Izzo, Stefano (2025) Design and Optimization of Convolutional Neural Networks for Complex Scenarios. [Tesi di dottorato]
![[thumbnail of PhD_Thesis_Izzo_Stefano.pdf]](https://www.fedoa.unina.it/style/images/fileicons/other.png "PhD_Thesis_Izzo_Stefano.pdf") |
Documento PDF
PhD_Thesis_Izzo_Stefano.pdf Visibile a [TBR] Amministratori dell'archivio Download (40MB) | Richiedi una copia |
| Tipologia del documento: | Tesi di dottorato |
|---|---|
| Lingua: | English |
| Titolo: | Design and Optimization of Convolutional Neural Networks for Complex Scenarios |
| Autori: | Autore Email Izzo, Stefano stefano.izzo@unina.it |
| Data: | 10 Febbraio 2025 |
| Numero di pagine: | 336 |
| Istituzione: | Università degli Studi di Napoli Federico II |
| Dipartimento: | Matematica e Applicazioni "Renato Caccioppoli" |
| Dottorato: | Matematica e Applicazioni |
| Ciclo di dottorato: | 37 |
| Coordinatore del Corso di dottorato: | nome email Nitsch, Carlo c.nitsch@unina.it |
| Tutor: | nome email Piccialli, Francesco [non definito] |
| Data: | 10 Febbraio 2025 |
| Numero di pagine: | 336 |
| Parole chiave: | Deep Learning, Convolutional Neural Networks, CNN, Machine Learning, Applied Mathematics |
| Settori scientifico-disciplinari del MIUR: | Area 01 - Scienze matematiche e informatiche > INF/01 - Informatica |
| Informazioni aggiuntive: | Sono 37° ciclo di dottorato |
| Depositato il: | 21 Ott 2025 20:28 |
| Ultima modifica: | 12 Ago 2026 05:38 |
| URI: | https://www.fedoa.unina.it/id/eprint/16707 |
Abstract
Nowadays, the rapid advancements in machine learning and neural network architectures have revolutionized data analysis, addressing complex challenges across diverse fields. This thesis delves into Convolutional Neural Networks (CNNs), highlighting their transformative potential through four targeted applications. The first study explores CNN-based classification of urban functional zones, enhancing spatial pattern analysis in high-resolution satellite imagery. The second focuses on precision agriculture, using image segmentation to optimize peach processing workflows and minimize operational costs. The third addresses the critical issue of speckle noise removal in medical and radar imaging, proposing a CNN architecture for enhanced image clarity while preserving structural details. Finally, the thesis introduces an innovative federated learning framework leveraging the Forward-Forward algorithm to address data privacy and heterogeneity in distributed environments. By integrating theoretical advancements and practical solutions, this work underscores CNNs' capability to solve emerging challenges and drive innovation in critical domains.
Downloads
Downloads per month over past year
Actions (login required)
 |
Modifica documento |
