The goal of this thesis is the identification and classification of spectral lines using deep learning methods. Specifically, a particular model of deep neural networks, convolutional neural networks, will be employed. The first chapter introduces the theoretical context, providing a description of the principles of machine learning, the characteristics of deep learning, and the functioning of deep neural networks. The second chapter describes the data processing procedure based on the WFC3 IR Spectroscopic Parallel Survey, which leads to the creation of the final dataset, along with a description of the variables involved and the strategy adopted to address the problem. Finally, the third chapter examines the practical implementation of the proposed model, presenting the results and analyzing its performance.
L’obiettivo di questa tesi è l’identificazione e la classificazione di linee spettrali attraverso metodi di deep learning. In particolare verrà utilizzato uno specifico modello di reti neurali profonde, le convolutional neural networks. Nel primo capitolo viene introdotto il contesto teorico, con una descrizione dei principi del machine learning, delle caratteristiche del deep learning e il funzionamento delle reti neurali profonde. Nel secondo capitolo si descrive il processo di elaborazione dei dati proveniente dal WFC3 IR Spectroscopic Parallel Survey, che conduce alla creazione del dataset finale, insieme alla descrizione delle variabili coinvolte e la strategia adottata per affrontare il problema. Infine nel terzo capitolo si analizza l’implementazione pratica del modello proposto, presentando i risultati e analizzandone le prestazioni.
Identificazione di linee spettrali mediante deep learning: un'analisi tramite diagrammi a stella
CIMA, AGNESE
2024/2025
Abstract
The goal of this thesis is the identification and classification of spectral lines using deep learning methods. Specifically, a particular model of deep neural networks, convolutional neural networks, will be employed. The first chapter introduces the theoretical context, providing a description of the principles of machine learning, the characteristics of deep learning, and the functioning of deep neural networks. The second chapter describes the data processing procedure based on the WFC3 IR Spectroscopic Parallel Survey, which leads to the creation of the final dataset, along with a description of the variables involved and the strategy adopted to address the problem. Finally, the third chapter examines the practical implementation of the proposed model, presenting the results and analyzing its performance.File | Dimensione | Formato | |
---|---|---|---|
Cima_Agnese.pdf
accesso riservato
Dimensione
555.54 kB
Formato
Adobe PDF
|
555.54 kB | Adobe PDF |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/84123