Loading...
213 results
Search Results
Now showing 1 - 10 of 213
- Deep-learning in identification of vocal pathologiesPublication . Teixeira, Felipe; Teixeira, João PauloThe work consists in a classification problem of four classes of vocal pathologies using one Deep Neural Network. Three groups of features extracted from speech of subjects with Dysphonia, Vocal Fold Paralysis, Laryngitis Chronica and controls were experimented. The best group of features are related with the source: relative jitter, relative shimmer, and HNR. A Deep Neural Network architecture with two levels were experimented. The first level consists in 7 estimators and second level a decision maker. In second level of the Deep Neural Network an accuracy of 39,5% is reached for a diagnosis among the 4 classes under analysis.
- Training neural networks by resilient backpropagation algorithm for tourism forecastingPublication . Fernandes, Paula Odete; Teixeira, João Paulo; Ferreira, João José; Azevedo, Susana GarridoThe main objective of this study is to presents a set of models for tourism destinations competitiveness, using the Artificial Neural Networks (ANN) methodology. The time series of two regions (North and Centre of Portugal) has used to predict the tourism demand. The prediction for two years ahead gives a mean absolute percentage error between 5 and 9 %. Therefore, the ANN model is adequate for modelling and prediction of the reference time series. This model is an important and useful framework for better planning and development of these two regions as they operate in highly competitive markets.
- Acoustic analysis of vocal dysphoniaPublication . Teixeira, João Paulo; Fernandes, Paula OdeteVoice acoustic analysis is becoming more and more usefúl in diagnosis of voice disorders or laryngological pathologies. The facility to record a voice sigiial is an advantage over other invasive techniques. This paper presents the statistical analyzes ofa set of voice parameters like jitter, shimmer and HNR over a 4 groups of subjects vvith dysphonia, fünctional dysphonia, hyperfünctional dysphonia, and psychogenic dysphonia and a control group. No statistical signifícance differences over pathologic groups were found but clear tendencies can be seen between pathologic and control group. The tendencies indicates this parameters as a good features to be used in an intelligent diagnosis system, moreover the jitter and shimmer parameters measured over different tones and vowels.
- Evaluation of a segmental durations model for TTSPublication . Teixeira, João Paulo; Freitas, Diamantino SilvaIn this paper we present a condensed description of a European Portuguese segmental duration’s model for TTS purposes and concentrate on its evaluation. This model is based on artificial neural networks. The evaluation of the model quality was made by comparison with read speech. The standard deviation reached in test set is 19.5 ms and the linear correlation coefficient is 0.84. The model is perceptually evaluated with 4.12 against 4.30 for natural human read speech in a scale of 5.
- Help system for medical diagnosis of the electrocardiogramPublication . Teixeira, João Paulo; Lopes, VandaThis presentation is part of a work that aims to create an interactive learning medical system of ECG events and pathologies diagnose. The system can be seen as an interactive game in wish the user will practice with ECG signal of several pathologies For this purpose an algorithm was developed that detects the events of the ECG in MATLAB. This paper is focus on the discussion of the algorithm, connections between the ECG and the heart physiology listing several pathologies and their respective ECG patterns. A preparatory smoothing of the signal is performed. The algorithm is now only applicable to the normal ECG. It is based in correlation of events within a period of the ECG, and finds the P wave and T wave searching the peaks under a confined region of the smoothed signal.
- Artificial neural networks versus Box Jenkins methodology in tourism demand analysisPublication . Fernandes, Paula Odete; Teixeira, João Paulo; Ferreira, João José; Azevedo, Susana GarridoSeveral empirical studies in the tourism area have been performed and published during the last decades. The researchers are unanimous upon considering that in the planning process, decisionmaking and control of the tourism sector, the forecast of the tourism demand assumes an important role. Nowadays, there is a great variety of methods for forecasting that have been developed and which can be applied in a set of situations presenting different characteristics and methodologies, going from simple approaches to more complex ones. In this context, the present study aims to explore and to evidence the usefulness of the Artificial Neural Networks methodology (ANN), in the analysis of the tourism demand, as an alternative to the Box-Jenkins methodology. ANN has been under attention in the area of business and economics since, in this field, it presents this methodology as a valid alternative to classical methods of forecasting allowing its application for problems in which the traditional ones would be difficult to use (Thawornwong & Enke, 2004). As referred by Hill et al. (1996) and Hansen et al. (1999), ANN shows 1 ability for improving time-series forecasts by mining additional information, diminishing their dimensionality, and reducing their complexity. In this way, for each methodology treatment, analysis and modeling of the tourism time-series: “Nights Spent in Hotel Accommodation per Month” registered between January 1987 and December 2006, was carried out since is one of the variables that better explains the effective tourism demand. The study was performed for the North and Center regions of Portugal. Considering the results, and according to the Criteria of MAPE for model evaluation in Lewis (1982), the ANN model presented an acceptable goodness of fit and good statistical properties and is, therefore, adequate for modelling and prediction of the reference time series, when compared to the results obtained by the methodology of Box-Jenkins.
- Forecasting omicron variant of Covid-19 with ANN model in european countries – number of cases, deaths, and ICU patientsPublication . Carvalho, Kathleen; Reis, Luis Paulo; Teixeira, João PauloAccurate predictions of time series are increasingly required to support judgments in a variety of decisions. Several predictive models are available to support these predictions, depending on how each field offers a data variety with varied behavior. The use of artificial neural networks (ANN) at the beginning of the COVID-19 pandemic was significant since the tool may offer forecasting data for various conditions and hence assist in governing critical choices. In this context, this paper describes a system for predicting the daily number of cases, fatalities, and Intensive Care Unit (ICU) patients for the next 28 days in five European countries: Portugal, the United Kingdom, France, Italy, and Germany. The database selection is based on comparable mitigation processes to analyze the impact of safety procedure flexibilization with the most recent numbers of COVID-19. Additionally, it is intended to check the algorithm’s adaptability to different variants throughout time. The network’s input data has been normalized to account for the size of the countries in the study and smoothed by seven days. The mean absolute error (MAE) was employed as a comparing criterion of two datasets, one with data from the beginning of the pandemic and another with data from the last year, since all variables (cases, deaths, and ICU patients) may be tendentious in percentage analysis. The best architecture produced a general MAE prediction for the 28 days ahead of 256,53 daily cases, 0,59 daily deaths, and 1,63 ICU patients, all numbers normalized by million people.
- Electroencephalogram cepstral distances in alzheimer’s disease diagnosisPublication . Rodrigues, Pedro Miguel; Freitas, Diamantino Silva; Teixeira, João PauloAlzheimer's disease (AD) represents one ofthe greatest public health challenges worldwide nowadays, because it affects millions of people ali o ver the world and it is expected that the disease will increase considerably in the near future. This study is the first application attempt of cepstral analysis on Electroencephalogram (EEG) signals to find new parameters in arder to achieve a better differentiation belween EEGs of AD patients and Control subjects. The results show that the methodology that uses a combined Wavelet (WT) Biorthogonal (Bior) 3.5 and cepstrum analysis was able to describe the EEG dynamics with a higher discriminative power than the other WTs/spectmm methodologies m previous studies. The most important significance figures were found in cepstral distances between cepstrums oftheta and alpha bands (p=0. 00006<0. 05).
- Modelação e previsão da procura turística em Moçambique: um estudo comparativo entre os modelos de redes neuronais artificiais e regressão linear múltiplaPublication . Constantino, Hortêncio; Fernandes, Paula Odete; Teixeira, João PauloO principal objetivo do presente trabalho assenta num estudo comparativo entre o modelo de Regressão Linear Múltipla e de Redes Neuronais Artificiais, para prever a procura turística em Moçambique. Utilizou-se para tal o número de dormidas mensais registadas nos estabelecimentos hoteleiros, para o período de Janeiro de 2004 a Dezembro de 2013. Para tal, foram selecionadas as variáveis explicativas: índice Harmonizado de Preços ao Consumidor, Produto Interno Bruto e Taxa de Câmbio para os principais mercados emissores: África Sul, Estados Unidos da América, Moçambique, Portugal e Reino Unido.
- Tourism time series forecast with artificial neural networksPublication . Teixeira, João Paulo; Fernandes, Paula OdeteThe modulation of tourism time series was used in this work for forecast purposes. The Tourism Revenue and Total Overnights registered in the hotels of the North region of Por- tugal were used for the experimented models. Several feed-forward Artificial Neural Networks (ANN) models using different input features and number of hidden nodes were experimented to forecast the Tourism time series. Empirical results indicate that the Dedicated ANN models perform better than models with several outputs. Generally the usage of previous 12 values of the same time series is very important to a good quality forecast. For the prediction of Tourism Revenue the Foreign Overnights and GDP of contributing countries are relevant. This time series was predicted with an error of 4.7% and a Pearson correlation of 0.98. The forecast of Total Overnights had an error of 6.0% and Pearson correlation of 0.98. Domestic Overnights are more predictable than Foreign Overnights.