Machine learning prediction of Gold Production in supernova / Ilhem Bellal

Public ISBD Titre : Machine learning prediction of Gold Production in supernova Type de document : document électronique Auteurs : Ilhem Bellal, Auteur ; Ouassila Boukhenfouf, Directeur de thèse Editeur : Setif:UFA Année de publication : 2025 Importance : 1 vol (61 f.) Format : 29 cm Langues : Anglais (eng) Catégories : Thèses & Mémoires:Physique Mots-clés : Machine learning Index. décimale : 530 - Physique Résumé : This topic dives into the cutting-edge application of artificial intelligence (AI), specifically Convolutional Neural Networks (CNNs), to analyze astrophysical spectra and detect heavy elements like gold in supernova remnants—super exciting stuff! It involves developing and training a CNN model using a synthetic dataset sourced from Kaggle, derived from the study "Artificial Intelligence Assisted Inversion of Synthetic Type Ia Supernova Spectra," and a newly constructed database compiled from open-source astronomical observations. This thesis kicks off with a brief intro to AI, machine learning (ML), and deep learning (DL), and highlights how ML can tackle supernova spectral analysis. It also walks through using Python, with libraries like TensorFlow/Keras, and Google Colab’s cloud-based, hardware-accelerated environment to build and train the CNN model. The trained CNN model predicts the presence of r-process–dominated elements, with training based on the Kaggle dataset and the newly created database. The results are promising, achieving an accuracy of 83.33% and a recall of 100% for r-process events, though performance varies due to data quality and hardware limitations. This work aims to enhance the CNN’s ability to identify heavy elements in supernova spectra, offering insights into nucleosynthesis while addressing challenges like data preprocessing and model optimization—setting the stage for future astrophysical discoveries. Note de contenu : Côte titre : MAPH/0714 Machine learning prediction of Gold Production in supernova [document électronique] / Ilhem Bellal, Auteur ; Ouassila Boukhenfouf, Directeur de thèse . - [S.l.] : Setif:UFA, 2025 . - 1 vol (61 f.) ; 29 cm. Langues : Anglais (eng) Catégories : Thèses & Mémoires:Physique Mots-clés : Machine learning Index. décimale : 530 - Physique Résumé : This topic dives into the cutting-edge application of artificial intelligence (AI), specifically Convolutional Neural Networks (CNNs), to analyze astrophysical spectra and detect heavy elements like gold in supernova remnants—super exciting stuff! It involves developing and training a CNN model using a synthetic dataset sourced from Kaggle, derived from the study "Artificial Intelligence Assisted Inversion of Synthetic Type Ia Supernova Spectra," and a newly constructed database compiled from open-source astronomical observations. This thesis kicks off with a brief intro to AI, machine learning (ML), and deep learning (DL), and highlights how ML can tackle supernova spectral analysis. It also walks through using Python, with libraries like TensorFlow/Keras, and Google Colab’s cloud-based, hardware-accelerated environment to build and train the CNN model. The trained CNN model predicts the presence of r-process–dominated elements, with training based on the Kaggle dataset and the newly created database. The results are promising, achieving an accuracy of 83.33% and a recall of 100% for r-process events, though performance varies due to data quality and hardware limitations. This work aims to enhance the CNN’s ability to identify heavy elements in supernova spectra, offering insights into nucleosynthesis while addressing challenges like data preprocessing and model optimization—setting the stage for future astrophysical discoveries. Note de contenu : Côte titre : MAPH/0714

Exemplaires (1)

Code-barres	Cote	Support	Localisation	Section	Disponibilité
MAPH/0714	MAPH/0714	Mémoire	Bibliothèque des sciences	Anglais	Disponible Disponible

---

1 (1 - 1 / 1)