theses on the topic 'Neural time series'
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 'Neural time series.'
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
Kajitani, Yoshio. "Forecasting time series with neural nets." Thesis, National Library of Canada = Bibliothèque nationale du Canada, 1999. http://www.collectionscanada.ca/obj/s4/f2/dsk2/ftp03/MQ39836.pdf.
Full text
2
Andreux, Mathieu. "Foveal autoregressive neural time-series modeling." Electronic Thesis or Diss., Paris Sciences et Lettres (ComUE), 2018. http://www.theses.fr/2018PSLEE073.
Full textAbstract:
Cette thèse s'intéresse à la modélisation non-supervisée de séries temporelles univariées. Nous abordons tout d'abord le problème de prédiction linéaire des valeurs futures séries temporelles gaussiennes sous hypothèse de longues dépendances, qui nécessitent de tenir compte d'un large passé. Nous introduisons une famille d'ondelettes fovéales et causales qui projettent les valeurs passées sur un sous-espace adapté au problème, réduisant ainsi la variance des estimateurs associés. Dans un deuxième temps, nous cherchons sous quelles conditions les prédicteurs non-linéaires sont plus performants que les méthodes linéaires. Les séries temporelles admettant une représentation parcimonieuse en temps-fréquence, comme celles issues de l'audio, réunissent ces conditions, et nous proposons un algorithme de prédiction utilisant une telle représentation. Le dernier problème que nous étudions est la synthèse de signaux audios. Nous proposons une nouvelle méthode de génération reposant sur un réseau de neurones convolutionnel profond, avec une architecture encodeur-décodeur, qui permet de synthétiser de nouveaux signaux réalistes. Contrairement à l'état de l'art, nous exploitons explicitement les propriétés temps-fréquence des sons pour définir un encodeur avec la transformée en scattering, tandis que le décodeur est entraîné pour résoudre un problème inverse dans une métrique adaptéeThis dissertation studies unsupervised time-series modelling. We first focus on the problem of linearly predicting future values of a time-series under the assumption of long-range dependencies, which requires to take into account a large past. We introduce a family of causal and foveal wavelets which project past values on a subspace which is adapted to the problem, thereby reducing the variance of the associated estimators. We then investigate under which conditions non-linear predictors exhibit better performances than linear ones. Time-series which admit a sparse time-frequency representation, such as audio ones, satisfy those requirements, and we propose a prediction algorithm using such a representation. The last problem we tackle is audio time-series synthesis. We propose a new generation method relying on a deep convolutional neural network, with an encoder-decoder architecture, which allows to synthesize new realistic signals. Contrary to state-of-the-art methods, we explicitly use time-frequency properties of sounds to define an encoder with the scattering transform, while the decoder is trained to solve an inverse problem in an adapted metric
3
Bonato, Tommaso. "Time Series Predictions With Recurrent Neural Networks." Bachelor's thesis, Alma Mater Studiorum - Università di Bologna, 2018.
Find full textAbstract:
L'obiettivo principale di questa tesi è studiare come gli algoritmi di apprendimento automatico (machine learning in inglese) e in particolare le reti neurali LSTM (Long Short Term Memory) possano essere utilizzati per prevedere i valori futuri di una serie storica regolare come, per esempio, le funzioni seno e coseno. Una serie storica è definita come una sequenza di osservazioni s_t ordinate nel tempo.
Inoltre cercheremo di applicare gli stessi principi per prevedere i valori di una serie storica prodotta utilizzando i dati di vendita di un prodotto cosmetico durante un periodo di tre anni.
Prima di arrivare alla parte pratica di questa tesi è necessario introdurre alcuni concetti fondamentali che saranno necessari per sviluppare l'architettura e il codice del nostro modello.
Sia nell'introduzione teorica che nella parte pratica l'attenzione sarà focalizzata sull'uso di RNN (Recurrent Neural Network o Rete Neurale Ricorrente) poiché sono le reti neurali più adatte a questo tipo di problema. Un particolare tipo di RNN, chiamato Long Short Term Memory (LSTM), sarà soggetto dello studio principale di questa tesi e verrà presentata e utilizzata anche una delle sue varianti chiamata Gated Recurrent Unit (GRU).
Questa tesi, in conclusione, conferma che LSTM e GRU sono il miglior tipo di rete neurale per le previsioni di serie temporali. Nell'ultima parte analizzeremo le differenze tra l'utilizzo di una CPU e una GPU durante la fase di training della rete neurale.
4
Brax, Christoffer. "Recurrent neural networks for time-series prediction." Thesis, University of Skövde, Department of Computer Science, 2000. http://urn.kb.se/resolve?urn=urn:nbn:se:his:diva-480.
Full textAbstract:
Recurrent neural networks have been used for time-series prediction with good results. In this dissertation recurrent neural networks are compared with time-delayed feed forward networks, feed forward networks and linear regression models on a prediction task. The data used in all experiments is real-world sales data containing two kinds of segments: campaign segments and non-campaign segments. The task is to make predictions of sales under campaigns. It is evaluated if more accurate predictions can be made when only using the campaign segments of the data.
Throughout the entire project a knowledge discovery process, identified in the literature has been used to give a structured work-process. The results show that the recurrent network is not better than the other evaluated algorithms, in fact, the time-delayed feed forward neural network showed to give the best predictions. The results also show that more accurate predictions could be made when only using information from campaign segments.
5
ABELEM, ANTONIO JORGE GOMES. "ARTIFICIAL NEURAL NETWORKS IN TIME SERIES FORECASTING." PONTIFÍCIA UNIVERSIDADE CATÓLICA DO RIO DE JANEIRO, 1994. http://www.maxwell.vrac.puc-rio.br/Busca_etds.php?strSecao=resultado&nrSeq=8489@1.
Full textAbstract:
COORDENAÇÃO DE APERFEIÇOAMENTO DO PESSOAL DE ENSINO SUPERIOREsta dissertação investiga a utilização de Redes Neurais Artificiais (RNAs) na previsão de séries temporais, em particular de séries financeiras, consideradas uma classe especial de séries temporais, caracteristicamente ruídos e sem periodicidade aparente. O trabalho envolve quatro partes principais: um estudo sobre redes neurais artificiais e séries temporais; a modelagem das RNAs para previsão de séries temporais; o desenvolvimento de um ambiente de simulação; e o estudo de caso. No estudo sobre Redes Neurais Artificiais e séries temporais fez-se um levantamento preliminar das aplicações de RNAs na previsão de séries. Constatou-se a predominância do uso do algoritmos de retropropagação do erro para o treinamento das redes, bem como dos modelos estatísticos de regressão, de médias móveis e de alisamento exponencial nas comparações com os resultados da rede. Na modelagem das RNAs de retropropagação do erro considerou-se três fatores determinantes no desempenho da rede: convergência, generalização e escalabilidade. Para o controle destes fatores usou-se mecanismos como; escolha da função de ativação dos neurônios - sigmóide ou tangente hiperbólica; escolha da função erro - MSE (Mean Square Error) ou MAD (Mean Absolutd Deviation); e escolha dos parâmetros de controle do gradiente descendente e do temapo de treinamento - taxa de aprendizado e termo de momento. Por fim, definiu-se a arquitetura da rede em função da técnica utilizada para a identificação de regularidades na série (windowing) e da otimização dos fatores indicadores de desempenho da rede. O ambiente de simulação foi desenvolvido em linguagem C e contém 3.600 linhas de códigos divididas em três módulos principais: interface com o usuário, simulação e funções secundárias. O módulo de interface com o usuário é responsável pela configuração e parametrização da rede, como também pela visualização gráfica dos resultados; módulo de simulação executa as fases de treinamento e testes das RNAs; o módulo de funções secundárias cuida do pré/pós-processamento dos dados, da manipulação de arquivos e dos cálculos dos métodos de avaliação empregados. No estudo de caso, as RNAs foram modeladas para fazer previsões da série do preço do ouro no mercado internacional. Foram feitas previsões univariadas single e multi-step e previsões multivariadas utilizando taxas de câmbio de moedas estrangeiras. Os métodos utilizandos para a avaliação do desempenho da rede foram: coeficiente U de Theil, MSE (Mean Square Error), NRMSE (Normalized Root Mean Square Error), POCID (Percentage Of Change In Direction), scattergram e comparação gráfica. Os resultados obtidos, além de avaliados com os métodos acima, foram comparados com o modelo de Box-Jenkins e comprovaram a superioridade das RNAs no tratamento de dados não-lineares e altamente ruidosos.
This dissertation investigates the use of Artificial Neural Nerworks (ANNs) in time series forecastig, especially financial time series, which are typically noisy and with no apparent periodicity. The dissertation covers four major parts: the study of Artificial Neural Networks and time series; the desing of ANNs applied to time series forecasting; the development of a simulation enironment; and a case study. The first part of this dissertation involved the study of Artficial Neural Netwrks and time series theory, resulting in an overview of ANNs utilization in time series forecasting. This overview confirmed the predominance of Backpropagations as the training algorithm, as well as the employment of statistical models, such as regression and moving average, for the Neural Network evaluation. In the design of ANNS, three performance measures were considered: covergence, generalization and scalability. To control these parameters, the following methods were applied: choice of activation function - sigmoid or hiperbolic tangent; choice of cost function - MSE (Mean Square Error) or MAD (Mean Absolute Deviation); choise of parameteres for controlling the gradiente descendent and learning times - the learning rate and momentum term; and network architecture. The simulation environment was developed in C language, with 3,600 lines of code distributed in three main modules: the user interface, the simulaton and the support functions modules. The user interface module is responsaible for the network configuration and for the graphical visualization. The simulation module performs the training and testing of ANNs. The support functions module takes care of the pre and pos processin, the files management and the metrics calculation. The case study concerned with the designing of an ANN to forescast the gold price in the international market. Two kinds of prediction were used: univariate - single and multi-step, and multivariate. The metrics used to evaluate the ANN performance were: U of Theil`s coeficient, MSE (Mean Square Error), NRMSE (Normalized Mean Saquare Error), POCID (Percentage Of Cnage In Direction), scattergram and graphical comparison. The results were also comapred with the Box-Jenkins model, confirming the superiority of ANN in handling non-linear and noisy data.
6
ZANDONADE, ELIANA. "USING NEURAL NETWORK IN TIME SERIES FORECASTING." PONTIFÍCIA UNIVERSIDADE CATÓLICA DO RIO DE JANEIRO, 1993. http://www.maxwell.vrac.puc-rio.br/Busca_etds.php?strSecao=resultado&nrSeq=8641@1.
Full textAbstract:
COORDENAÇÃO DE APERFEIÇOAMENTO DO PESSOAL DE ENSINO SUPERIOREste trabalho associa previsão de Séries Temporais a uma nova metodologia de processamento de informação: REDE NEURAL. Usaremos o modelo de Retropropagação, que consiste em uma Rede Neural multicamada com as unidades conectadas apenas com a unidades conectadas apenas com as unidades da camada subseqüente e com a informação passando em uma única direção. Aplicaremos o modelo de retropropagação na análise de quatro séries temporais: uma série ruidosa. Uma série com tendência, uma série sazonal e uma série de Consumo de Energia Elétrica da cidade de Uruguaiana, RS. Os resultados obtidos serão comparados com os modelos ARIMA de Box e Jenkins e um modelo com intervenção
This work join the Times-Séries Forecasting to a new information processing metodoligy: NEURAL NETWORK. We will use the Back-Propagation model, that consist in an arquitecture of a feed-forward network with hidden layers. We will apply the Back-Propagation model in an analysis to four times series: a noisy series, a series with trend, a seasonal series and an electrical energy consuption series of Uruguaiana, RS. The results will be compare with the Box and jenkins´ ARIMA models and a model with intervention.
7
MELLEM, MARCELO TOURASSE NASSIM. "AUTOREGRESSIVE-NEURAL HYBRID MODELS FOR TIME SERIES." PONTIFÍCIA UNIVERSIDADE CATÓLICA DO RIO DE JANEIRO, 1997. http://www.maxwell.vrac.puc-rio.br/Busca_etds.php?strSecao=resultado&nrSeq=14541@1.
Full textAbstract:
COORDENAÇÃO DE APERFEIÇOAMENTO DO PESSOAL DE ENSINO SUPERIOREste trabalho apresenta um modelo linear por partes chamado de modelo ARN. Trata-se de uma estrutura híbrida que envolve modelos autoregressivos e redes neurais. Este modelo é comparado com o modelo AR de coeficientes fixos e com a rede neural estática aplicada à previsão. Os resultados mostram que o ARN consegue identificar a estrutura não-linear dos dados simulados e que na maioria dos casos ele possui melhor habilidade preditiva do que os modelos supracitados.
In this thesis we develop a piece-wise linear model named ARN model. Our model has a hybrid structure which combines autoregressive models and neural networks. We compare our model to the fixed-coefficient AR model and to the prediction static neural network. Our results show that ARN is able to find the non-linear structure of simulated data and in most cases it performs better than the methods mentioned above.
8
Han, Ying. "Analysing time series using artificial neural networks." Thesis, University of the West of Scotland, 2003. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.398318.
Full text
9
Rana, Md Mashud. "Energy time series prediction." Thesis, The University of Sydney, 2014. http://hdl.handle.net/2123/11745.
Full textAbstract:
Reliable operations and economical utilization of power systems require electricity load forecasting at a wide range of forecasting horizons. The objective of this thesis is two-fold: developing accurate prediction models for electricity load forecasting, and quantifying the load forecasting uncertainty. At first, we consider the task of feature selection for electricity load forecasting. We propose a two-step approach - identifying a set of candidate features based on the data characteristics and then selecting a subset of them using four different methods. We evaluate the performance of these methods using state-of-the-art prediction algorithms. The results show that all feature selection methods are able to identify small subsets of highly relevant features for electricity load forecasting. We then present a generic approach for very short term electricity load forecasting. It combines multilevel wavelet packet transform, a non-linear feature selection method based on mutual information, and machine learning prediction algorithms. The evaluation shows that the proposed approach is robust and outperforms several non-wavelet based approaches. We also propose a novel approach for forecasting the daily load profile. The proposed approach uses mutual information for feature selection and an ensemble of neural networks for building a prediction model. The evaluation using two years of electricity load data for Australia, Portugal and Spain shows that it provides accurate predictions. Finally, we present LUBEX, a neural networks based approach for forecasting prediction intervals to quantify the uncertainty associated with electricity load prediction. LUBEX extends an existing method (LUBE) by including an advanced feature selection method and using an ensemble of neural networks. A comprehensive evaluation using 24 different case studies shows that LUBEX is able to generate high quality prediction intervals.
10
SOTO, CLAVER PARI. "TEMPORAL NEURAL NETWORKS FOR TREATING TIME VARIANT SERIES." PONTIFÍCIA UNIVERSIDADE CATÓLICA DO RIO DE JANEIRO, 1999. http://www.maxwell.vrac.puc-rio.br/Busca_etds.php?strSecao=resultado&nrSeq=7437@1.
Full textAbstract:
COORDENAÇÃO DE APERFEIÇOAMENTO DO PESSOAL DE ENSINO SUPERIORAs RNA Temporais, em função de sua estrutura, consideram o tempo na sua operação, incorporando memória de curto prazo distribuída na rede em todos os neurônios escondidos e em alguns dos casos nos neurônios de saída. Esta classe de redes é utilizada para representar melhor a natureza temporal dos sistemas dinâmicos. Em contraste, a RNA estática tem uma estrutura apropriada para tarefas de reconhecimento de padrões, classificação e outras de natureza estática ou estacionária tendo sido utilizada com sucesso em diversas aplicações. O objetivo desta tese, portanto foi estudar a teoria e avaliar o desempenho das Redes Neurais Temporais em comparação com as Redes Neurais Estáticas, em aplicações de sistemas dinâmicos. O desenvolvimento desta pesquisa envolveu 3 etapas principais: pesquisa bibliográfica das metodologias desenvolvidas para RNA Temporais; seleção e implementação de modelos para a avaliação destas redes; e estudo de casos. A pesquisa bibliográfica permitiu compila e classificar os principais trabalhos sobre RNA Temporais. Tipicamente, estas redes podem ser classificadas em dois grupos: Redes com Atraso no Tempo e Redes Recorrentes. Para a análise de desempenho, selecionou-se uma redee de cada grupo para implementação. Do primeiro grupo foi selecionada a Rede FIR, onde as sinapses são filtros FIR (Finite-duration Impulse Response) que representam a natureza temporal do problema. A rede FIR foi selecionada por englobar praticamente, todos os outros métodos de sua classe e apresentar um modelo matemático mais formal. Do segundo grupo, considerou-se a rede recorrente de Elman que apresenta realimentação global de cada um dos neurônios escondidos para todos eles. No estudo de casos testou-se o desempenho das redes selecionadas em duas linhas de aplicação: previsão de séries temporais e processamento digital de sinais. No caso de previsão de séries temporais, foram utilizadas séries de consumo de energia elétrica, comparando-se os resultados com os encontrados na literatura a partir de métodos de Holt-Winters, Box & Jenkins e RNA estáticas. No caso da aplicação das RNA em processamento digital de sinais, utilizou-se a filtragem de ruído em sinais de voz onde foram feitas comparações com os resultados apresentados pelo filtro neural convencional, que é uma rede feed-forward multicamada com o algoritmo de retropropagação para o aprendizado. Este trabalho demonstrou na prática que as RNA temporais conseguem capturar as características dos processos temporais de forma mais eficiente que as RNA Estatísticas e outros métodos tradicionais, podendo aprender diretamente o comportamento não estacionário das séries temporais. Os resultados demonstraram que a rede neural FIR e a rede Elman aprendem melhor a complexidade dos sinais de voz.
This dissertation investigates the development of Artificial Neural Network (ANN) in the solution of problems where the patterns presented to the network have a temporary relationship to each other, such as time series forecast and voice processing. Temporary ANN considers the time in its operation, incorporating memory of short period distributed in the network in all the hidden neurons and in the output neurons in some cases. This class of network in better used to represent the temporary nature of the dynamic systems. In contrast, Static ANN has a structure adapted for tasks of pattern recognition, classification and another static or stationary problems, achieving great success in several applications. Considered an universal approximator, Static ANN has also been used in applications of dynamic systems, through some artifices in the input of the network and through statistical data pre- processings. The objective of this work is, therefore to study the theory and evaluate the performance of Temporal ANN, in comparison with Static ANN, in applications of dynamics systems. The development of this research involved 3 main stages: bibliographical research of the methodologies developed for Temporal ANN; selection and implementation of the models for the evaluation of these networks; and case studies. The bibliographical research allowed to compile and to classify the main on Temporal ANN, Typically, these network was selected, where the synapses are filters FIR (Finite-duration Impulse Response) that represent the temporary nature of the problem. The FIR network has been selected since it includes practically all other methods of its class, presenting a more formal mathematical model. On the second group, the Elman recurrent network was considered, that presents global feedback of each neuron in the hidden layer to all other neurons in this layer. In the case studies the network selected have been tested in two application: forecast of time series and digital signal processing. In the case of forecast, result of electric energy consumption time series prediction were compared with the result found in the literature such as Holt-Winters, Box & Jenkins and Static ANN methods. In the case of the application of processing where the comparisons were made with the results presented by the standard neural filter, made of a multilayer feed-forward network with the back propagation learning algorithm. This work showed in practice that Temporal ANN captures the characteristics of the temporary processes in a more efficient way that Static ANN and other methods, being able to learn the non stationary behavior of the temporary series directly. The results showed that the FIR neural network and de Elman network learned better the complexity of the voice signals.
11
Lombardi, Alessandro. "Multiple time series forecasting with Graph Neural Networks." Master's thesis, Alma Mater Studiorum - Università di Bologna, 2021. http://amslaurea.unibo.it/24729/.
Full textAbstract:
Time series forecasting aims to predict future values to support organizations making strategic decisions. This problem has been studied for decades due to its relevance in almost all industries and areas, ranging from financial data to product demand. Recently, modern solutions based on deep learning have gained popularity among academia and industry, mainly due to the necessity to automatize the forecasting of multiple time series and exploit external explanatory variables. Considering the recent successes of Graph Neural Networks (GNNs) in modelling graph data, this study extends previous works based on time series forecasting from visibility graphs. In particular, in the first direction, the link prediction task, targeted by local random walks in the related work, is resolved by custom GNNs. In the second direction, a new strategy based on graph regression using GNNs is proposed to learn graph representations able to combine hidden historical patterns and external features. The M5 competition dataset is used to compare the proposed models to the related work and other traditional and machine learning benchmarks. Final results show promising performances on various higher levels of the M5 competition and delineate multiple limitations from the related work.
12
Liotta, Andrea <1987>. "evolutionary wavelet neural networks for time series forecasting." Master's Degree Thesis, Università Ca' Foscari Venezia, 2013. http://hdl.handle.net/10579/3447.
Full text
13
Dodd, Tony. "Prior knowledge for time series modelling." Thesis, University of Southampton, 2000. https://eprints.soton.ac.uk/254110/.
Full text
14
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.
15
Sarishvili, Alex. "Neural network based lag selection for multivariate time series." [S.l. : s.n.], 2002. http://deposit.ddb.de/cgi-bin/dokserv?idn=966609611.
Full text
16
Andoh, Charles. "Risk analysis of financial time series using neural networks." [S.l.] : [s.n.], 2005. http://deposit.ddb.de/cgi-bin/dokserv?idn=974193461.
Full text
17
Aamodt, Rune. "Using Artificial Neural Networks To Forecast Financial Time Series." Thesis, Norwegian University of Science and Technology, Department of Computer and Information Science, 2010. http://urn.kb.se/resolve?urn=urn:nbn:no:ntnu:diva-10907.
Full textAbstract:
This thesis investigates the application of artificial neural networks (ANNs) for forecasting financial time series (e.g. stock prices).The theory of technical analysis dictates that there are repeating patterns that occur in the historic prices of stocks, and that identifying these patterns can be of help in forecasting future price developments. A system was therefore developed which contains several ``agents'', each producing recommendations on the stock price based on some aspect of technical analysis theory. It was then tested if ANNs, using these recommendations as inputs, could be trained to forecast stock price fluctuations with some degree of precision and reliability.The predictions of the ANNs were evaluated by calculating the Pearson correlation between the predicted and actual price changes, and the ``hit rate'' (how often the predicted and the actual change had the same sign). Although somewhat mixed overall, the empirical results seem to indicate that at least some of the ANNs were able to learn enough useful features to have significant predictive power. Tests were performed with ANNs forecasting over different time frames, including intraday. The predictive performance was seen to decline on the shorter time scales.
18
Chan, Lipton. "Time-series prediction using evolutionary lateral-delay neural networks." Thesis, University of Glasgow, 2003. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.272850.
Full text
19
Ghazali, Rozaida. "Higher order neural networks for financial time series prediction." Thesis, Liverpool John Moores University, 2007. http://researchonline.ljmu.ac.uk/5879/.
Full textAbstract:
Neural networks have been shown to be a promising tool for forecasting financial times series. Numerous research and applications of neural networks in business have proven their advantage in relation to classical methods that do not include artificial intelligence. What makes this particular use of neural networks so attractive to financial analysts and traders is the fact that governments and companies benefit from it to make decisions on investment and trading. However, when the number of inputs to the model and the number of training examples becomes extremely large, the training procedure for ordinary neural network architectures becomes tremendously slow and unduly tedious. To overcome such time-consuming operations, this research work focuses on using various Higher Order Neural Networks (HONNs) which have a single layer of learnable weights, therefore reducing the networks' complexity. In order to predict the upcoming trends of univariate financial time series signals, three HONNs models; the Pi-Sigma Neural Network, the Functional Link Neural Network, and the Ridge Polynomial Neural Network were used, as well as the Multilayer Perceptron. Furthermore, a novel neural network architecture which comprises of a feedback connection in addition to the feedforward Ridge Polynomial Neural Network was constructed. The proposed network combines the properties of both higher order and recurrent neural networks, and is called Dynamic Ridge Polynomial Neural Network (DRPNN). Extensive simulations covering ten financial time series were performed. The forecasting performance of various feedforward HONNs models, the Multilayer Perceptron and the novel DRPNN was compared. Simulation results indicate that HONNs, particularly the DRPNN in most cases demonstrated advantages in capturing chaotic movement in the financial signals with an improvement in the profit return over other network models. The relative superiority of DRPNN to other networks is not just its ability to attain high profit return, but rather to model the training set with fast learning and convergence. The network offers fast training and shows considerable promise as a forecasting tool. It is concluded that DRPNN do have the capability to forecast the financial markets, and individual investor could benefit from the use of this forecasting.
20
Pan, Lingxue. "Resampling in neural networks with application to financial time series." Thesis, National Library of Canada = Bibliothèque nationale du Canada, 2000. http://www.collectionscanada.ca/obj/s4/f2/dsk2/ftp02/NQ47406.pdf.
Full text
21
Burton, Holly. "Reservoir inflow forecasting using time series and neural network models." Thesis, National Library of Canada = Bibliothèque nationale du Canada, 1999. http://www.collectionscanada.ca/obj/s4/f2/dsk1/tape8/PQDD_0017/MQ54220.pdf.
Full text
22
Burton, Holly. "Reservoir inflow forecasting using time series and neural network models." Thesis, McGill University, 1998. http://digitool.Library.McGill.CA:80/R/?func=dbin-jump-full&object_id=29800.
Full textAbstract:
In practice, the reservoir net inflow is computed based upon the application of the water balance equation to the reservoir system since it is difficult to obtain direct and reliable measurements of this variable. The net inflow process has been thus found to possess a random behaviour because it is related to the stochastic nature of various physical processes involved in the water balance computation (e.g., precipitation, evaporation, etc.). Therefore, the aim of this research is to propose a forecasting method that can accurately and efficiently predict the random reservoir inflow series. The proposed forecasting methods considered were the linear regression, the exponential smoothing technique, the periodic autoregressive moving average (PARMA) method, and the neural network procedure. An illustrative application was carried out using 25 years (1970--1994) of monthly rainfall and inflow data from the Pedu-Muda reservoirs in Kedah, Malaysia. The first 18 years (1970--1987) were used for calibration while the remaining 7 years (1988--1994) were used for verification of the proposed models. (Abstract shortened by UMI.)
23
MACHADO, MARIA AUGUSTA SOARES. "IDENTIFICATION OF NON-SEASONAL TIME SERIES THROUGH FUZZY NEURAL NETWORKS." PONTIFÍCIA UNIVERSIDADE CATÓLICA DO RIO DE JANEIRO, 2000. http://www.maxwell.vrac.puc-rio.br/Busca_etds.php?strSecao=resultado&nrSeq=7554@1.
Full textAbstract:
CONSELHO NACIONAL DE DESENVOLVIMENTO CIENTÍFICO E TECNOLÓGICOObservando a dificuldade de batimento (match) dos padrões de comportamento das funções de autocorrelação e de autocorrelação parcial teóricas com as respectivas funções e as autocorrelação e de autocorrelação parcial estimadas de uma séries temporal, aliada ao fato da dificuldade em definir um número em específico como delimitador inequívoco do que seja um lag significativo, tornam clara a dose de julgamento subjetivo a ser realizado por um especialista de análise de séries temporais na tomada de decisão sobre a estrutura de Box & Jenkins adequada a ser escolhida para modelar o processo estocástico sendo estudado. A matemática nebulosa permite a criação de sistemas de inferências nebulosas (inferência dedutiva) e representa o conhecimento de forma explícita, através de regras nebulosas, possibilitando, facilmente, o entendimento do sistema em estudo. Por outro lado, um modelo de redes neurais representa o conhecimento de forma implícita, adquirido através de exemplos (dados), possuindo excelente capacidade de generalização (inferência indutiva). Esta tese apresenta um sistema especialista composto de cinco redes neurais nebulosas do tipo retropropagação para o auxílio na análise de séries temporais não sazonais. O sistema indica ao usuário a estrutura mais adequada, dentre as estruturas AR(1), MA (1), AR(2), MA(2) e ARMA(1,1), tomando como base a menor distância Euclidiana entre os valores esperados e as saídas das redes neurais nebulosas.
It is well known the difficulties associated with the tradicional procedure for model identification of the Box & Jenkins model through the pattern matching of the theoretical and estimated ACF and PACF. The decision on the acceptance of the null hypothesis of zero ACF (or PACF) for a given lag is based on a strong asymptotic result, particularly for the PACF, leading, sometimes, to wrong decisions on the identified order of the models. The fuzzy logic allows one to infer system governed by incomplete or fuzzy knowledge (deductive inference) using a staighforward formulation of the problem via fuzzy mathematics. On the other hand, the neural network represent the knowledge in a implicit manner and has a great generalization capacity (inductive inference). In this thesis we built a specialist system composed of 5 fuzzy neural networks to help on the automatic identificationof the following Box & Jenkins ARMA structure AR(1), MA(1), AR(2), MA(2) and ARMA (1,1), through the Euclidian distance between the estimated output of the net and the corresponding patterns of each one of the five structures.
24
Aupke, Phil. "Uncertainity in Renewable Energy Time Series Prediction using Neural Networks." Thesis, Karlstads universitet, Institutionen för matematik och datavetenskap (from 2013), 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:kau:diva-82714.
Full textAbstract:
With the increasing demand for solar energy, the forecast of the PV station energy production has to be as precisely as possible. To make the prediction more robust, also correlated infor- mation about the weather can be added to the previous energy production of the PV station. This thesis is part of a project, which has the goal to build an energy marketplace for a smart energy grid between households. To make the decisions of the prosumer more accurate, a forecast for the PV station energy production has to be as accurate as possible. Because not every household or even some smart grids will contain a weather station, also interpolated weather information has to be considered. The objective of this work is the evaluation of the accuracy difference between precise weather information, located directly at the PV station and interpolated weather data. The errors of the data were recorded due to misfunctions in the sensors and were cleared with the usage of winsorization. The unnecessary weather features have been detected with several feature selection methods. For the forecast of the energy production three established machine learning algorithms were used: Random Forest, LSTM and Facebook Prophet. For the com- parison of the performance different performance metrics were used. The validation of the three models was carried out by a walk-forward cross validation with unseen data. Further- more, for each of the two datasets one of the three machine learning model were trained. For the performance measurement i.e., the LSTM model trained on precise weather information also received the interpolated data as an input for the prediction and vice versa. As a conclu- sion, the Random Forest model performed better than the other two model types, with an av- erage normalized error of 0.15. Whereas the LSTM model received an error of 0.37 and the Prophet model 0.58. For the difference between interpolated and actual weather information the results prove, that the uncertainity in those variables also affects the prediction of the PV station energy outcome. The LSTM model MSE increased by 14 percent and the Random Forest results with an increasement of 16 percent. The end of the thesis includes a discussion about the results and possible tasks for future work takes place.
25
Haddad, Josef, and Carl Piehl. "Unsupervised anomaly detection in time series with recurrent neural networks." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2019. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-259655.
Full textAbstract:
Artificial neural networks (ANN) have been successfully applied to a wide range of problems. However, most of the ANN-based models do not attempt to model the brain in detail, but there are still some models that do. An example of a biologically constrained ANN is Hierarchical Temporal Memory (HTM). This study applies HTM and Long Short-Term Memory (LSTM) to anomaly detection problems in time series in order to compare their performance for this task. The shape of the anomalies are restricted to point anomalies and the time series are univariate. Pre-existing implementations that utilise these networks for unsupervised anomaly detection in time series are used in this study. We primarily use our own synthetic data sets in order to discover the networks’ robustness to noise and how they compare to each other regarding different characteristics in the time series. Our results shows that both networks can handle noisy time series and the difference in performance regarding noise robustness is not significant for the time series used in the study. LSTM outperforms HTM in detecting point anomalies on our synthetic time series with sine curve trend but a conclusion about the overall best performing network among these two remains inconclusive.Artificiella neurala nätverk (ANN) har tillämpats på många problem. Däremot försöker inte de flesta ANN-modeller efterlikna hjärnan i detalj. Ett exempel på ett ANN som är begränsat till att efterlikna hjärnan är Hierarchical Temporal Memory (HTM). Denna studie tillämpar HTM och Long Short-Term Memory (LSTM) på avvikelsedetektionsproblem i tidsserier för att undersöka vilka styrkor och svagheter de har för detta problem. Avvikelserna i denna studie är begränsade till punktavvikelser och tidsserierna är i endast en variabel. Redan existerande implementationer som utnyttjar dessa nätverk för oövervakad avvikelsedetektionsproblem i tidsserier används i denna studie. Vi använder främst våra egna syntetiska tidsserier för att undersöka hur nätverken hanterar brus och hur de hanterar olika egenskaper som en tidsserie kan ha. Våra resultat visar att båda nätverken kan hantera brus och prestationsskillnaden rörande brusrobusthet var inte tillräckligt stor för att urskilja modellerna. LSTM presterade bättre än HTM på att upptäcka punktavvikelser i våra syntetiska tidsserier som följer en sinuskurva men en slutsats angående vilket nätverk som presterar bäst överlag är fortfarande oavgjord.
26
Kourentzes, Nikolaos. "Input variable selection for time series forecasting with artificial neural networks : an empirical evaluation across varying time series frequencies." Thesis, Lancaster University, 2009. http://eprints.lancs.ac.uk/60234/.
Full textAbstract:
Over the last two decades there has been an increase in the research of artificial neural networks (ANNs) to forecasting problems. Both in theoretical and empirical works, ANNs have shown evidence of good performance, in many cases outperforming established statistical benchmarks. This thesis starts by reviewing the advances in ANNs for time series forecasting, assessing their performance in the literature, analysing the current state of the art, the modelling issues that have been solved and which are still critical for forecasting with ANNs, thereby indicating future research directions. The specification of the input vector is identified as the most crucial unresolved modelling issue for ANNs’ accuracy. Notably, there is no rigorous empirical evaluation of the multiple published input variable selection methodologies. This problem is addressed from four different perspectives. A rigorous evaluation of several published methodologies, along with new proposed variations, is performed on low frequency data, exploring which input variable selection methodologies perform best. This analysis concludes that regression based methodologies outperformed other linear and nonlinear ones. The best way to code deterministic seasonality in the inputs of the ANNs is explored, a topic overlooked in the ANN literature, and a parsimonious encoding based on seasonal indices is proposed. The effect of the frequency of the time series on specifying the inputs for ANNs for forecasting is evaluated, revealing several challenges in modelling high frequency time series and providing evidence that the performance of several input variable specification methodologies is not consistent for different data frequencies. This leads to an evaluation of methodologies to select input variables for ANNs solely for high frequency data. Regression based methodologies are found to perform best, in agreement with the evaluation on low frequency dataset, while the ranking of the remaining methodologies is found to be inconsistent for different data frequencies.
27
Howells, Timothy Paul. "Pattern recognition in physiological time-series data using Bayesian neural networks." Thesis, University of Edinburgh, 2003. http://hdl.handle.net/1842/24717.
Full textAbstract:
This thesis describes the application of Bayesian techniques to the analysis of a large database of physiological time series data collected during the management of patients following traumatic brain injury at the Western General Hospital in Edinburgh. The study can be divided into three main sections: • Model validation using simulated data: Techniques are developed that show that under certain conditions the distribution of network outputs generated by these Bayesian neural networks correctly models the desired conditional probability density functions for a wide range of simple problems for which exact solutions can be derived. This provides the basis for using these models in a scientific context. • Model validation using real data. Statistical prognostic modelling for head injured patients is well advanced using simple demographic and clinical features. The Bayesean techniques developed in the previous section are applied to this problem, and the results are compared to those obtained using standard statistical techniques. • Application of these models to physiological data. The models are now applied to the full database and used to interpret the data and provide new insight into the risk factors for head injured patients in intensive care.
28
Sandström, Carl. "An evolutionary approach to time series forecasting with artificial neural networks." Thesis, KTH, Skolan för datavetenskap och kommunikation (CSC), 2015. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-168224.
Full textAbstract:
In this paper an evolutionary approach to forecasting the stock market is tested and compared with backpropagation. An neuroevolutionary algorithm is implemented and backtested measuring returns and the normalized-mean-square-error for each algorithm on selected stocks from NASDAQ. The results are not entirely conclusive and further investigation would be needed to say definitely, but it seems as a neuroevolutionary approach could outperform backpropagation for time series prediction.
29
Al-Hitmi, Mohammed Abdulla E. "Non-linear data analysis and neural networks for time series prediction." Thesis, University of Sheffield, 2001. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.370084.
Full text
30
FIALHO, MARCELLO MOREIRA STUCKERT. "APPLICATION OF INTERVAL NEURAL NETWORKS TO TIME SERIES FORECASTING AND TRADING." PONTIFÍCIA UNIVERSIDADE CATÓLICA DO RIO DE JANEIRO, 1996. http://www.maxwell.vrac.puc-rio.br/Busca_etds.php?strSecao=resultado&nrSeq=9297@1.
Full textAbstract:
COORDENAÇÃO DE APERFEIÇOAMENTO DO PESSOAL DE ENSINO SUPERIOREsta dissertação apresenta uma proposta de arquitetura de redes neurais de intervalos para previsão de séries financeiras. O desempenho desta arquitetura é analisado através de testes de previsão para algumas séries de mercado. Como contribuição adicional é apresentado um algoritmo de trading automático. Este algoritmo é avaliado aplicando-o à séries de mercado, para mensuração de lucros percentuais. Por fim, dados de previsão, obtidos pela rede proposta, são utilizadas para a otimização do trading.
This text presents a new Neural network architeture to be employed in the forecast of financial series. The architecture´s performance is evaluated through benchmarks, using data from financial series. As an additional contribution, an automatic trading algorithm, which is also evaluated through benchmarks, is presented. Finally, forecast data, obtained with the proposed NN architecture, is used to improve the trading algorithm´s performance.
31
Taskaya-Temizel, Tugba. "Configuration of neural networks to model seasonal and cyclic time series." Thesis, University of Surrey, 2006. http://epubs.surrey.ac.uk/844482/.
Full textAbstract:
Time series often exhibit periodical patterns that can be analysed by conventional statistical techniques. These techniques rely upon an appropriate choice of model parameters that are often difficult to determine. Whilst neural networks also require an appropriate parameter configuration, they offer a way in which non-linear patterns may be modelled. However, evidence from a limited number of experiments has been used to argue that periodical patterns cannot be modelled using such networks. Researchers have argued that combining models for forecasting gives better estimates than single time series models particularly for seasonal and cyclic series. For example, a hybrid architecture comprising an autoregressive integrated moving average model (ARIMA) and a neural network is a well-known technique that has recently been shown to give better forecasts by taking advantage of each model's capabilities. However, this assumption carries the danger of underestimating the relationship between the model's linear and non-linear components, particularly by assuming that individual forecasting techniques are appropriate, say, for modelling the residuals. In this thesis, we show that such combinations do not necessarily outperform individual forecasts. On the contrary, we show that the combined forecast can underperform significantly compared to its constituents'. We also present a method to overcome the perceived limitations of neural networks by determining the configuration parameters of a time delayed neural network from the seasonal data it is being used to model. The motivation of our method is that Occam's razor should guide us in selecting a simpler solution compared to a complex solution. Our method uses a fast Fourier transform to calculate the number of input tapped delays, with results demonstrating improved performance as compared to that of other linear and hybrid seasonal modelling techniques on twelve benchmark time series. Keywords: neural networks, time series, cycles, ARIMA-NN hybrids, Fourier, TDNN.
32
Börjesson, Lukas. "Forecasting Financial Time Series through Causal and Dilated Convolutional Neural Networks." Thesis, Linköpings universitet, Institutionen för datavetenskap, 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-167331.
Full textAbstract:
In this paper, predictions of future price movements of a major American stock index was made by analysing past movements of the same and other correlated indices. A model that has shown very good results in speech recognition was modified to suit the analysis of financial data and was then compared to a base model, restricted by assumptions made for an efficient market. The performance of any model, that is trained by looking at past observations, is heavily influenced by how the division of the data into train, validation and test sets is made. This is further exaggerated by the temporal structure of the financial data, which means that the causal relationship between the predictors and the response is dependent in time. The complexity of the financial system further increases the struggle to make accurate predictions, but the model suggested here was still able to outperform the naive base model by more than 20 percent. The model is, however, too primitive to be used as a trading system, but suitable modifications, in order to turn the model into one, will be discussed in the end of the paper.
33
Moradi, Mahdi. "TIME SERIES FORECASTING USING DUAL-STAGE ATTENTION-BASED RECURRENT NEURAL NETWORK." OpenSIUC, 2020. https://opensiuc.lib.siu.edu/theses/2701.
Full textAbstract:
AN ABSTRACT OF THE RESEARCH PAPER OFMahdi Moradi, for the Master of Science degree in Computer Science, presented on April 1, 2020, at Southern Illinois University Carbondale.TITLE: TIME SERIES FORECASTING USING DUAL-STAGE ATTENTION-BASED RECURRENT NEURAL NETWORKMAJOR PROFESSOR: Dr. Banafsheh Rekabdar
34
Hartvigsen, Thomas. "Adaptively-Halting RNN for Tunable Early Classification of Time Series." Digital WPI, 2018. https://digitalcommons.wpi.edu/etd-theses/1257.
Full textAbstract:
Early time series classification is the task of predicting the class label of a time series before it is observed in its entirety. In time-sensitive domains where information is collected over time it is worth sacrificing some classification accuracy in favor of earlier predictions, ideally early enough for actions to be taken. However, since accuracy and earliness are contradictory objectives, a solution to this problem must find a task-dependent trade-off. There are two common state-of-the-art methods. The first involves an analyst selecting a timestep at which all predictions must be made. This does not capture earliness on a case-by-case basis, so if the selecting timestep is too early, all later signals are missed, and if a signal happens early, the classifier still waits to generate a prediction. The second method is the exhaustive search for signals, which encodes no timing information and is not scalable to high dimensions or long time series. We design the first early classification model called EARLIEST to tackle this multi-objective optimization problem, jointly learning (1) to decide at which time step to halt and generate predictions and (2) how to classify the time series. Each of these is learned based on the task and data features. We achieve an analyst-controlled balance between the goals of earliness and accuracy by pairing a recurrent neural network that learns to classify time series as a supervised learning task with a stochastic controller network that learns a halting-policy as a reinforcement learning task. The halting-policy dictates sequential decisions, one per timestep, of whether or not to halt the recurrent neural network and classify the time series early. This pairing of networks optimizes a global objective function that incorporates both earliness and accuracy. We validate our method via critical clinical prediction tasks in the MIMIC III database from the Beth Israel Deaconess Medical Center along with another publicly available time series classification dataset. We show that EARLIEST out-performs two state-of-the-art LSTM-based early classification methods. Additionally, we dig deeper into our model's performance using a synthetic dataset which shows that EARLIEST learns to halt when it observes signals without having explicit access to signal locations. The contributions of this work are three-fold. First, our method is the first neural network-based solution to early classification of time series, bringing the recent successes of deep learning to this problem. Second, we present the first reinforcement-learning based solution to the unsupervised nature of early classification, learning the underlying distributions of signals without access to this information through trial and error. Third, we propose the first joint-optimization of earliness and accuracy, allowing learning of complex relationships between these contradictory goals.
35
Wasnik, Sachinkumar. "Fatigue Detection in EEG Time Series Data Using Deep Learning." Thesis, The University of Sydney, 2021. https://hdl.handle.net/2123/24917.
Full textAbstract:
Fatigue has widespread effects on the brain’s executive function, reaction time and information processing, causing loss of alertness, that affect safety, and productivity. There are various subjective and behavioural methods to measure fatigue. However, none of them is precise. The work in this thesis employs physiological measures such as heart rate, blood pressure, and breathing that are objective and quantitative indicators. These are thought
to provide reliable measures of fatigue and may be easier to deploy in real world scenarios, compared to the subjective or behavioural methods. In particular, electroencephalogram (EEG) signals have the advantage of being able to measure fatigue in the early stages, and therefore have great potential in the design of early warning system to detect fatigue.
Traditional computational models trained using EEG data show potential improvement in detecting fatigue but require a significant number of electrodes, making deployment in a real-world fatigue detection scenario difficult (e.g., on a driver who is on the road). This project aims to develop computational models to perform fatigue detection using sparse EEG data from only two electrodes. The resulting algorithms could potentially be deployed in pragmatic situations (e.g., embedded in a wearable device), making the contribution of this study useful for real- world scenarios
In machine learning approaches, the area of deep learning has shown excellent performance in tackling problems of image classification and speech recognition. This project introduces the application of deep learning methods in early warning systems of fatigue detection. EEG data of patients suffering from mild to severe Obstructive Sleep Apnoea (OSA) are used in this study. These patients performed a driving simulation test under varying conditions of sleep deprivation, with their wake EEG and driving performance variables continuously monitored. The data collected during a driving simulation test of 57 sleep-deprived subjects were used for training and evaluating the computational models. The principal machine learning task was to employ the EEG data as input and predict the probability of a crash (crash / no crash) before the actual crash event.
After testing a preliminary EEG-K-Nearest Neighbour (EEG-KNN) as proof of concept to test data cleaning and pre-processing, two deep learning models were introduced, EEG-Deep Neural Network (EEG-DNN) and EEG Convolutional Neural Network (EEG-CNN). The Least Absolute Sum of Squares Operator (LASSO) was applied as a feature selection method in EEG-KNN to overcome the curse of dimensionality and identify promising features. EEG-KNN was used to predict a crash in the short-term (i.e., 5-second preduration), while EEG-DNN and EEG-CNN were used to predict a crash in the longer term (i.e., 6-minute pre-duration and 3- minute pre-duration respectively). Techniques such as dropout regularisation and early stopping were used to improve the performance of EEG-DNN and EEG-CNN on the test data.
The Receiver Operating Curve (ROC) is widely used to assess the performance of a classifier and compare the number of true positives (actual crash events) to the number of false positives. The metric considered for the evaluation of computational models on test data is the area under the ROC curve (AUROC). A larger value indicates better classification performance. The EEG-KNN in this study achieved an AUROC of 0.77 in short-term fatigue
detection. The Deep learning model, EEG-DNN significantly improved the performance of crash prediction and achieved a sensitivity level of 87%. Further, the EEG-CNN was used to reduce the number of electrodes required to detect fatigue. The EEG-CNN achieved an AUROC of 0.95.
This project has developed a data framework and computational models to detect fatigue ahead of crash events, making intervention possible in the real-world scenarios. The proposed computational models utilised a lower number of electrodes and worked with sparse EEG data to detect fatigue, thus enabling a practical, effective and easy-to-use solution to be devised.
36
Zhai, Yusheng. "Time series forecasting competition among three sophisticated paradigms /." Electronic version (Microsoft Word), 2005. http://dl.uncw.edu/etd/2005/zhaiy/yushengzhai.html.
Full text
37
Setyawati, Bina R. "Multi-layer feed forward neural networks for foreign exchange time series forecasting." Morgantown, W. Va. : [West Virginia University Libraries], 2005. https://eidr.wvu.edu/etd/documentdata.eTD?documentid=4180.
Full textAbstract:
Thesis (Ph. D.)--West Virginia University, 2005.Title from document title page. Document formatted into pages; contains xii, 185 p. : ill. (some col.). Includes abstract. Includes bibliographical references (p. 140-146).
38
Gallant, Peter Joseph. "A hybrid evolutionary algorithm to train neural networks as time-series predictors." Thesis, National Library of Canada = Bibliothèque nationale du Canada, 2001. http://www.collectionscanada.ca/obj/s4/f2/dsk3/ftp04/NQ59526.pdf.
Full text
39
Novak, Martina. "A neural network approach for simulation and forecasting of chaotic time series." Thesis, Georgia Institute of Technology, 2002. http://hdl.handle.net/1853/19087.
Full text
40
MEDEIROS, MARCELO CUNHA. "A LINEAR-NEURAL HYBRID MODEL FOR ANALYSIS AND FORECASTING OF TIME-SERIES." PONTIFÍCIA UNIVERSIDADE CATÓLICA DO RIO DE JANEIRO, 1998. http://www.maxwell.vrac.puc-rio.br/Busca_etds.php?strSecao=resultado&nrSeq=14540@1.
Full textAbstract:
CONSELHO NACIONAL DE DESENVOLVIMENTO CIENTÍFICO E TECNOLÓGICOEsta dissertação apresenta um modelo não linear auto-regressivo com variáveis exógenas (ARX), para análise e previsão de séries temporais. Os coeficientes do modelo são estimados pela saída de uma rede neural feed-forward, treinada por um algoritmo híbrido de otimização. Os resultados obtidos são comparados tanto com modelos lineares, quanto com não lineares.
This thesis presents a non linear autoregressive model with exogeneous variables (ARX), for time series analysis and forecasting. The coefficients of the model are given by the output of a feed-forward neural network. The results are compared with both linear and non linear models.
41
Chen, Tiankai M. Eng Massachusetts Institute of Technology. "Anomaly detection in semiconductor manufacturing through time series forecasting using neural networks." Thesis, Massachusetts Institute of Technology, 2018. http://hdl.handle.net/1721.1/120245.
Full textAbstract:
Thesis: M. Eng. in Advanced Manufacturing and Design, Massachusetts Institute of Technology, Department of Mechanical Engineering, 2018.Cataloged from PDF version of thesis.
Includes bibliographical references (pages 92-94).
Semiconductor manufacturing provides unique challenges to the anomaly detection problem. With multiple recipes and multivariate data, it is difficult for engineers to reliably detect anomalies in the manufacturing process. An experimental study into anomaly detection through time series forecasting is carried out with application to a plasma etch case study. The study is performed on three predictive models with increasing complexity for comparison. The three models are namely: Autoregressive Integrated Moving Average (ARIMA), Multi-Layer Perceptron (MLP) and Long Short Term Memory (LSTM). ARIMA is a statistical model while MLP and LSTM are neural network models. The results from the control experiment, under supervised training, shows the validity of MLP and LSTM in detecting anomalies through time series forecasting with a recall accuracy of 92% for the best model. Conversely, the ARIMA model has a relatively poor performance due to the inability to model the data correctly. Experimental results also display the ability of neural network models to adapt to training sets of multiple recipes. Furthermore, downsampling is explored to reduce training times and has been found to have minor effects on the accuracy of the model. Moreover, an unsupervised approach towards anomaly detection is found to have little success in detecting anomalous points in the data.
by Tiankai Chen.
M. Eng. in Advanced Manufacturing and Design
42
McDonald, Scott. "Applications of self-organising fuzzy neural networks in financial time series analysis." Thesis, Ulster University, 2016. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.694650.
Full textAbstract:
The forecasting of financial time series is a major research area in statistics, econometrics and, increasingly, computational intelligence. Financial data are known to be extremely complex, nonstationary, and nonlinear in their composition. Machine learning algorithms have shown themselves to be capable of modelling complex datasets, particularly when compared with traditional statistical models. In particular, artificial neural networks are one of the most popular models in the literature. This thesis explores the usage of a particular type of neural network, namely a self organising fuzzy neural network (SOFNN), for financial forecasting applications. A general overview of the computational methods used in the experimental chapters, as well as a review of the existing literature, are presented in Chapters 2 and 3. Chapter 4 investigates the usage of the SOFNN applied to stock price prediction, using random forests and a multi-objective genetic algorithm to automate input variable and parameter selection. In Chapter 5, the efficacy of combining linear statistical models and nonlinear machine learning models is investigated. The effects of combining the forecasts of various models into groups, or ensembles, are also evaluated. Finally, in Chapter 6, an Interval Type 2 (IT2) SOFNN is designed and implemented. It is a more general form of the networks used in Chapters 4 and 5, ba'3ed on Type 2 fuzzy logic. The accuracy and robustness of the Hew model's forecasts are evaluated using a number of financial time series. The results of this work show that the SOFNN is a suitable choice for forecasting financial data. Its dynamic structure and online learning algorithm are particularly useful when dealing with complex, nonstationary data. However, no single model can be expected to he superior to all others in every situation. It is shown that comhining the forecasts of multiple models, even those trained on t.he same datasets, can improve overall forecasting accuracy. Finally, the suitability of the IT2 SOFNN for predicting stock prices is established. The increased modelling capabilities of the more general Type 2 fuzzy membership functions, as well as an increased robustness to noise, makes it an attractive choice for this application.
43
Winn, David. "An analysis of neural networks and time series techniques for demand forecasting." Thesis, Rhodes University, 2007. http://hdl.handle.net/10962/d1004362.
Full textAbstract:
This research examines the plausibility of developing demand forecasting techniques which are consistently and accurately able to predict demand. Time Series Techniques and Artificial Neural Networks are both investigated. Deodorant sales in South Africa are specifically studied in this thesis. Marketing techniques which are used to influence consumer buyer behaviour are considered, and these factors are integrated into the forecasting models wherever possible. The results of this research suggest that Artificial Neural Networks can be developed which consistently outperform industry forecasting targets as well as Time Series forecasts, suggesting that producers could reduce costs by adopting this more effective method.
44
Tse, Peter W. "Neural networks for machine fault diagnosis and life span prediction." Thesis, University of Sussex, 1997. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.390518.
Full text
45
Vendramin, Nicoló. "Unsupervised Anomaly Detection on Multi-Process Event Time Series." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2018. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-254885.
Full textAbstract:
Establishing whether the observed data are anomalous or not is an important task that has been widely investigated in literature, and it becomes an even more complex problem if combined with high dimensional representations and multiple sources independently generating the patterns to be analyzed. The work presented in this master thesis employs a data-driven pipeline for the definition of a recurrent auto-encoder architecture to analyze, in an unsupervised fashion, high-dimensional event time-series generated by multiple and variable processes interacting with a system. Facing the above mentioned problem the work investigates whether it is possible or not to use a single model to analyze patterns produced by different sources. The analysis of log files that record events of interaction between users and the radio network infrastructure is employed as realworld case-study for the given problem. The investigation aims to verify the performances of a single machine learning model applied to the learning of multiple patterns developed through time by distinct sources. The work proposes a pipeline, to deal with the complex representation of the data source and the definition and tuning of the anomaly detection model, that is based on no domain-specific knowledge and can thus be adapted to different problem settings. The model has been implemented in four different variants that have been evaluated over both normal and anomalous data, gathered partially from real network cells and partially from the simulation of anomalous behaviours. The empirical results show the applicability of the model for the detection of anomalous sequences and events in the described conditions, with scores reaching above 80% in terms of F1-score, and varying depending on the specific threshold setting. In addition, their deeper interpretation gives insights about the difference between the variants of the model and thus, their limitations and strong points.Att fastställa huruvida observerade data är avvikande eller inte är en viktig uppgift som har studerats ingående i litteraturen och problemet blir ännu mer komplext, om detta kombineras med högdimensionella representationer och flera källor som oberoende genererar de mönster som ska analyseras. Arbetet som presenteras i denna uppsats använder en data-driven pipeline för definitionen av en återkommande auto-encoderarkitektur för att analysera, på ett oövervakat sätt, högdimensionella händelsetidsserier som genereras av flera och variabla processer som interagerar med ett system. Mot bakgrund av ovanstående problem undersöker arbetet om det är möjligt eller inte att använda en enda modell för att analysera mönster som producerats av olika källor. Analys av loggfiler som registrerar händelser av interaktion mellan användare och radionätverksinfrastruktur används som en fallstudie för det angivna problemet. Undersökningen syftar till att verifiera prestandan hos en enda maskininlärningsmodell som tillämpas för inlärning av flera mönster som utvecklats över tid från olika källor. Arbetet föreslår en pipeline för att hantera den komplexa representationen hos datakällorna och definitionen och avstämningen av anomalidetektionsmodellen, som inte är baserad på domänspecifik kunskap och därför kan anpassas till olika probleminställningar. Modellen har implementerats i fyra olika varianter som har utvärderats med avseende på både normala och avvikande data, som delvis har samlats in från verkliga nätverksceller och delvis från simulering av avvikande beteenden. De empiriska resultaten visar modellens tillämplighet för detektering av avvikande sekvenser och händelser i det föreslagna ramverket, med F1-score över 80%, varierande beroende på den specifika tröskelinställningen. Dessutom ger deras djupare tolkning insikter om skillnaden mellan olika varianter av modellen och därmed deras begränsningar och styrkor.
46
Kasderidis, Stathis P. "A compartmental model neuron, its networks and application to time series." Thesis, King's College London (University of London), 1999. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.313657.
Full text
47
Capanni, Niccolo Francesco. "The functionality of spatial and time domain artificial neural models." Thesis, Robert Gordon University, 2006. http://hdl.handle.net/10059/241.
Full textAbstract:
This thesis investigates the functionality of the units used in connectionist Artificial Intelligence systems. Artificial Neural Networks form the foundation of the research and their units, Artificial Neurons, are first compared with alternative models. This initial work is mainly in the spatial-domain and introduces a new neural model, termed a Taylor Series neuron. This is designed to be flexible enough to assume most mathematical functions. The unit is based on Power Series theory and a specifically implemented Taylor Series neuron is demonstrated. These neurons are of particular usefulness in evolutionary networks as they allow the complexity to increase without adding units. Training is achieved via various traditiona and derived methods based on the Delta Rule, Backpropagation, Genetic Algorithms and associated evolutionary techniques. This new neural unit has been presented as a controllable and more highly functional alternative to previous models. The work on the Taylor Series neuron moved into time-domain behaviour and through the investigation of neural oscillators led to an examination of single-celled intelligence from which the later work developed. Connectionist approaches to Artificial Intelligence are almost always based on Artificial Neural Networks. However, another route towards Parallel Distributed Processing was introduced. This was inspired by the intelligence displayed by single-celled creatures called Protoctists (Protists). A new system based on networks of interacting proteins was introduced. These networks were tested in pattern-recognition and control tasks in the time-domain and proved more flexible than most neuron models. They were trained using a Genetic Algorithm and a derived Backpropagation Algorithm. Termed "Artificial BioChemical Networks" (ABN) they have been presented as an alternative approach to connectionist systems.
48
Tadjuidje, Kamgaing Joseph. "Competing neural networks as models for non stationary financial time series changepoint analysis /." [S.l. : s.n.], 2005. http://deposit.ddb.de/cgi-bin/dokserv?idn=974108014.
Full text
49
Silver-Warner, Stephen John. "Associative memory neural networks : an investigation with application to chaotic time series prediction." Thesis, Brunel University, 1997. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.362486.
Full text
50
Kingdon, Jason Conrad. "Feed forward neural networks and genetic algorithms for automated financial time series modelling." Thesis, University College London (University of London), 1995. http://discovery.ucl.ac.uk/1318052/.
Full textAbstract:
This thesis presents an automated system for financial time series modelling. Formal and applied methods are investigated for combining feed-forward Neural Networks and Genetic Algorithms (GAs) into a single adaptive/learning system for automated time series forecasting. Four important research contributions arise from this investigation: i) novel forms of GAs are introduced which are designed to counter the representational bias associated with the conventional Holland GA, ii) an experimental methodology for validating neural network architecture design strategies is introduced, iii) a new method for network pruning is introduced, and iv) an automated method for inferring network complexity for a given learning task is devised. These methods provide a general-purpose applied methodology for developing neural network applications and are tested in the construction of an automated system for financial time series modelling. Traditional economic theory has held that financial price series are random. The lack of a priori models on which to base a computational solution for financial modelling provides one of the hardest tests of adaptive system technology. It is shown that the system developed in this thesis isolates a deterministic signal within a Gilt Futures prices series, to a confidences level of over 99%, yielding a prediction accuracy of over 60% on a single run of 1000 out-of-sample experiments. An important research issue in the use of feed-forward neural networks is the problems associated with parameterisation so as to ensure good generalisation. This thesis conducts a detailed examination of this issue. A novel demonstration of a network's ability to act as a universal functional approximator for finite data sets is given. This supplies an explicit formula for setting a network's architecture and weights in order to map a finite data set to arbitrary precision. It is shown that a network's ability to generalise is extremely sensitive to many parameter choices and that unless careful safeguards are included in the experimental procedure over-fitting can occur. This thesis concentrates on developing automated techniques so as to tackle these problems. Techniques for using GAs to parameterise neural networks are examined. It is shown that the relationship between the fitness function, the GA operators and the choice of encoding are all instrumental in determining the likely success of the GA search. To address this issue a new style of GA is introduced which uses multiple encodings in the course of a run. These are shown to out-perform the Holland GA on a range of standard test functions. Despite this innovation it is argued that the direct use of GAs to neural network parameterisation runs the risk of compounding the network sensitivity issue. Moreover, in the absence of a precise formulation of generalisation a less direct use of GAs to network parameterisation is examined. Specifically a technique, artficia1 network generation (ANG), is introduced in which a GA is used to artificially generate test learning problems for neural networks that have known network solutions. ANG provides a means for directly testing i) a neural net architecture, ii) a neural net training process, and iii) a neural net validation procedure, against generalisation. ANG is used to provide statistical evidence in favour of Occam's Razor as a neural network design principle. A new method for pruning and inferring network complexity for a given learning problem is introduced. Network Regression Pruning (NRP) is a network pruning method that attempts to derive an optimal network architecture by starting from what is considered an overly large network. NRP differs radically from conventional pruning methods in that it attempts to hold a trained network's mapping fixed as pruning proceeds. NRP is shown to be extremely successful at isolating optimal network architectures on a range of test problems generated using ANG. Finally, NRP and techniques validated using ANG are combined to implement an Automated Neural network Time series Analysis System (ANTAS). ANTAS is applied to the gilt futures price series The Long Gilt Futures Contract (LGFC).