Titre :	A Deep Learning Model for Predication
Type de document :	texte imprimé
Auteurs :	Ammar Assif Menaouel, Auteur ; Seif Eddine Ayad, Auteur ; Daifi,ahlem, Directeur de thèse
Année de publication :	2022
Importance :	1 vol (85 f .)
Format :	29cm
Langues :	Français (fre)
Catégories :	Thèses & Mémoires:Informatique
Mots-clés :	Machine Learning Deep Learning
Index. décimale :	004 Informatique
Résumé :	Time series forecasting involves developing a predictive model on data where there is an ordered relationship between observations. In fact, there are many challenges when forecasting one or more possible future observations because forecasting models add the complexity of order or temporal dependence between observations. Traditionally, time series forecasting has been dominated by linear methods like ARIMA because they are well understood and effective on many problems. However, this linear relationship excludes more complex joint distributions and many real-world problems have multiple input variables. In this thesis, we focus on developing a deep learning model as it has already been proved that they are effective on more complex time series forecasting problems with multiple input variables. Our proposed model is based on LSTM autoencoder (Long Short Term Memory), which extract complex nonlinear relationships and perform well for mutivariates inputs. In addition, the ability of the autoencoder to project the data in latent space help to deal with the limitation of missing data. We conducted our experiments on a data set of people with type 1 diabetes in order to predict their blood glucose level for a period of 5 minutes to an hour. We have obtained a good result, which we will work on improving in the upcoming works, InchaaAllah
Côte titre :	MAI/0582
En ligne :	https://drive.google.com/file/d/1dgX0lB3xwdmu5pvcqrhPGFAWV7YK3fZF/view?usp=share [...]
Format de la ressource électronique :	pdf

A Deep Learning Model for Predication [texte imprimé] / Ammar Assif Menaouel, Auteur ; Seif Eddine Ayad, Auteur ; Daifi,ahlem, Directeur de thèse . - 2022 . - 1 vol (85 f .) ; 29cm.
Langues : Français (fre)

Catégories :	Thèses & Mémoires:Informatique
Mots-clés :	Machine Learning Deep Learning
Index. décimale :	004 Informatique
Résumé :	Time series forecasting involves developing a predictive model on data where there is an ordered relationship between observations. In fact, there are many challenges when forecasting one or more possible future observations because forecasting models add the complexity of order or temporal dependence between observations. Traditionally, time series forecasting has been dominated by linear methods like ARIMA because they are well understood and effective on many problems. However, this linear relationship excludes more complex joint distributions and many real-world problems have multiple input variables. In this thesis, we focus on developing a deep learning model as it has already been proved that they are effective on more complex time series forecasting problems with multiple input variables. Our proposed model is based on LSTM autoencoder (Long Short Term Memory), which extract complex nonlinear relationships and perform well for mutivariates inputs. In addition, the ability of the autoencoder to project the data in latent space help to deal with the limitation of missing data. We conducted our experiments on a data set of people with type 1 diabetes in order to predict their blood glucose level for a period of 5 minutes to an hour. We have obtained a good result, which we will work on improving in the upcoming works, InchaaAllah
Côte titre :	MAI/0582
En ligne :	https://drive.google.com/file/d/1dgX0lB3xwdmu5pvcqrhPGFAWV7YK3fZF/view?usp=share [...]
Format de la ressource électronique :	pdf