To see the other types of publications on this topic, follow the link: LSTM modeli.

Dissertations / Theses on the topic 'LSTM modeli'

Create a spot-on reference in APA, MLA, Chicago, Harvard, and other styles

Select a source type:

Consult the top 50 dissertations / theses for your research on the topic 'LSTM modeli.'

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

Li, Edwin. "LSTM Neural Network Models for Market Movement Prediction." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2018. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-231627.

Full text
Abstract:
Interpreting time varying phenomena is a key challenge in the capital markets. Time series analysis using autoregressive methods has been carried out over the last couple of decades, often with reassuring results. However, such methods sometimes fail to explain trends and cyclical fluctuations, which may be characterized by long-range dependencies or even dependencies between the input features. The purpose of this thesis is to investigate whether recurrent neural networks with LSTM-cells can be used to capture these dependencies, and ultimately be used as a complement for index trading decisions. Experiments are made on different setups of the S&amp;P-500 stock index, and two distinct models are built, each one being an improvement of the previous model. The first model is a multivariate regression model, and the second model is a multivariate binary classifier. The output of each model is used to reason about the future behavior of the index. The experiment shows for the configuration provided that LSTM RNNs are unsuitable for predicting exact values of daily returns, but gives satisfactory results when used to predict the direction of the movement.<br>Att förstå och kunna förutsäga hur index varierar med tiden och andra parametrar är ett viktigt problem inom kapitalmarknader. Tidsserieanalys med autoregressiva metoder har funnits sedan årtionden tillbaka, och har oftast gett goda resultat. Dessa metoder saknar dock möjligheten att förklara trender och cykliska variationer i tidsserien, något som kan karaktäriseras av tidsvarierande samband, men även samband mellan parametrar som indexet beror utav. Syftet med denna studie är att undersöka om recurrent neural networks (RNN) med long short-term memory-celler (LSTM) kan användas för att fånga dessa samband, för att slutligen användas som en modell för att komplettera indexhandel. Experimenten är gjorda mot en modifierad S&amp;P-500 datamängd, och två distinkta modeller har tagits fram. Den ena är en multivariat regressionsmodell för att förutspå exakta värden, och den andra modellen är en multivariat klassifierare som förutspår riktningen på nästa dags indexrörelse. Experimenten visar för den konfiguration som presenteras i rapporten att LSTM RNN inte passar för att förutspå exakta värden för indexet, men ger tillfredsställande resultat när modellen ska förutsäga indexets framtida riktning.
APA, Harvard, Vancouver, ISO, and other styles
2

Tang, Hao. "Bidirectional LSTM-CNNs-CRF Models for POS Tagging." Thesis, Uppsala universitet, Institutionen för lingvistik och filologi, 2018. http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-362823.

Full text
Abstract:
In order to achieve state-of-the-art performance for part-of-speech(POS) tagging, the traditional systems require a significant amount of hand-crafted features and data pre-processing. In this thesis, we present a discriminative word embedding, character embedding and byte pair encoding (BPE) hybrid neural network architecture to implement a true end-to-end system without feature engineering and data pre-processing. The neural network architecture is a combination of bidirectional LSTM, CNNs, and CRF, which can achieve a state-of-the-art performance for a wide range of sequence labeling tasks. We evaluate our model on Universal Dependencies (UD) dataset for English, Spanish, and German POS tagging. It outperforms other models with 95.1%, 98.15%, and 93.43% accuracy on testing datasets respectively. Moreover, the largest improvements of our model appear on out-of-vocabulary corpora for Spanish and German. According to statistical significance testing, the improvements of English on testing and out-of-vocabulary corpora are not statistically significant. However, the improvements of the other more morphological languages are statistically significant on their corresponding corpora.
APA, Harvard, Vancouver, ISO, and other styles
3

Backer-Meurke, Henrik, and Marcus Polland. "Predicting Road Rut with a Multi-time-series LSTM Model." Thesis, Högskolan Dalarna, Institutionen för information och teknik, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:du-37599.

Full text
Abstract:
Road ruts are depressions or grooves worn into a road. Increases in rut depth are highly undesirable due to the heightened risk of hydroplaning. Accurately predicting increases in road rut depth is important for maintenance planning within the Swedish Transport Administration. At the time of writing this paper, the agency utilizes a linear regression model and is developing a feed-forward neural network for road rut predictions. The aim of the study was to evaluate the possibility of using a Recurrent Neural Network to predict road rut. Through design science research, an artefact in the form of a LSTM model was designed, developed, and evaluated.The dataset consisted of multiple-multivariate short time series where research was limited. Case studies were conducted which inspired the conceptual design of the model. The baseline LSTM model proposed in this paper utilizes the full dataset in combination with time-series individualization through an added index feature. Additional features thought to correlate with rut depth was also studied through multiple training set variations. The model was evaluated by calculating the Root Mean Squared Error (RMSE) and the Mean Absolute Error (MAE) for each training set variation. The baseline model predicted rut depth with a MAE of 0.8110 (mm) and a RMSE of 1.124 (mm) outperforming a control set without the added index. The feature with the highest correlation to rut depth was curvature with a MAEof 0.8031 and a RMSE of 1.1093. Initial finding shows that there is a possibility of utilizing an LSTM model trained on multiple-multivariate time series to predict rut depth. Time series individualization through an added index feature yielded better results than control, indicating that it had the desired effect on model performance.
APA, Harvard, Vancouver, ISO, and other styles
4

Vitali, Greta <1995&gt. "“Forecasting Stock Index Volatility: A comparison between GARCH and LSTM models”." Master's Degree Thesis, Università Ca' Foscari Venezia, 2019. http://hdl.handle.net/10579/15933.

Full text
Abstract:
The financial world is characterized by the uncertainty of events and this phenomenon can expose operators to huge financial risks. Thus, there is a need to measure this uncertainty, with the aim to predict it and to make adequate plans of action. The concept of uncertainty is often associated with the definition of volatility, which is a measure of the variation of stock prices of a financial instrument during the time. But modelling volatility is not a trivial task, because of the essence of financial stock prices, which usually present volatility clusters, fat tails, nonnormality and structural breaks in the distribution. A popular class of models able to capture many of these stylized facts is the ARCH/GARCH family. As a matter of fact, a GARCH model is able to explain the time-varying variance and the presence of clusters in the series of the returns. Nevertheless, it requires some constraints on both parameters and distributions of returns to obtain satisfactory results. An attractive solution is given by some mathematical models based on artificial intelligence. Indeed, the artificial neural networks, resembling the human brain, are able to make predictions of future volatility due to their ability to be self-adaptive and to be a universal approximator of any underlying nonlinear function of financial data. The aim of this thesis is to make a comparison between the forecasting capabilities of a GARCH(1,1) model and a Long Short-Term Memory network. In particular, the objective is to predict the volatility of the Dow Jones Industrial Average Index, demonstrating the superiority of the neural network with respect to the well-established GARCH model.
APA, Harvard, Vancouver, ISO, and other styles
5

Wang, Nancy. "Spectral Portfolio Optimisation with LSTM Stock Price Prediction." Thesis, KTH, Matematisk statistik, 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-273611.

Full text
Abstract:
Nobel Prize-winning modern portfolio theory (MPT) has been considered to be one of the most important and influential economic theories within finance and investment management. MPT assumes investors to be riskaverse and uses the variance of asset returns as a proxy of risk to maximise the performance of a portfolio. Successful portfolio management reply, thus on accurate risk estimate and asset return prediction. Risk estimates are commonly obtained through traditional asset pricing factor models, which allow the systematic risk to vary over time domain but not in the frequency space. This approach can impose limitations in, for instance, risk estimation. To tackle this shortcoming, interest in applications of spectral analysis to financial time series has increased lately. Among others, the novel spectral portfolio theory and the spectral factor model which demonstrate enhancement in portfolio performance through spectral risk estimation [1][11]. Moreover, stock price prediction has always been a challenging task due to its non-linearity and non-stationarity. Meanwhile, Machine learning has been successfully implemented in a wide range of applications where it is infeasible to accomplish the needed tasks traditionally. Recent research has demonstrated significant results in single stock price prediction by artificial LSTM neural network [6][34]. This study aims to evaluate the combined effect of these two advancements in a portfolio optimisation problem and optimise a spectral portfolio with stock prices predicted by LSTM neural networks. To do so, we began with mathematical derivation and theoretical presentation and then evaluated the portfolio performance generated by the spectral risk estimates and the LSTM stock price predictions, as well as the combination of the two. The result demonstrates that the LSTM predictions alone performed better than the combination, which in term performed better than the spectral risk alone.<br>Den nobelprisvinnande moderna portföjlteorin (MPT) är utan tvekan en av de mest framgångsrika investeringsmodellerna inom finansvärlden och investeringsstrategier. MPT antar att investerarna är mindre benägna till risktagande och approximerar riskexponering med variansen av tillgångarnasränteavkastningar. Nyckeln till en lyckad portföljförvaltning är därmed goda riskestimat och goda förutsägelser av tillgångspris. Riskestimering görs vanligtvis genom traditionella prissättningsmodellerna som tillåter risken att variera i tiden, dock inte i frekvensrummet. Denna begränsning utgör bland annat ett större fel i riskestimering. För att tackla med detta har intresset för tillämpningar av spektraanalys på finansiella tidsserier ökat de senast åren. Bland annat är ett nytt tillvägagångssätt för att behandla detta den nyintroducerade spektralportföljteorin och spektralfak- tormodellen som påvisade ökad portföljenprestanda genom spektralriskskattning [1][11]. Samtidigt har prediktering av aktierpriser länge varit en stor utmaning på grund av dess icke-linjära och icke-stationära egenskaper medan maskininlärning har kunnat använts för att lösa annars omöjliga uppgifter. Färska studier har påvisat signifikant resultat i aktieprisprediktering med hjälp av artificiella LSTM neurala nätverk [6][34]. Detta arbete undersöker kombinerade effekten av dessa två framsteg i ett portföljoptimeringsproblem genom att optimera en spektral portfölj med framtida avkastningar predikterade av ett LSTM neuralt nätverk. Arbetet börjar med matematisk härledningar och teoretisk introduktion och sedan studera portföljprestation som genereras av spektra risk, LSTM aktieprispredikteringen samt en kombination av dessa två. Resultaten visar på att LSTM-predikteringen ensam presterade bättre än kombinationen, vilket i sin tur presterade bättre än enbart spektralriskskattningen.
APA, Harvard, Vancouver, ISO, and other styles
6

Nääs, Starberg Filip, and Axel Rooth. "Predicting a business application's cloud server CPU utilization using the machine learning model LSTM." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-301247.

Full text
Abstract:
Cloud Computing sees increased adoption as companies seek to increase flexibility and reduce cost. Although the large cloud service providers employ a pay-as-you-go pricing model and enable customers to scale up and down quickly, there is still room for improvement. Workload in the form of CPU utilization often fluctuates which leads to unnecessary cost and environmental impact for companies. To help mitigate this issue, the aim of this paper is to predict future CPU utilization using a long short-term memory (LSTM) machine learning model. By predicting utilization up to 30 minutes into the future, companies are able to scale their capacity just in time and avoid unnecessary cost and damage to the environment. The study is divided into two parts. The first part analyses how well the LSTM model performs when predicting one step at a time compared with a state-of-the-art model. The second part analyses the accuracy of the LSTM when making predictions up to 30 minutes into the future. To allow for an objective analysis of results, the LSTM is compared with a standard RNN, which is similar to the LSTM in its inherit algorithmic structure. To conclude, the results suggest that LSTM may be a useful tool for reducing cost and unnecessary environmental impact for business applications hosted on a public cloud.<br>Användandet av molntjänster ökar bland företag som önskar förbättrad flexibilitet och sänkta kostnader. De stora molntjänstleverantörerna använder en prismodell där kostnaden är direkt kopplad till användningen, och låter kunderna snabbt ställa om sin kapacitet, men det finns ändå förbättringsmöjligheter. CPU-behoven fluktuerar ofta vilket leder till meningslösa kostnader och onödig påverkan på klimatet när kapacitet är outnyttjad. För att lindra detta problem används i denna rapport en LSTM maskininlärningsmodell för att förutspå framtida CPU-utnyttjande. Genom att förutspå utnyttjandet upp till 30 minuter in i framtiden hinner företag ställa om sin kapacitet och undvika onödig kostnad och klimatpåverkan. Arbetet ¨ar uppdelat i två delar. Först en del där LSTM-modellen förutspår ett tidssteg åt gången. Därefter en del som analyserar träffsäkerheten för LSTM flera tidssteg in i framtiden, upp till 30 tidssteg. För att möjliggöra en objektiv utvärdering så jämfördes LSTM-modellen med ett standard recurrent neural network (RNN) vilken liknar LSTM i sin struktur. Resultaten i denna studie visar att LSTM verkar vara ¨överlägsen RNN, både när det gäller att förutspå ett tidssteg in i framtiden och när det gäller flera tidssteg in i framtiden. LSTM-modellen var kapabel att förutspå CPU-utnyttjandet 30 minuter in i framtiden med i hög grad bibehållen träffsäkerhet, vilket också var målet med studien. Sammanfattningsvis tyder resultaten på att denna LSTM-modell, och möjligen liknande LSTM-modeller, har potential att användas i samband med företagsapplikationer då man önskar att reducera onödig kostnad och klimatpåverkan.
APA, Harvard, Vancouver, ISO, and other styles
7

Werngren, Simon. "Comparison of different machine learning models for wind turbine power predictions." Thesis, Uppsala universitet, Avdelningen för systemteknik, 2018. http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-362332.

Full text
Abstract:
The goal of this project is to compare different machine learning algorithms ability to predict wind power output 48 hours in advance from earlier power data and meteorological wind speed predictions. Three different models were tested, two autoregressive integrated moving average (ARIMA) models one with exogenous regressors one without and one simple LSTM neural net model. It was found that the ARIMA model with exogenous regressors was the most accurate while also beingrelatively easy to interpret and at 1h 45min 32s had a comparatively short training time. The LSTM was less accurate, harder to interpretand took 14h 3min 5s to train. However the LSTM only took 32.7s to create predictions once the model was trained compared to the 33min13.7s it took for the ARIMA model with exogenous regressors to deploy.Because of this fast deployment time the LSTM might be preferable in certain situations. The ARIMA model without exogenous regressors was significantly less accurate than the other two without significantly improving on the other ARIMA model in any way
APA, Harvard, Vancouver, ISO, and other styles
8

Shojaee, Ali B. S. "Bacteria Growth Modeling using Long-Short-Term-Memory Networks." University of Cincinnati / OhioLINK, 2021. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1617105038908441.

Full text
APA, Harvard, Vancouver, ISO, and other styles
9

Anese, Gianluca <1995&gt. "Explanatory power of GARCH models using news-based investor sentiment: Applications of LSTM networks for text classification." Master's Degree Thesis, Università Ca' Foscari Venezia, 2020. http://hdl.handle.net/10579/16940.

Full text
Abstract:
Many authors have shown that investors are not fully rational, as the traditional Efficient Markets Hypothesis suggests, and that investor sentiment can have an impact on stock prices. As investor sentiment is not directly measurable, different proxies have been used by researchers. In addition, progress in natural language processing has contributed to the development of new sentiment measures based on text sources obtained by news providers and social media. This work deals with a classification problem on financial news data and defines a reliable proxy for investor sentiment using both dictionary – based and supervised Machine Learning techniques. In particular, LSTMs networks have been adopted. The resulting sentiment proxies have been used as exogenous variables in the mean and variance equations of a Generalized Autoregressive Conditional Heteroskedasticity model in order to prove the existence of a relationship among them and stock returns and among them and volatility.
APA, Harvard, Vancouver, ISO, and other styles
10

Suresh, Sreerag. "An Analysis of Short-Term Load Forecasting on Residential Buildings Using Deep Learning Models." Thesis, Virginia Tech, 2020. http://hdl.handle.net/10919/99287.

Full text
Abstract:
Building energy load forecasting is becoming an increasingly important task with the rapid deployment of smart homes, integration of renewables into the grid and the advent of decentralized energy systems. Residential load forecasting has been a challenging task since the residential load is highly stochastic. Deep learning models have showed tremendous promise in the fields of time-series and sequential data and have been successfully used in the field of short-term load forecasting at the building level. Although, other studies have looked at using deep learning models for building energy forecasting, most of those studies have looked at limited number of homes or an aggregate load of a collection of homes. This study aims to address this gap and serve as an investigation on selecting the better deep learning model architecture for short term load forecasting on 3 communities of residential buildings. The deep learning models CNN and LSTM have been used in the study. For 15-min ahead forecasting for a collection of homes it was found that homes with a higher variance were better predicted by using CNN models and LSTM showed better performance for homes with lower variances. The effect of adding weather variables on 24-hour ahead forecasting was studied and it was observed that adding weather parameters did not show an improvement in forecasting performance. In all the homes, deep learning models are shown to outperform the simple ANN model.<br>Master of Science<br>Building energy load forecasting is becoming an increasingly important task with the rapid deployment of smart homes, integration of renewables into the grid and the advent of decentralized energy systems. Residential load forecasting has been a challenging task since residential load is highly stochastic. Deep learning models have showed tremendous promise in the fields of time-series and sequential data and have been successfully used in the field of short-term load forecasting. Although, other studies have looked at using deep learning models for building energy forecasting, most of those studies have looked at only a single home or an aggregate load of a collection of homes. This study aims to address this gap and serve as an analysis on short term load forecasting on 3 communities of residential buildings. Detailed analysis on the model performances across all homes have been studied. Deep learning models have been used in this study and their efficacy is measured compared to a simple ANN model.
APA, Harvard, Vancouver, ISO, and other styles
11

Parthiban, Dwarak Govind. "On the Softmax Bottleneck of Word-Level Recurrent Language Models." Thesis, Université d'Ottawa / University of Ottawa, 2020. http://hdl.handle.net/10393/41412.

Full text
Abstract:
For different input contexts (sequence of previous words), to predict the next word, a neural word-level language model outputs a probability distribution over all the words in the vocabulary using a softmax function. When the log of probability outputs for all such contexts are stacked together, the resulting matrix is a log probability matrix which can be denoted as Q_theta, where theta denotes the model parameters. When language modeling is formulated as a matrix factorization problem, the matrix to be factorized Q_theta is expected to be high-rank as natural language is highly context-dependent. But existing softmax based word-level language models have a limitation of not being able to produce such matrices; this is known as the softmax bottleneck. There are several works that attempted to overcome the limitations introduced by softmax bottleneck, such as the models that can produce high-rank Q_theta. During the process of reproducing the results of these works, we observed that the rank of Q_theta does not always positively correlate with better performance (i.e., lower test perplexity). This puzzling observation triggered us to conduct a systematic investigation to check the influence of rank of Q_theta on better performance of a language model. We first introduce a new family of activation functions called the Generalized SigSoftmax (GSS). By controlling the parameters of GSS, we were able to construct language models that can produce Q_theta with diverse ranks (i.e., low, medium, and high ranks). For models that use GSS with different parameters, we observe that rank does not have a strong positive correlation with perplexity on the test data, reinforcing the support of our initial observation. By inspecting the top-5 predictions made by different models for a selected set of input contexts, we observe that a high-rank Q_theta does not guarantee a strong qualitative performance. Then, we conduct experiments to check if there are any other additional benefits in having models that can produce high-rank Q_theta. We expose that Q_theta rather suffers from the phenomenon of fast singular value decay. Additionally, we also propose an alternative metric to denote the rank of any matrix known as epsilon-effective rank, which can be useful to approximately quantify the singular value distribution when different values for epsilon are used. We conclude by showing that it is the regularization which has played a positive role in the performance of these high-rank models in comparison to the chosen baselines, and there is no single model yet which truly gains improved expressiveness just because of breaking the softmax bottleneck.
APA, Harvard, Vancouver, ISO, and other styles
12

Hellstenius, Sasha. "Model comparison of patient volume prediction in digital health care." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2018. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-229908.

Full text
Abstract:
Accurate predictions of patient volume are an essential tool to improve resource allocation and doctor utilization in the traditional, as well as the digital health care domain. Varying methods for patient volume prediction within the traditional health care domain has been studied in contemporary research, while the concept remains underexplored within the digital health care domain. In this paper, an evaluation of how two different non-linear state-of-the-art time series prediction models compare when predicting patient volume within the digital health care domain is presented. The models compared are the feed forward Multi-layer Percepron (MLP) and the recursive Long Short-Term Memory (LSTM) network. The results imply that the prediction problem itself is straightforward, while also indicating that there are significant differences in prediction accuracy between the evaluated models. The conclusions presented state that that the LSTM model offers substantial prediction advantages that outweigh the complexity overhead for the given problem.<br>En korrekt förutsägelse av patientvolym är essentiell för att förbättra resursallokering av läkare inom traditionell liksom digital vård. Olika metoder för förutsägelse av patientvolym har undersökts inom den traditionella vården medan liknande studier inom den digitala sektorn saknas. I denna uppsats undersöks två icke-linjära moderna metoder för tidsserieanalys av patientvolym inom den digitala sjukvården. Modellerna som undersöks är multi-lagersperceptronen (MLP) samt Long Short-Term Memory (LSTM) nätverket. Resultaten som presenteras indikerar att problemet i sig är okomplicerat samtidigt som det visar sig finnas signifikanta skillnader i korrektheten av förutsägelser mellan de olika modellerna. Slutsatserna som presenteras pekar på att LSTM-modellen erbjuder signifikanta fördelar som överväger komplexitets- och prestandakostnaden.
APA, Harvard, Vancouver, ISO, and other styles
13

Cissoko, Mamadou Ben Hamidou. "Adaptive time-aware LSTM for predicting and interpreting ICU patient trajectories from irregular data." Electronic Thesis or Diss., Strasbourg, 2024. http://www.theses.fr/2024STRAD012.

Full text
Abstract:
En médecine prédictive personnalisée, modéliser avec précision la maladie et les processus de soins d'un patient est crucial en raison des dépendances temporelles à long terme inhérentes. Cependant, les dossiers de santé électroniques (DSE) se composent souvent de données épisodiques et irrégulières, issues des admissions hospitalières sporadiques, créant des schémas uniques pour chaque séjour hospitalier.Par conséquent, la construction d'un modèle prédictif personnalisé nécessite une considération attentive de ces facteurs pour capturer avec précision le parcours de santé du patient et aider à la prise de décision clinique.LSTM sont efficaces pour traiter les données séquentielles comme les DSE, mais ils présentent deux limitations majeures : l'incapacité à interpréter les résultats des prédictions et à prendre en compte des intervalles de temps irréguliers entre les événements consécutifs. Pour surmonter ces limitations, nous introduisons de nouveaux réseaux neuronaux à mémoire dynamique profonde appelés Multi-Way Adaptive et Adaptive Multi-Way Interpretable Time-Aware LSTM (MWTA-LSTM etAMITA), conçus pour les données séquentielles collectées de manière irrégulière.L'objectif principal des deux modèles est de tirer parti des dossiers médicaux pour mémoriser les trajectoires de maladie et les processus de soins, estimer les états de maladie actuels et prédire les risques futurs, offrant ainsi un haut niveau de précision et de pouvoir prédictif<br>In personalized predictive medicine, accurately modeling a patient's illness and care processes is crucial due to the inherent long-term temporal dependencies. However, Electronic Health Records (EHRs) often consist of episodic and irregularly timed data, stemming from sporadic hospital admissions, which create unique patterns for each hospital stay. Consequently, constructing a personalized predictive model necessitates careful consideration of these factors to accurately capture the patient's health journey and assist in clinical decision-making. LSTM networks are effective for handling sequential data like EHRs, but they face two significant limitations: the inability to interpret prediction results and to take into account irregular time intervals between consecutive events. To address limitations, we introduce novel deep dynamic memory neural networks called Multi-Way Adaptive and Adaptive Multi-Way Interpretable Time-Aware LSTM (MWTA-LSTM and AMITA) designed for irregularly collected sequential data. The primary objective of both models is to leverage medical records to memorize illness trajectories and care processes, estimate current illness states, and predict future risks, thereby providing a high level of precision and predictive power
APA, Harvard, Vancouver, ISO, and other styles
14

Vera, Barberán José María. "Adding external factors in Time Series Forecasting : Case study: Ethereum price forecasting." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-289187.

Full text
Abstract:
The main thrust of time-series forecasting models in recent years has gone in the direction of pattern-based learning, in which the input variable for the models is a vector of past observations of the variable itself to predict. The most used models based on this traditional pattern-based approach are the autoregressive integrated moving average model (ARIMA) and long short-term memory neural networks (LSTM). The main drawback of the mentioned approaches is their inability to react when the underlying relationships in the data change resulting in a degrading predictive performance of the models. In order to solve this problem, various studies seek to incorporate external factors into the models treating the system as a black box using a machine learning approach which generates complex models that require a large amount of data for their training and have little interpretability. In this thesis, three different algorithms have been proposed to incorporate additional external factors into these pattern-based models, obtaining a good balance between forecast accuracy and model interpretability. After applying these algorithms in a study case of Ethereum price time-series forecasting, it is shown that the prediction error can be efficiently reduced by taking into account these influential external factors compared to traditional approaches while maintaining full interpretability of the model.<br>Huvudinstrumentet för prognosmodeller för tidsserier de senaste åren har gått i riktning mot mönsterbaserat lärande, där ingångsvariablerna för modellerna är en vektor av tidigare observationer för variabeln som ska förutsägas. De mest använda modellerna baserade på detta traditionella mönsterbaserade tillvägagångssätt är auto-regressiv integrerad rörlig genomsnittsmodell (ARIMA) och långa kortvariga neurala nätverk (LSTM). Den huvudsakliga nackdelen med de nämnda tillvägagångssätten är att de inte kan reagera när de underliggande förhållandena i data förändras vilket resulterar i en försämrad prediktiv prestanda för modellerna. För att lösa detta problem försöker olika studier integrera externa faktorer i modellerna som behandlar systemet som en svart låda med en maskininlärningsmetod som genererar komplexa modeller som kräver en stor mängd data för deras inlärning och har liten förklarande kapacitet. I denna uppsatsen har tre olika algoritmer föreslagits för att införliva ytterligare externa faktorer i dessa mönsterbaserade modeller, vilket ger en bra balans mellan prognosnoggrannhet och modelltolkbarhet. Efter att ha använt dessa algoritmer i ett studiefall av prognoser för Ethereums pristidsserier, visas det att förutsägelsefelet effektivt kan minskas genom att ta hänsyn till dessa inflytelserika externa faktorer jämfört med traditionella tillvägagångssätt med bibehållen full tolkbarhet av modellen.
APA, Harvard, Vancouver, ISO, and other styles
15

Ahmed, War, and Mehrdad Bahador. "The accuracy of the LSTM model for predicting the S&P 500 index and the difference between prediction and backtesting." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2018. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-229415.

Full text
Abstract:
In this paper the question of the accuracy of the LSTM algorithm for predicting stock prices is being researched. The LSTM algorithm is a form of deep learning algorithm. The algorithm takes in a set of data as inputs and finds a pattern to dissolve an output. Our results point to that using backtesting as the sole method to verify the accuracy of a model can fallible. For the future, researchers should take a fresh approach by using real-time testing. We completed this by letting the algorithm make predictions on future data. For the accuracy of the model we reached the conclusion that having more parameters improves accuracy.<br>I detta arbete forskas det kring hur bra prognoser man kan ge genom att använda sig av LSTM algoritmen för att förutspå aktiekurser. LSTM-algoritmen är en form av djupinlärnigsmetod, där man ger algoritmen en del typer av data som input och hittar ett mönster i datan vilket ger ett resultat. I vårt resultat kom vi fram till man ej ska förlita sig på backtesting för att verifiera sina resultat utan även använda modellen till att göra prognoser på framtida data. Vi kan även tillägga att tillförlitlighet ökar om man använder sig av flera faktorer i modellen.
APA, Harvard, Vancouver, ISO, and other styles
16

Rostami, Jako, and Fredrik Hansson. "Time Series Forecasting of House Prices: An evaluation of a Support Vector Machine and a Recurrent Neural Network with LSTM cells." Thesis, Uppsala universitet, Statistiska institutionen, 2019. http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-385823.

Full text
Abstract:
In this thesis, we examine the performance of different forecasting methods. We use dataof monthly house prices from the larger Stockholm area and the municipality of Uppsalabetween 2005 and early 2019 as the time series to be forecast. Firstly, we compare theperformance of two machine learning methods, the Long Short-Term Memory, and theSupport Vector Machine methods. The two methods forecasts are compared, and themodel with the lowest forecasting error measured by three metrics is chosen to be comparedwith a classic seasonal ARIMA model. We find that the Long Short-Term Memorymethod is the better performing machine learning method for a twelve-month forecast,but that it still does not forecast as well as the ARIMA model for the same forecast period.
APA, Harvard, Vancouver, ISO, and other styles
17

Anstensrud, Ole-Petter Bård. "Pricing a Bermudan Swaption using the LIBOR Market Model : A LSM approach." Thesis, Norwegian University of Science and Technology, Department of Mathematical Sciences, 2008. http://urn.kb.se/resolve?urn=urn:nbn:no:ntnu:diva-9787.

Full text
Abstract:
<p>This study will focus on the pricing of interest rate derivatives within the framework of the LIBOR Market Model. First we introduce the mathematical and financial foundations behind the basic theory. Then we give a rather rigouros introduction to the LIBOR Market Model and show how to calibrate the model to a real data set. We use the model to price a basic swaption contract before we choose to concentrate on a more exotic Bermudan swaption. We use the Least Squares Monte Carlo (LSM) algorithm to handle the early exercise features of the Bermuda swaption. All major results are vizualised and the C++ implementation code is enclosed in appendix B.</p>
APA, Harvard, Vancouver, ISO, and other styles
18

Goncalves, Juliana Bittencourt. "Calibração do modelo de superfície noah lsm: aplicação em uma região agrícola no sul do Brasil." Universidade Federal de Santa Maria, 2016. http://repositorio.ufsm.br/handle/1/3934.

Full text
Abstract:
Conselho Nacional de Desenvolvimento Científico e Tecnológico<br>In this study, simulated to net radiation and energy flux in a region with rotation of crops, for two distinct periods: Period 1 (01 / Feb / 2009 to 31 / Jan / 2010) and period 2 (14 / Dec / 2009 to 28 / Apr / 2010). In these simulations we used the NOAH LSM surface model. For the period 1 initially, without any calibration simulations were performed only with the input of the local weather conditions, an adjustment of an experiment controlfile file and spin up for the stabilization of the initial conditions. In these simulations, the results were very poor, indicating a need to test the sensitivity of the model especially because of the launch conditions of temperature and soil moisture. After these tests it was found that the initial predictions of impact can be considerable conditions for the two cases. It is noticed that the soil moisture changes generate greater impact in the model that temperature variations boot. As a result, proposed a calibration for the model. The calibration method was to make some simulations manually varying the parameters of soil and vegetation, or both, according to the deficiencies of the NOAH LSM. The tests were carried out until they could get a more optimized forecast for the period studied. The initial analysis of the local conditions of the experimental site was very important for calibration, as it allowed establishing previous parameters corresponding to values close to those parameters when calibrated. Simulation results after calibration applied satisfactorily exhibited liquid radiation and heat flows. So it can be said that the calibration is proposed representing characteristics of vegetation and soil correctly. Nevertheless, the corrections that the model still needs, especially in sensible and latent heat fluxes, may be associated with representation in heat distribution processes and water, or by the fact that the colder months had considerable regime rains. So when there is cloud cover, the model still has problems in representation. Results for long periods of data, as in this work, may lose some of representativeness due to the seasonality of the vegetation parameters, for which varied the parameters for periods culture and fallow. The most important contribution made in this work was a model fit for an agricultural ecosystem area and validate it for the future, it may be used as an initial boundary condition in numerical weather prediction models. The implementation variations in LAI and albedo parameter applied in the simulations of period 2 (soybean) improved the description of the heat flux and net radiation.<br>Neste estudo simulou-se a radiação líquida e os fluxos de energia para uma região com rotações de cultivos agrícolas, para dois períodos distintos: Período 1 (01/Fev/2009 até 31/Jan/2010) e período 2 (14/Dez/2009 até 28/Abr/2010). Nestas simulações utilizou-se o modelo de superfície NOAH LSM. Inicialmente, para o período 1, foram feitas simulações sem nenhuma calibração, apenas com a entrada das condições meteorológicas locais, um ajuste do arquivo controlfile e um experimento spin up para a estabilização das condições iniciais. Nestas simulações, os resultados foram muito insatisfatórios, indicando uma necessidade de testar a sensibilidade do modelo principalmente frente às condições de inicialização da temperatura e da umidade do solo. Após estes testes verificou-se que impactos das previsões às condições iniciais podem ser consideráveis para os dois casos. Percebe-se que as variações de umidade do solo geram maior impacto no modelo devido à temperatura do solo que é simulada. Na sequência, propôs-se uma calibração para o modelo. O método de calibração consistiu em fazer algumas simulações variando-se manualmente os parâmetros de solo e vegetação, ou ambos, de acordo com as deficiências do NOAH LSM. Assim, os testes foram realizados até que se conseguisse uma previsão mais otimizada para o período estudado. A análise inicial das condições locais do sítio experimental foi de suma importância para a calibração, pois ela possibilitou estabelecer parâmetros prévios que correspondem a valores próximos dos parâmetros quando calibrados. Os resultados das simulações, após a calibração aplicada, representaram satisfatoriamente a radiação líquida e os fluxos de calor. Portanto, pode-se dizer que a calibração proposta está representando as características de vegetação e de solo de forma correta. Apesar disso, as correções que o modelo ainda necessita, principalmente nos fluxos de calor sensível e latente, podem estar associadas a representação nos processos de distribuição do calor e da água, ou ainda pelo fato de que os meses mais frios tiveram um considerável regime de chuvas. Assim, quando há nebulosidade, o modelo ainda apresenta problemas na representação. Os resultados para períodos longos de dados, os quais foram considerados neste trabalho podem perder um pouco da representatividade em função da sazonalidade dos parâmetros de vegetação, motivo pelo qual variou-se os parâmetros para períodos com cultura e com pousios. A contribuição mais importante realizada neste trabalho foi um ajuste do modelo para uma região de ecossistema agrícola e a sua validação para que futuramente, possa ser utilizado como condição de contorno inicial em modelos de previsão numérica do Tempo. A implementação das variações diárias no parâmetro IAF e no albedo, aplicada nas simulações do período 2 na cultura de soja, melhorou a descrição dos fluxos de calor e da radiação líquida.
APA, Harvard, Vancouver, ISO, and other styles
19

Cheng, Kun. "Deformable models for adaptive radiotherapy planning." Thesis, University of Edinburgh, 2016. http://hdl.handle.net/1842/22893.

Full text
Abstract:
Radiotherapy is the most widely used treatment for cancer, with 4 out of 10 cancer patients receiving radiotherapy as part of their treatment. The delineation of gross tumour volume (GTV) is crucial in the treatment of radiotherapy. An automatic contouring system would be beneficial in radiotherapy planning in order to generate objective, accurate and reproducible GTV contours. Image guided radiotherapy (IGRT) acquires patient images just before treatment delivery to allow any necessary positional correction. Consequently, real-time contouring system provides an opportunity to adopt radiotherapy on the treatment day. In this thesis, freely deformable models (FDM) and shape constrained deformable models (SCDMs) were used to automatically delineate the GTV for brain cancer and prostate cancer. Level set method (LSM) is a typical FDM which was used to contour glioma on brain MRI. A series of low level image segmentation methodologies are cascaded to form a case-wise fully automatic initialisation pipeline for the level set function. Dice similarity coefficients (DSCs) were used to evaluate the contours. Results shown a good agreement between clinical contours and LSM contours, in 93% of cases the DSCs was found to be between 60% and 80%. The second significant contribution is a novel development to the active shape model (ASM), a profile feature was selected from pre-computed texture features by minimising the Mahalanobis distance (MD) to obtain the most distinct feature for each landmark, instead of conventional image intensity. A new group-wise registration scheme was applied to solve the correspondence definition within the training data. This ASM model was used to delineated prostate GTV on CT. DSCs for this case was found between 0.75 and 0.91 with the mean DSC 0.81. The last contribution is a fully automatic active appearance model (AAM) which captures image appearance near the GTV boundary. The image appearance of inner GTV was discarded to spare the potential disruption caused by brachytherapy seeds or gold markers. This model outperforms conventional AAM at the prostate base and apex region by involving surround organs. The overall mean DSC for this case is 0.85.
APA, Harvard, Vancouver, ISO, and other styles
20

Van, den Bergh F., Wyk MA Van, Wyk BJ Van, and G. Udahemuka. "A comparison of data-driven and model-driven approaches to brightness temperature diurnal cycle interpolation." SAIEE Africa Research Journal, 2007. http://encore.tut.ac.za/iii/cpro/DigitalItemViewPage.external?sp=1001082.

Full text
Abstract:
This paper presents two new schemes for interpolating missing samples in satellite diurnal temperature cycles (DTCs). The first scheme, referred to here as the cosine model, is an improvement of the model proposed in [2] and combines a cosine and exponential function for modelling the DTC. The second scheme uses the notion of a Reproducing Kernel Hilbert Space (RKHS) interpolator [1] for interpolating the missing samples. The application of RKHS interpolators to the DTC interpolation problem is novel. Results obtained by means of computer experiments are presented.
APA, Harvard, Vancouver, ISO, and other styles
21

Mukhedkar, Dhananjay. "Polyphonic Music Instrument Detection on Weakly Labelled Data using Sequence Learning Models." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-279060.

Full text
Abstract:
Polyphonic or multiple music instrument detection is a difficult problem compared to detecting single or solo instruments in an audio recording. As music is time series data it be can modelled using sequence learning methods within deep learning. Recently, temporal convolutional networks (TCN) have shown to outperform conventional recurrent neural networks (RNN) on various sequence modelling tasks. Though there have been significant improvements in deep learning methods, data scarcity becomes a problem in training large scale models. Weakly labelled data is an alternative where a clip is annotated for presence or absence of instruments without specifying the times at which an instrument is sounding. This study investigates how TCN model compares to a Long Short-Term Memory (LSTM) model while trained on weakly labelled dataset. The results showed successful training of both models along with generalisation on a separate dataset. The comparison showed that TCN performed better than LSTM, but only marginally. Therefore, from the experiments carried out it could not be explicitly concluded if TCN is convincingly a better choice over LSTM in the context of instrument detection, but definitely a strong alternative.<br>Polyfonisk eller multipel musikinstrumentdetektering är ett svårt problem jämfört med att detektera enstaka eller soloinstrument i en ljudinspelning. Eftersom musik är tidsseriedata kan den modelleras med hjälp av sekvensinlärningsmetoder inom djup inlärning. Nyligen har ’Temporal Convolutional Network’ (TCN) visat sig överträffa konventionella ’Recurrent Neural Network’ (RNN) på flertalet sekvensmodelleringsuppgifter. Även om det har skett betydande förbättringar i metoder för djup inlärning, blir dataknapphet ett problem vid utbildning av storskaliga modeller. Svagt märkta data är ett alternativ där ett klipp kommenteras för närvaro av frånvaro av instrument utan att ange de tidpunkter då ett instrument låter. Denna studie undersöker hur TCN-modellen jämförs med en ’Long Short-Term Memory’ (LSTM) -modell medan den tränas i svagt märkta datasätt. Resultaten visade framgångsrik utbildning av båda modellerna tillsammans med generalisering i en separat datasats. Jämförelsen visade att TCN presterade bättre än LSTM, men endast marginellt. Därför kan man från de genomförda experimenten inte uttryckligen dra slutsatsen om TCN övertygande är ett bättre val jämfört med LSTM i samband med instrumentdetektering, men definitivt ett starkt alternativ.
APA, Harvard, Vancouver, ISO, and other styles
22

Ujihara, Rintaro. "Multi-objective optimization for model selection in music classification." Thesis, KTH, Optimeringslära och systemteori, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-298370.

Full text
Abstract:
With the breakthrough of machine learning techniques, the research concerning music emotion classification has been getting notable progress combining various audio features and state-of-the-art machine learning models. Still, it is known that the way to preprocess music samples and to choose which machine classification algorithm to use depends on data sets and the objective of each project work. The collaborating company of this thesis, Ichigoichie AB, is currently developing a system to categorize music data into positive/negative classes. To enhance the accuracy of the existing system, this project aims to figure out the best model through experiments with six audio features (Mel spectrogram, MFCC, HPSS, Onset, CENS, Tonnetz) and several machine learning models including deep neural network models for the classification task. For each model, hyperparameter tuning is performed and the model evaluation is carried out according to pareto optimality with regard to accuracy and execution time. The results show that the most promising model accomplished 95% correct classification with an execution time of less than 15 seconds.<br>I och med genombrottet av maskininlärningstekniker har forskning kring känsloklassificering i musik sett betydande framsteg genom att kombinera olikamusikanalysverktyg med nya maskinlärningsmodeller. Trots detta är hur man förbehandlar ljuddatat och valet av vilken maskinklassificeringsalgoritm som ska tillämpas beroende på vilken typ av data man arbetar med samt målet med projektet. Denna uppsats samarbetspartner, Ichigoichie AB, utvecklar för närvarande ett system för att kategorisera musikdata enligt positiva och negativa känslor. För att höja systemets noggrannhet är målet med denna uppsats att experimentellt hitta bästa modellen baserat på sex musik-egenskaper (Mel-spektrogram, MFCC, HPSS, Onset, CENS samt Tonnetz) och ett antal olika maskininlärningsmodeller, inklusive Deep Learning-modeller. Varje modell hyperparameteroptimeras och utvärderas enligt paretooptimalitet med hänsyn till noggrannhet och beräkningstid. Resultaten visar att den mest lovande modellen uppnådde 95% korrekt klassificering med en beräkningstid på mindre än 15 sekunder.
APA, Harvard, Vancouver, ISO, and other styles
23

Almqvist, Olof. "A comparative study between algorithms for time series forecasting on customer prediction : An investigation into the performance of ARIMA, RNN, LSTM, TCN and HMM." Thesis, Högskolan i Skövde, Institutionen för informationsteknologi, 2019. http://urn.kb.se/resolve?urn=urn:nbn:se:his:diva-16974.

Full text
Abstract:
Time series prediction is one of the main areas of statistics and machine learning. In 2018 the two new algorithms higher order hidden Markov model and temporal convolutional network were proposed and emerged as challengers to the more traditional recurrent neural network and long-short term memory network as well as the autoregressive integrated moving average (ARIMA). In this study most major algorithms together with recent innovations for time series forecasting is trained and evaluated on two datasets from the theme park industry with the aim of predicting future number of visitors. To develop models, Python libraries Keras and Statsmodels were used. Results from this thesis show that the neural network models are slightly better than ARIMA and the hidden Markov model, and that the temporal convolutional network do not perform significantly better than the recurrent or long-short term memory networks although having the lowest prediction error on one of the datasets. Interestingly, the Markov model performed worse than all neural network models even when using no independent variables.
APA, Harvard, Vancouver, ISO, and other styles
24

Mirshekarianbabaki, Sadegh. "Blood Glucose Level Prediction via Seamless Incorporation of Raw Features Using RNNs." Ohio University / OhioLINK, 2018. http://rave.ohiolink.edu/etdc/view?acc_num=ohiou1523988526094778.

Full text
APA, Harvard, Vancouver, ISO, and other styles
25

Sibelius, Parmbäck Sebastian. "HMMs and LSTMs for On-line Gesture Recognition on the Stylaero Board : Evaluating and Comparing Two Methods." Thesis, Linköpings universitet, Artificiell intelligens och integrerade datorsystem, 2019. http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-162237.

Full text
Abstract:
In this thesis, methods of implementing an online gesture recognition system for the novel Stylaero Board device are investigated. Two methods are evaluated - one based on LSTMs and one based on HMMs - on three kinds of gestures: Tap, circle, and flick motions. A method’s performance was measured in its accuracy in determining both whether any of the above listed gestures were performed and, if so, which gesture, in an online single-pass scenario. Insight was acquired regarding the technical challenges and possible solutions to the online aspect of the problem. Poor performance was, however, observed in both methods, with a likely culprit identified as low quality of training data, due to an arduous and complex gesture performance capturing process. Further research improving on the process of gathering data is suggested.
APA, Harvard, Vancouver, ISO, and other styles
26

Annecchini, Andrea. "Valutazione di derivati Americani in modelli multi-dimensionali." Master's thesis, Alma Mater Studiorum - Università di Bologna, 2019. http://amslaurea.unibo.it/20743/.

Full text
Abstract:
La valutazione del prezzo di opzioni americane è un problema molto diffuso in ambito finanziario. In questo elaborato viene presentato l'algoritmo LSM basato sulla simulazione e una sua modifica che permette di eliminare un tipo di bias. Viene studiata la convergenza dei coefficienti di regressione che compaiono nell'algoritmo e vengono fatti diversi esperimenti numerici per valutare alcuni tipi di opzioni americane.
APA, Harvard, Vancouver, ISO, and other styles
27

Santos, Julio Cesar Grimalt dos. "Cálculo do Value at Risk (VaR) para o Ibovespa, pós crise de 2008, por meio dos modelos de heterocedasticidade condicional (GARCH) e de volatilidade estocástica (Local Scale Model - LSM)." reponame:Repositório Institucional do FGV, 2015. http://hdl.handle.net/10438/13521.

Full text
Abstract:
Submitted by JULIO CESAR GRIMALT DOS SANTOS (grimbil@hotmail.com) on 2015-02-23T21:08:49Z No. of bitstreams: 1 Dissertação Final.pdf: 1416129 bytes, checksum: fcbac3f948355bac6f5b59569bf2610a (MD5)<br>Approved for entry into archive by Janete de Oliveira Feitosa (janete.feitosa@fgv.br) on 2015-03-04T16:04:21Z (GMT) No. of bitstreams: 1 Dissertação Final.pdf: 1416129 bytes, checksum: fcbac3f948355bac6f5b59569bf2610a (MD5)<br>Approved for entry into archive by Marcia Bacha (marcia.bacha@fgv.br) on 2015-03-12T19:06:25Z (GMT) No. of bitstreams: 1 Dissertação Final.pdf: 1416129 bytes, checksum: fcbac3f948355bac6f5b59569bf2610a (MD5)<br>Made available in DSpace on 2015-03-12T19:06:41Z (GMT). No. of bitstreams: 1 Dissertação Final.pdf: 1416129 bytes, checksum: fcbac3f948355bac6f5b59569bf2610a (MD5) Previous issue date: 2015-02-10<br>O objetivo deste estudo é propor a implementação de um modelo estatístico para cálculo da volatilidade, não difundido na literatura brasileira, o modelo de escala local (LSM), apresentando suas vantagens e desvantagens em relação aos modelos habitualmente utilizados para mensuração de risco. Para estimação dos parâmetros serão usadas as cotações diárias do Ibovespa, no período de janeiro de 2009 a dezembro de 2014, e para a aferição da acurácia empírica dos modelos serão realizados testes fora da amostra, comparando os VaR obtidos para o período de janeiro a dezembro de 2014. Foram introduzidas variáveis explicativas na tentativa de aprimorar os modelos e optou-se pelo correspondente americano do Ibovespa, o índice Dow Jones, por ter apresentado propriedades como: alta correlação, causalidade no sentido de Granger, e razão de log-verossimilhança significativa. Uma das inovações do modelo de escala local é não utilizar diretamente a variância, mas sim a sua recíproca, chamada de 'precisão' da série, que segue uma espécie de passeio aleatório multiplicativo. O LSM captou todos os fatos estilizados das séries financeiras, e os resultados foram favoráveis a sua utilização, logo, o modelo torna-se uma alternativa de especificação eficiente e parcimoniosa para estimar e prever volatilidade, na medida em que possui apenas um parâmetro a ser estimado, o que representa uma mudança de paradigma em relação aos modelos de heterocedasticidade condicional.<br>The objective of this study is to propose the implementation of a statistical model to calculate the volatility not widespread in Brazilian literature, LSM, with its advantages and disadvantages compared to the models commonly used for risk measurement. To estimate the parameters will be used daily prices of Ibovespa in the period from January 2009 to December 2014, and to measure the empirical accuracy of the models out of sample tests will be performed, comparing the VaR obtained for the period from January to December 2014. Explanatory variables were introduced in an attempt to improve the models, and we chose to its corresponding American Ibovespa, the Dow Jones index, for presenting characteristics such as high correlation, causality in the Granger sense, and reason for significant log-likelihood. One of the local scale model innovation is not directly use the variance, but its reciprocal, called 'precision' series, which follows a kind of multiplicative random walk. LSM captured all financial series of stylized facts, and the results were favorable to use, so the model becomes an efficient and economical alternative specification for estimating and predicting volatility, to the extent that only one parameter has to be estimated, which represents a paradigm shift in the models of conditional heteroscedasticity.
APA, Harvard, Vancouver, ISO, and other styles
28

Albert, Florea George, and Filip Weilid. "Deep Learning Models for Human Activity Recognition." Thesis, Malmö universitet, Fakulteten för teknik och samhälle (TS), 2019. http://urn.kb.se/resolve?urn=urn:nbn:se:mau:diva-20201.

Full text
Abstract:
AMI Meeting Corpus (AMI) -databasen används för att undersöka igenkännande av gruppaktivitet. AMI Meeting Corpus (AMI) -databasen ger forskare fjärrstyrda möten och naturliga möten i en kontorsmiljö; mötescenario i ett fyra personers stort kontorsrum. För attuppnågruppaktivitetsigenkänninganvändesbildsekvenserfrånvideosoch2-dimensionella audiospektrogram från AMI-databasen. Bildsekvenserna är RGB-färgade bilder och ljudspektrogram har en färgkanal. Bildsekvenserna producerades i batcher så att temporala funktioner kunde utvärderas tillsammans med ljudspektrogrammen. Det har visats att inkludering av temporala funktioner både under modellträning och sedan förutsäga beteende hos en aktivitet ökar valideringsnoggrannheten jämfört med modeller som endast använder rumsfunktioner[1]. Deep learning arkitekturer har implementerats för att känna igen olika mänskliga aktiviteter i AMI-kontorsmiljön med hjälp av extraherade data från the AMI-databas.Neurala nätverks modellerna byggdes med hjälp av KerasAPI tillsammans med TensorFlow biblioteket. Det finns olika typer av neurala nätverksarkitekturer. Arkitekturerna som undersöktes i detta projektet var Residual Neural Network, Visual GeometryGroup 16, Inception V3 och RCNN (LSTM). ImageNet-vikter har använts för att initialisera vikterna för Neurala nätverk basmodeller. ImageNet-vikterna tillhandahålls av Keras API och är optimerade för varje basmodell [2]. Basmodellerna använder ImageNet-vikter när de extraherar funktioner från inmatningsdata. Funktionsextraktionen med hjälp av ImageNet-vikter eller slumpmässiga vikter tillsammans med basmodellerna visade lovande resultat. Både Deep Learning användningen av täta skikt och LSTM spatio-temporala sekvens predikering implementerades framgångsrikt.<br>The Augmented Multi-party Interaction(AMI) Meeting Corpus database is used to investigate group activity recognition in an office environment. The AMI Meeting Corpus database provides researchers with remote controlled meetings and natural meetings in an office environment; meeting scenario in a four person sized office room. To achieve the group activity recognition video frames and 2-dimensional audio spectrograms were extracted from the AMI database. The video frames were RGB colored images and audio spectrograms had one color channel. The video frames were produced in batches so that temporal features could be evaluated together with the audio spectrogrames. It has been shown that including temporal features both during model training and then predicting the behavior of an activity increases the validation accuracy compared to models that only use spatial features [1]. Deep learning architectures have been implemented to recognize different human activities in the AMI office environment using the extracted data from the AMI database.The Neural Network models were built using the Keras API together with TensorFlow library. There are different types of Neural Network architectures. The architecture types that were investigated in this project were Residual Neural Network, Visual Geometry Group 16, Inception V3 and RCNN(Recurrent Neural Network). ImageNet weights have been used to initialize the weights for the Neural Network base models. ImageNet weights were provided by Keras API and was optimized for each base model[2]. The base models uses ImageNet weights when extracting features from the input data.The feature extraction using ImageNet weights or random weights together with the base models showed promising results. Both the Deep Learning using dense layers and the LSTM spatio-temporal sequence prediction were implemented successfully.
APA, Harvard, Vancouver, ISO, and other styles
29

Keisala, Simon. "Using a Character-Based Language Model for Caption Generation." Thesis, Linköpings universitet, Interaktiva och kognitiva system, 2019. http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-163001.

Full text
Abstract:
Using AI to automatically describe images is a challenging task. The aim of this study has been to compare the use of character-based language models with one of the current state-of-the-art token-based language models, im2txt, to generate image captions, with focus on morphological correctness. Previous work has shown that character-based language models are able to outperform token-based language models in morphologically rich languages. Other studies show that simple multi-layered LSTM-blocks are able to learn to replicate the syntax of its training data. To study the usability of character-based language models an alternative model based on TensorFlow im2txt has been created. The model changes the token-generation architecture into handling character-sized tokens instead of word-sized tokens. The results suggest that a character-based language model could outperform the current token-based language models, although due to time and computing power constraints this study fails to draw a clear conclusion. A problem with one of the methods, subsampling, is discussed. When using the original method on character-sized tokens this method removes characters (including special characters) instead of full words. To solve this issue, a two-phase approach is suggested, where training data first is separated into word-sized tokens where subsampling is performed. The remaining tokens are then separated into character-sized tokens. Future work where the modified subsampling and fine-tuning of the hyperparameters are performed is suggested to gain a clearer conclusion of the performance of character-based language models.
APA, Harvard, Vancouver, ISO, and other styles
30

Boukachaba, Niama. "Apport des observations satellitaires hyperspectrales infrarouges IASI au-dessus des continents dans le modèle météorologique à échelle convective AROME." Phd thesis, Toulouse, INPT, 2017. http://oatao.univ-toulouse.fr/19257/1/BOUKACHABA_Niama.pdf.

Full text
Abstract:
Le sondeur infrarouge hyperspectral IASI (Interféromètre Atmosphérique de Sondage Infrarouge, développé conjointement par le CNES et EUMETSAT et embarqué à bord des satellites défilants Metop A, Metop B et très prochainement Metop C (2006, 2012 et 2018, respectivement)) apporte une très grande quantité d’informations permettant, entre autres, de décrire finement les paramètres de surface (température et émissivité sur une large gamme de longueurs d’onde). Néanmoins, les prévisions de température des surfaces continentales ne sont pas encore suffisamment réalistes pour utiliser l’information infrarouge en basse troposphère et proche de la surface au-dessus des continents car les radiances sensibles à ces régions sont fortement affectées par la variation des paramètres de surface (tels que la température, l’émissivité et l’humidité) et par la présence des nuages. Ceci peut conduire à un écart parfois important entre les observations et les simulations, conduisant à un rejet important des observations et à une mauvaise détection nuageuse. De ce fait, l’objectif principal de la thèse est l’amélioration des analyses et des prévisions par l’augmentation des observations IASI assimilées sur les continents dans le modèle à aire limitée AROME. La première partie du travail s’est focalisée sur l’identification du canal IASI le plus approprié à la restitution de la tempèrature de surface (Ts). En poursuivant les travaux de thèse de [Vincensini, 2013], cinq canaux IASI localisés entre 901.50 cm−1 et 1115.75 cm−1 ont été sélectionnés pour une meilleure prise en compte des basses couches de l’atmosphère plus particulièrement en termes de température et d’humidité. La restitution de la Ts s’est faite par inversion de l’équation du transfert radiatif [Karbou et al., 2006] en utilisant le modèle de transfert radiatif RTTOV et l’atlas d’émissivité développé par l’université de Wisconsin. Le canal IASI 1194 (943.25 cm−1) a été retenu pour la restitution des Ts suite à une série de comparaisons effectuées entre la Ts restituée à partir des différents canaux IASI sélectionnés et celle de l’ébauche. Aussi, des comparaisons ont été réalisées entre les Ts restituées à partir de IASI et celles restituées à partir de SEVIRI et de AVHRR. La seconde partie du travail a reposé sur l’étude de l’impact de l’utilisation de la Ts restituée à partir du canal IASI 1194 dans les processus de simulation et d’assimilation des canaux IASI utilisés dans les modèles de prévision numérique du temps de Météo-France (AROME et ARPEGE). La Ts restituée à partir du canal IASI 1194 a été intégrée dans le modèle RTTOV pour améliorer les simulations des autres observations IASI sensibles à la surface. L’impact sur la détection nuageuse issue de l’algorithme de [McNally and Watts, 2003] a également été évalué. Par la suite, d’autres expériences ont été menés pour étudier l’impact de l’utilisation des Ts restituées sur l’assimilation de données et sur l’amélioration de la sélection des canaux IASI sur terre dans le modèle AROME. L’impact sur les analyses et les prévisions ont été également décrits.
APA, Harvard, Vancouver, ISO, and other styles
31

Jain, Monika. "Regularized ensemble correlation filter tracking." Thesis, Queensland University of Technology, 2022. https://eprints.qut.edu.au/229266/1/Monika_Jain_Thesis.pdf.

Full text
Abstract:
Visual Object Tracking is the task of tracking an object within a video. Broadly, most tracking algorithms can be classified into neural network based, correlation filter based, and hybrid. This thesis investigates various methods to improve tracking using correlation filters. The thesis contributes four novel trackers. The first tracker uses an appearance model pool to avoid faulty filter updates. Next, the appearance feature channel weights are learned using the graph-based similarity followed by modelling sparse spatio-temporal variations. At last, non-linearity of the appearance features is captured. The thesis also presents extensive evaluation of the proposed trackers on standard datasets.
APA, Harvard, Vancouver, ISO, and other styles
32

Ridhagen, Markus, and Petter Lind. "A comparative study of Neural Network Forecasting models on the M4 competition data." Thesis, Uppsala universitet, Statistiska institutionen, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-445568.

Full text
Abstract:
The development of machine learning research has provided statistical innovations and further developments within the field of time series analysis. This study seeks to investigate two different approaches on artificial neural network models based on different learning techniques, and answering how well the neural network approach compares with a basic autoregressive approach, as well as how the artificial neural network models compare to each other. The models were compared and analyzed in regards to the univariate forecast accuracy on 20 randomly drawn time series from two different time frequencies from the M4 competition dataset. Forecasting was made dependent on one time lag (t-1) and forecasted three and six steps ahead respectively. The artificial neural network models outperformed the baseline Autoregressive model, showing notably lower mean average percentage error overall. The Multilayered perceptron models performed better than the Long short-term memory model overall, whereas the Long short-term memory model showed improvement on longer prediction time dimensions. As the training were done univariately  on a limited set of time steps, it is believed that the one layered-approach gave a good enough approximation on the data, whereas the added layer couldn’t fully utilize its strengths of processing power. Likewise, the Long short-term memory model couldn’t fully demonstrate the advantagements of recurrent learning. Using the same dataset, further studies could be made with another approach to data processing. Implementing an unsupervised approach of clustering the data before analysis, the same models could be tested with multivariate analysis on models trained on multiple time series simultaneously.
APA, Harvard, Vancouver, ISO, and other styles
33

Mealey, Thomas C. "Binary Recurrent Unit: Using FPGA Hardware to Accelerate Inference in Long Short-Term Memory Neural Networks." University of Dayton / OhioLINK, 2018. http://rave.ohiolink.edu/etdc/view?acc_num=dayton1524402925375566.

Full text
APA, Harvard, Vancouver, ISO, and other styles
34

Li, Yuntao. "Federated Learning for Time Series Forecasting Using Hybrid Model." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2019. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-254677.

Full text
Abstract:
Time Series data has become ubiquitous thanks to affordable edge devices and sensors. Much of this data is valuable for decision making. In order to use these data for the forecasting task, the conventional centralized approach has shown deficiencies regarding large data communication and data privacy issues. Furthermore, Neural Network models cannot make use of the extra information from the time series, thus they usually fail to provide time series specific results. Both issues expose a challenge to large-scale Time Series Forecasting with Neural Network models. All these limitations lead to our research question:Can we realize decentralized time series forecasting with a Federated Learning mechanism that is comparable to the conventional centralized setup in forecasting performance?In this work, we propose a Federated Series Forecasting framework, resolving the challenge by allowing users to keep the data locally, and learns a shared model by aggregating locally computed updates. Besides, we design a hybrid model to enable Neural Network models utilizing the extra information from the time series to achieve a time series specific learning. In particular, the proposed hybrid outperforms state-of-art baseline data-central models with NN5 and Ericsson KPI data. Meanwhile, the federated settings of purposed model yields comparable results to data-central settings on both NN5 and Ericsson KPI data. These results together answer the research question of this thesis.<br>Tidseriedata har blivit allmänt förekommande tack vare överkomliga kantenheter och sensorer. Mycket av denna data är värdefull för beslutsfattande. För att kunna använda datan för prognosuppgifter har den konventionella centraliserade metoden visat brister avseende storskalig datakommunikation och integritetsfrågor. Vidare har neurala nätverksmodeller inte klarat av att utnyttja den extra informationen från tidsserierna, vilket leder till misslyckanden med att ge specifikt tidsserierelaterade resultat. Båda frågorna exponerar en utmaning för storskalig tidsserieprognostisering med neurala nätverksmodeller. Alla dessa begränsningar leder till vår forskningsfråga:Kan vi realisera decentraliserad tidsserieprognostisering med en federerad lärningsmekanism som presterar jämförbart med konventionella centrala lösningar i prognostisering?I det här arbetet föreslår vi ett ramverk för federerad tidsserieprognos som löser utmaningen genom att låta användaren behålla data lokalt och lära sig en delad modell genom att aggregera lokalt beräknade uppdateringar. Dessutom utformar vi en hybrid modell för att möjliggöra neurala nätverksmodeller som kan utnyttja den extra informationen från tidsserierna för att uppnå inlärning av specifika tidsserier. Den föreslagna hybrida modellen presterar bättre än state-of-art centraliserade grundläggande modeller med NN5och Ericsson KPIdata. Samtidigt ger den federerade ansatsen jämförbara resultat med de datacentrala ansatserna för både NN5och Ericsson KPI-data. Dessa resultat svarar tillsammans på forskningsfrågan av denna avhandling.
APA, Harvard, Vancouver, ISO, and other styles
35

Nováčik, Tomáš. "Rekurentní neuronové sítě pro rozpoznávání řeči." Master's thesis, Vysoké učení technické v Brně. Fakulta informačních technologií, 2016. http://www.nusl.cz/ntk/nusl-255371.

Full text
Abstract:
This master thesis deals with the implementation of various types of recurrent neural networks via programming language lua using torch library. It focuses on finding optimal strategy for training recurrent neural networks and also tries to minimize the duration of the training. Furthermore various types of regularization techniques are investigated and implemented into the recurrent neural network architecture. Implemented recurrent neural networks are compared on the speech recognition task using AMI dataset, where they model the acustic information. Their performance is also compared to standard feedforward neural network. Best results are achieved using BLSTM architecture. The recurrent neural network are also trained via CTC objective function on the TIMIT dataset. Best result is again achieved using BLSTM architecture.
APA, Harvard, Vancouver, ISO, and other styles
36

Burg, Antoine. "Multivariate extensions for mortality modelling." Electronic Thesis or Diss., Université Paris sciences et lettres, 2025. http://www.theses.fr/2025UPSLD002.

Full text
Abstract:
Au cours des deux derniers siècles, l’espérance de vie tout autour du globe a connu un accroissement considérable. Si la tendance sur le long terme est plutôt régulière, l’amélioration de la longévité peut être décomposée sur le court-terme en plusieurs phases, que l’on peut relier le plus souvent aux progrès médicaux et à la diminution de causes de mortalité particulières. L’année 2020 marque un tournant du fait de l’ampleur de la pandémie Covid-19 et de ses conséquences. Ses effets directs et indirects sur l’économie et les systèmes de santé se manifestent également au travers des autres causes majeures de décès. Pour comprendre et anticiper les risques liés à la mortalité, il devient de plus en plus nécessaire pour les acteurs de la réassurance de raisonner et modéliser en termes de causes de décès. Ce type de modélisation pose néanmoins des défis spécifiques, issues de la nature multivariée des modèles, dont la complexité dépasse celle des outils classiques de l’actuaire. Nous proposons dans cette thèse plusieurs axes pour étendre la modélisation de la mortalité à un cadre multivarié. Ces axes sont abordés sous forme d’articles de recherche. La première étude porte sur des aspects techniques des distributions multivariées au sein de modèles linéaires généralisés. Lorsque les variables explicatives sont catégorielles, nous proposons de nouveaux estimateurs pour les distributions multinomiale, multinomiale négative et de Dirichlet sous forme de formules fermées, qui permettent notamment un gain considérable en temps de calcul. Ces estimateurs sont utilisés dans la seconde étude pour proposer une nouvelle méthode d’estimation des paramètres de modèles de mortalité. Cette méthode prolonge le cadre existant pour la mortalité toute cause, et permet de traiter toutes les problématiques de modélisation de mortalité en une seule étape, en particulier par cause de décès. Le troisième axe porte sur les projections de mortalité. Nous étudions des réseaux de neurones spécifiquement adaptés aux séries temporelles. Nous montrons par des exemples concrets auxquels peut faire face l’actuaire que ces modèles sont suffisamment flexibles et robustes, offrant une alternative crédible aux modèles classiques<br>Over the past two centuries, life expectancy around the globe has increased considerably. While the long-term trend is fairly regular, the improvement in longevity can be broken down into several phases in the short term, which can most often be linked to medical progress and the reduction in specific causes of mortality. The year 2020 marks a turning point due to the scale of the Covid-19 pandemic and its consequences. Its direct and indirect effects on the economy and healthcare systems will also be felt through other major causes of death. To understand and anticipate mortality-related risks, it is becoming increasingly necessary for reinsurance players to reason and model in terms of causes of death. However, this type of modeling poses specific challenges. By its very nature, it involves multivariate models, whose complexity exceeds that of conventional actuary tools. In this thesis, we propose several avenues for extending mortality modeling to a multivariate framework. These are presented in the form of research articles. The first study deals with technical aspects of multivariate distributions within generalized linear models. When the explanatory variables are categorical, we propose new estimators for the multinomial, negative multinomial and Dirichlet distributions in the form of closed formulas, which notably enable considerable savings in computation time. These estimators are used in the second study to propose a new method for estimating the parameters of mortality models. This method extends the existing framework for all-cause mortality, and enables all mortality modeling issues to be addressed in a single step, particularly by cause-of-death. The third axis concerns mortality forecasts. We study neural networks specifically adapted to time series. Based on concrete use cases, we show that these models are sufficiently flexible and robust to offer a credible alternative to conventional models
APA, Harvard, Vancouver, ISO, and other styles
37

Soares, Leonardo dos Reis Leano. "Estudo do comportamento de sinais OSL de BeO e Al2O3:C usando o Modelo OTOR Simplificado e Método dos Mínimos Quadrados." Universidade de São Paulo, 2018. http://www.teses.usp.br/teses/disponiveis/43/43134/tde-02112018-132426/.

Full text
Abstract:
A dosimetria das radiações alfa, beta e gama é importante para diversas áreas aplicadas, sendo utilizada na proteção radiológica de pacientes e profissionais que se expõem a esses tipos de radiações. Com estudos dosimétricos pode-se obter melhores estimativas de dose absorvida, e ter mais precisão na estimativa de riscos populacionais. As técnicas de Termoluminescência (TL) e Luminescência Oticamente Estimulada (OSL) são utilizadas para essas aplicações dosimétricas. Estudos recentes têm mostrado que alguns materiais dosimétricos conhecidos como óxido de alumínio dopado com carbono (Al$_2$O$_3$:C) e óxido de berílio (BeO) sofrem mudanças no formato observado dos sinais OSL com relação as taxas de dose e tipos de radiação. O principal objetivo desse trabalho foi analisar os formatos dessas curvas e verificar quantitativamente, se existem ou não mudanças nos formatos dos sinais OSL dos dosímetros irradiados com diferentes tipos de radiação e taxas de dose. Sob o modelo de uma armadilha e um centro de recombinação (OTOR) foram estudados os sinais OSL com estímulo contínuo (CW-OSL). O modelo OTOR é simples, mas não possui solução analítica e as soluções computacionais são custosas pelo número grande de variáveis e parâmetros. Nesse trabalho, foi necessário realizar algumas simplificações para obtenção de um modelo ainda mais simples para ajuste nos dados. O modelo OTOR simples apresenta um comportamento de decaimento exponencial na descrição do sinal CW-OSL. Uma outra abordagem de extensão do modelo OTOR-simples foi a utilização do modelo com duas armadilhas independentes e um centro de recombinação, que resulta em dois decaimentos exponenciais. Para obtenção dos parâmetros que descrevem o sinal CW-OSL com esses modelos, foi utilizado o método dos mínimos quadrados (MMQ), com refinamento dos parâmetros pelo método de Gauss. O modelo de dois decaimentos exponenciais mostrou-se superior em qualidade com análise do parâmetro $\\chi^2$ e do comportamento dos resíduos em relação ao modelo de um decaimento exponencial para ambos os materiais utilizados. Com os ajustes, foi possível verificar diferenças nos comportamentos do sinal CW-OSL das amostras irradiadas em diferentes situações. As diferenças observadas nos comportamentos são apresentadas pelos parâmetros de decaimento ou de sinal inicial, ou pelas relações entre esses. Os parâmetros ajustados mostram que os sinais OSL provenientes do Al$_2$O$_3$:C e do BeO irradiados com alfa, beta e gama apresentam diferenças significativas nos comportamentos. As diferenças verificadas pelos ajustes dos sinais CW-OSL apresentados pelos dosímetros irradiados com beta e gama podem ter sido em parte causadas por efeito de fading, que afeta de maneira distinta os formatos das curvas e parâmetros ajustados. Nas irradiações com radiação gama com faixas de doses (de 22 a 122 mGy) e taxas doses absorvidas (de 0.024 a 1.66 Gy/s) não foram observados diferenças significativas nos sinais OSL.<br>The dosimetry of alpha, beta and gamma radiation is important in various applied areas, it is used in radiation protection of patients and professionals who are exposed at this kind of radiation. With dosimetric studies, it is possible to better estimate the absorbed dose, and population risks. Thermoluminescence (TL) and Optically Stimulated Luminescence (OSL) techniques are used for these dosimetric applications. Recent studies have shown that some known dosimetric materials as carbon doped aluminum oxide (Al$_2$O$_3$:C) and berilium oxide (BeO) undergo changes in OSL signal behavior related to dose rates and types of radiation. The main objective of this work was to analise the formats of these curves and quantitatively verify whether or not there are changes in OSL signal of the dosimeters irradiated with different types of radiation and dose rates. Under the model of one trap one recombination center (OTOR) the continuous wave OSL (CW-OSL) signals were studied. The OTOR model is the simplest model, but has no analytical solution and the computational solutions are costly by the large number of variables and parameters. In this work, it was necessary to make some simplifications in order to obtain a simple model that could be fitted to the data. The simple-OTOR model shows an exponential decay behavior in the CW-OSL signal description. Another extension approach to the simple-OTOR model was the model with two independent traps and one recombination center, that results in two exponential decays. To obtain the parameters that describe the CW-OSL signal with these models, the least square method (LSM) was used, with parameter refinement by Gauss method. For both the materials the two exponential decay model proved to be superior in quality to the one exponential decay by the analysis of the parameter $\\chi^2$ and the behavior of the residuals. With the fittings, it was possible to verify differences in the behavior of the CW-OSL signal of the samples irradiated in different situations. These differences observed are presented in the decay or initial signal parameters, or in their ratios. Fitted parameters show that OSL signals from Al$_2$O$_3$:C and BeO irradiated with alpha, beta and gamma exhibit significant differences in behavior. The differences verified by the fittings of the CW-OSL signals presented by beta and gamma irradiated dosimeters may in part have been caused by fading effect, which affects in a different way the shapes of the curves and fitted parameters. Gamma irradiation with dose and absorbed dose rate ranges from 22 to 122 mGy and from 0.024 to 1.66Gy/s respectively did not produce significant differences in OSL signals.
APA, Harvard, Vancouver, ISO, and other styles
38

Lund, Max. "Duplicate Detection and Text Classification on Simplified Technical English." Thesis, Linköpings universitet, Institutionen för datavetenskap, 2019. http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-158714.

Full text
Abstract:
This thesis investigates the most effective way of performing classification of text labels and clustering of duplicate texts in technical documentation written in Simplified Technical English. Pre-trained language models from transformers (BERT) were tested against traditional methods such as tf-idf with cosine similarity (kNN) and SVMs on the classification task. For detecting duplicate texts, vector representations from pre-trained transformer and LSTM models were tested against tf-idf using the density-based clustering algorithms DBSCAN and HDBSCAN. The results show that traditional methods are comparable to pre-trained models for classification, and that using tf-idf vectors with a low distance threshold in DBSCAN is preferable for duplicate detection.
APA, Harvard, Vancouver, ISO, and other styles
39

Alsulami, Khalil Ibrahim D. "Application-Based Network Traffic Generator for Networking AI Model Development." University of Dayton / OhioLINK, 2021. http://rave.ohiolink.edu/etdc/view?acc_num=dayton1619387614152354.

Full text
APA, Harvard, Vancouver, ISO, and other styles
40

Holcner, Jonáš. "Strojový překlad pomocí umělých neuronových sítí." Master's thesis, Vysoké učení technické v Brně. Fakulta informačních technologií, 2018. http://www.nusl.cz/ntk/nusl-386020.

Full text
Abstract:
The goal of this thesis is to describe and build a system for neural machine translation. System is built with recurrent neural networks - encoder-decoder architecture in particular. The result is a nmt library used to conduct experiments with different model parameters. Results of the experiments are compared with system built with the statistical tool Moses.
APA, Harvard, Vancouver, ISO, and other styles
41

Eriksson, Henrik. "Federated Learning in Large Scale Networks : Exploring Hierarchical Federated Learning." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-292744.

Full text
Abstract:
Federated learning faces a challenge when dealing with highly heterogeneous data and it can sometimes be inadequate to adopt an approach where a single model is trained for usage at all nodes in the network. Different approaches have been investigated to succumb this issue such as adapting the trained model to each node and clustering the nodes in the network and train a different model for each cluster where the data is less heterogeneous. In this work we study the possibilities to improve the local model performance utilizing the hierarchical setup that comes with clustering the participating clients in the network. Experiments are carried out featuring a Long Short-Term Memory network to perform time series forecasting to evaluate different approaches utilizing the hierarchical setup and comparing them to standard federated learning approaches. The experiments are done using a dataset collected by Ericsson AB consisting of handovers recorded at base stations in an European city. The hierarchical approaches didn’t show any benefit over common two-level approaches.<br>Federated Learning står inför en utmaning när det gäller att hantera data med en hög grad av heterogenitet och det kan i vissa fall vara olämpligt att använda sig av en approach där en och samma modell är tränad för att användas av alla noder i nätverket. Olika approacher för att hantera detta problem har undersökts som att anpassa den tränade modellen till varje nod och att klustra noderna i nätverket och träna en egen modell för varje kluster inom vilket datan är mindre heterogen. I detta arbete studeras möjligheterna att förbättra prestandan hos de lokala modellerna genom att dra nytta av den hierarkiska anordning som uppstår när de deltagande noderna i nätverket grupperas i kluster. Experiment är utförda med ett Long Short-Term Memory-nätverk för att utföra tidsserieprognoser för att utvärdera olika approacher som drar nytta av den hierarkiska anordningen och jämför dem med vanliga federated learning-approacher. Experimenten är utförda med ett dataset insamlat av Ericsson AB. Det består av "handoversfrån basstationer i en europeisk stad. De hierarkiska approacherna visade inga fördelar jämfört med de vanliga två-nivåapproacherna.
APA, Harvard, Vancouver, ISO, and other styles
42

Ramesh, Chandra D. S. "Turbulent Mixed Convection." Thesis, Indian Institute of Science, 2000. https://etd.iisc.ac.in/handle/2005/236.

Full text
Abstract:
Turbulent mixed convection is a complicated flow where the buoyancy and shear forces compete with each other in affecting the flow dynamics. This thesis deals with the near wall dynamics in a turbulent mixed convection flow over an isothermal horizontal heated plate. We distinguish between two types of mixed convection ; low-speed mixed convection (LSM) and high-speed mixed convection (HSM). In LSM the entire boundary layer, including the near-wall region, is dominated by buoyancy; in HSM the near-wall region, is dominated by shear and the outer region by buoyancy. We show that the value of the parameter (* = ^ determines whether the flow is LSM or HSM. Here yr is the friction length scale and L is the Monin-Obukhov length scale. In the present thesis we proposed a model for the near-wall dynamics in LSM. We assume the coherent structure near-wall for low-speed mixed convection to be streamwise aligned periodic array of laminar plumes and give a 2d model for the near wall dynamics, Here the equation to solve for the streamwise velocity is linear with the vertical and spanwise velocities given by the free convection model of Theerthan and Arakeri [1]. We determine the profiles of streamwise velocity, Reynolds shear stress and RMS of the fluctuations of the three components of velocity. From the model we obtain the scaling for wall shear stress rw as rw oc (UooAT*), where Uoo is the free-stream velocity and AT is the temperature difference between the free-stream and the horizontal surface.A similar scaling for rw was obtained in the experiments of Ingersoll [5] and by Narasimha et al [11] in the atmospheric boundary layer under low wind speed conditions. We also derive a formula for boundary layer thickness 5(x) which predicts the boundary layer growth for the combination free-stream velocity Uoo and AT in the low-speed mixed convection regime.
APA, Harvard, Vancouver, ISO, and other styles
43

Ramesh, Chandra D. S. "Turbulent Mixed Convection." Thesis, Indian Institute of Science, 2000. http://hdl.handle.net/2005/236.

Full text
Abstract:
Turbulent mixed convection is a complicated flow where the buoyancy and shear forces compete with each other in affecting the flow dynamics. This thesis deals with the near wall dynamics in a turbulent mixed convection flow over an isothermal horizontal heated plate. We distinguish between two types of mixed convection ; low-speed mixed convection (LSM) and high-speed mixed convection (HSM). In LSM the entire boundary layer, including the near-wall region, is dominated by buoyancy; in HSM the near-wall region, is dominated by shear and the outer region by buoyancy. We show that the value of the parameter (* = ^ determines whether the flow is LSM or HSM. Here yr is the friction length scale and L is the Monin-Obukhov length scale. In the present thesis we proposed a model for the near-wall dynamics in LSM. We assume the coherent structure near-wall for low-speed mixed convection to be streamwise aligned periodic array of laminar plumes and give a 2d model for the near wall dynamics, Here the equation to solve for the streamwise velocity is linear with the vertical and spanwise velocities given by the free convection model of Theerthan and Arakeri [1]. We determine the profiles of streamwise velocity, Reynolds shear stress and RMS of the fluctuations of the three components of velocity. From the model we obtain the scaling for wall shear stress rw as rw oc (UooAT*), where Uoo is the free-stream velocity and AT is the temperature difference between the free-stream and the horizontal surface.A similar scaling for rw was obtained in the experiments of Ingersoll [5] and by Narasimha et al [11] in the atmospheric boundary layer under low wind speed conditions. We also derive a formula for boundary layer thickness 5(x) which predicts the boundary layer growth for the combination free-stream velocity Uoo and AT in the low-speed mixed convection regime.
APA, Harvard, Vancouver, ISO, and other styles
44

Elistratova, Vera. "Conception optimale d’une gamme de moteurs synchrones à démarrage direct à haute performance énergétique." Thesis, Ecole centrale de Lille, 2015. http://www.theses.fr/2015ECLI0022/document.

Full text
Abstract:
Ce travail a pour objectif de développer un outil analytique multi-physiques de dimensionnement d’une gamme de moteurs « hybrides » à démarrage direct, intégrant les avantages des deux technologies : l’auto-démarrage de la technologie asynchrone et les bonnes performances énergétique en régime permanent de la technologie synchrone à aimants permanents en répondant aux nouveaux enjeux d’efficacité énergétique et en ajoutant à cela les aspects économiques.La validation de cet outil est effectuée par des modèles éléments finis créés avec un logiciel commercial ANSYS/Maxwell et par des essais expérimentaux réalisés à l’aide de deux prototypes LSPMSM 7.5kW<br>This work aims to develop a multi-physical generic model (and a pre-design software) for a range of LSPMSMs which would integrate the advantages of both technologies: self-start asynchronous technology and good energy performance of synchronous permanent magnet technology. The validation of this model is carried out by finite element commercial software ANSYS / Maxwell and by experimental tests using two 7.5kW.LSPMSM prototypes
APA, Harvard, Vancouver, ISO, and other styles
45

Kau, Wei-Hao, and 高偉豪. "Time series prediction using LSTM Network Models." Thesis, 2018. http://ndltd.ncl.edu.tw/handle/ey3638.

Full text
Abstract:
碩士<br>國立中山大學<br>應用數學系研究所<br>106<br>As recent computing hardware technology has undergone rapid and significant advances, complex methods that require a lot of computing power have been realized, which has led to the development of more machine learning methods and neural network models. This paper discusses the Long short-term memory (LSTM) network model of recurrent neural networks. The first part of this paper introduces the basic concept of LSTM and its training method. The second part discusses short- and long-term prediction, and compares their differences with the conventional time series model. The third part compares the prediction performance of the conventional time series model and LSTM by analysing simulated data. In the empirical study, a Long short-term memory network model is fitted for bicycle rental data in Kaohsiung, and predictive analysis is performed.
APA, Harvard, Vancouver, ISO, and other styles
46

Yen-LiangLin and 林彥良. "PM 2.5 Prediction based on LSTM Model." Thesis, 2018. http://ndltd.ncl.edu.tw/handle/8gayas.

Full text
Abstract:
碩士<br>國立成功大學<br>工程科學系<br>106<br>Recently, pollution conditions of particulate matter 2.5 in Taiwan have become more severe day by day. Several other cities in Asia such as Beijing and Delhi are also facing the same pollution problem, which draws attention to government and experts. Due to the human activities in Asia such as industrialization and animal husbandry, air pollution condition has been getting worse, increases the possibility of population suffering from cardiovascular disease. Particular matter pollution has become a problem we cannot ignore in modern society. Currently, official meteorological department applies traditional statistic model to predict meteorology trend. Traditional statistic model such like ARIMA has certain accuracy on time series data. However, nowadays along with the calculate ability of computer and chips progressing, application field of neural network and deep learning has become much more extensive. Recurrent neural network had been developed to deal with time sequence data. Long short term memory model has a longer time range memorize ability than recurrent neural network, meanwhile has been frequently applied on forecasting and analyzation. This thesis utilizes the long short term memory model to predict future particular matter hourly average concentration, in hope that government and the departments concerned could take actions on the pollution phenomenon, improve the air pollution problem.
APA, Harvard, Vancouver, ISO, and other styles
47

TSAI, YI-TING, and 蔡宜廷. "Air Pollution Forecasting using LSTM with Aggregation Model." Thesis, 2018. http://ndltd.ncl.edu.tw/handle/5p88t6.

Full text
Abstract:
碩士<br>國立臺北大學<br>資訊工程學系<br>107<br>In developed countries or developing countries, the effects of air pollutants on the health of the public are consistent. PM2.5 is a suspended particle In the airborne particulate pollutants. There is no impact on the human body for the concentration threshold of suspended particulates. The concentration of suspended particulates is for humans with diseases of the respiratory system. The impact is also different, so no standard or regulation can completely protect the public. Therefore, predicting the value of PM2.5 in the future is an important issue. This paper uses data provided by the Environmental Protection Agency and Central Weather Bureau from 2013 to 2017. Dividing a data set into data sets of three different sources of pollution. The Aggregation Model uses three types time series data sets of different pollution sources to establish three LSTM (Long Short-Term Memory networks) sub-neural networks to obtain prediction characteristics in three different pollution sources. Combining the predicted features produced by the three sub-neural networks, and outputting prediction feature to the fully connected layer, respectively, giving different weights by the hidden layers from back propagation, and finally obtaining the PM2.5 values predicted in the next 1 to 8 hours. After comparing with the existing methods ANN and LSTM, the accuracy of the next hour is 0.15 less than the LSTM and 0.11 in the MAE. The RMSE is reduced by 0.75 and the MAE error is reduced by 0.54 after comparison with the ANN.
APA, Harvard, Vancouver, ISO, and other styles
48

TSAI, CHENG-HAN, and 蔡承翰. "Application of LSTM Model to Water Stage Forecasting." Thesis, 2019. http://ndltd.ncl.edu.tw/handle/mzv86q.

Full text
Abstract:
碩士<br>逢甲大學<br>水利工程與資源保育學系<br>107<br>In recent years, abnormal weather conditions are observed and the heavy rainfall events are increasing, which is different from the previous disaster characteristics. The river stage rises rapidly and various areas have suffered from floods, resulting in losses of life and property. In previous studies, neural network was often used for prediction but ordinary neural networks cannot preserve the previous information during the prediction which limits the long-term prediction ability. Recurrent Neural Network (RNN) is a suitable choice to overcome this limitation. An RNN model has internal self-looped cells, allowing the RNN to remember information that time series conveyed. RNN was facing gradient explosion and gradient vanish while doing deep learning. To overcome the problem of RNN this study used Long Short-Term Memory model as main structure to build a precipitation-water stage forecasting model. According to the way that we used data, we build three categories model to investigate the different of using average rainfall and distributed rainfall as training data and explore the possibility of application upstream hydrological data as training data to forecast downstream water stage. All the models have good forecasting performance, indicating that the proposed forecasting models had potential to be used to other watersheds.
APA, Harvard, Vancouver, ISO, and other styles
49

Tsung-LingHsu and 許琮苓. "Stock Market Trend Prediction Based on LSTM Model." Thesis, 2019. http://ndltd.ncl.edu.tw/handle/9zca24.

Full text
Abstract:
碩士<br>國立成功大學<br>工程科學系<br>107<br>The stock market has always been an important economic indicator for the society. It is believed that the fluctuation of the stock market seems changed according to some cyclic regularity. For a stock investor, how to find the cyclic regularity of a stock market is a most important issue. However, due to many factors affecting the stock market, it is difficult to obtain accurate predictions. At present, many algorithms have been applied for predicting stock market trends. Due to both local/regional and global economic performance will affect the stock market. The degree of influence is also different for different stock market. So it’s still difficult to define an accepted model for different stock market. In this study, a neural network model called long short-term memory (LSTM) is proposed for predicting the price trend of four companies in the Taiwan stock. Five basic transaction factors for each stock and the three most common statistical indicators in Taiwan stock market are considered. The transaction factors for a specific stock are its opening price, closing price, highest price, lowest price, and transaction volume. The common statistical indicators are Relative Strength Index (RSI), Exponential Moving Average (EMA). For considering the public news attention for the specific company, the information from Google trend is also considered as positive or negative influence with different weighting according to the keywords represented in the news. The data set applied for this study is the trade volume of the Taiwan stock market from January 2010 to February 2019. A window with 60 consecutive trading days is considered to predict the open price of the next (the 61th) trading day, then the window is moved forward to next day for predicting the opening price of the 62th day. The results show that the trend of up or down of the prediction accurate is 63% for the proposed model.
APA, Harvard, Vancouver, ISO, and other styles
50

Hsu, Yi-Kuan, and 許以觀. "A Factory-aware Attentional LSTM Model for PM2.5 Prediction." Thesis, 2019. http://ndltd.ncl.edu.tw/handle/fcx28w.

Full text
Abstract:
碩士<br>國立交通大學<br>資訊管理研究所<br>107<br>With air quality issues becoming a global concern, many countries is facing lot of air pollution problems. While monitoring stations have been established to collect air quality information, and scientists have been committed to the study of air quality predictions, but few studies have taken the different monitoring areas and industrial features into account. In this paper, we propose a deep neural network for PM2.5 predictions, named FAA-LSTM, collecting air quality data from three types of monitors and factory data that is highly related to air quality. A spatial transformation component is designed to obtain the local factors by segmenting the monitoring areas into grids and we consider the influence of neighboring factory data over local PM2.5 grids by adopting attention mechanism to find out the importance. Next, the factor of global air quality station is considered. We combine these heterogeneous data and feed it into a long short-term memory neural network to extract the hidden features and forecast PM2.5 concentrations. In this research, we evaluate our model FAA-LSTM with data from EPA and Academia Sinica in Taichung, surpassing the results of multiple methods, including linear regression, support vector regression, multi-layer perceptron and LSTM.
APA, Harvard, Vancouver, ISO, and other styles
We offer discounts on all premium plans for authors whose works are included in thematic literature selections. Contact us to get a unique promo code!