For the last decades, neural networks have been an increasing topic of interest in machine learning due to their capability to approximate any type of function. Over time, various models with different purposes have been proposed, thus reaching numerous fields of application and disciplines of interest. Only recently, the knowledge of neural networks has been combined with the mathematics of ordinary differential equations (ODE). This paper's aim is to present the neural network model introduced in 2019 by Ricky T. Q. Chen et al. called Neural ODE, analyzing its advantages, limitations, and some of its applications. It is also intended to examine the impact and evolution this model has had from its introduction until today.
Le reti neurali sono da alcuni decenni un argomento di grande interesse nel machine learning per la loro capacità di approssimare qualsiasi funzione. Nel tempo sono stati proposti vari modelli con diversi scopi, raggiungendo così numerosi campi d'applicazione e discipline di interesse. Solo recentemente le conoscenze sulle reti neurali si sono unite alla matematica delle equazioni differenziali ordinarie (ODE). L’obiettivo di questo elaborato è quello di presentare il modello di rete neurale introdotto nel 2019 da Ricky T. Q. Chen et al. denominato Neural ODE, analizzandone i vantaggi, le limitazioni e alcune sue applicazioni. Si intende, inoltre, esaminare l’impatto e l' evoluzione che tale modello ha avuto dalla sua introduzione ad oggi.
Equazioni differenziali ordinarie applicate alle reti neurali
CALIFANO, STEFANO
2021/2022
Abstract
For the last decades, neural networks have been an increasing topic of interest in machine learning due to their capability to approximate any type of function. Over time, various models with different purposes have been proposed, thus reaching numerous fields of application and disciplines of interest. Only recently, the knowledge of neural networks has been combined with the mathematics of ordinary differential equations (ODE). This paper's aim is to present the neural network model introduced in 2019 by Ricky T. Q. Chen et al. called Neural ODE, analyzing its advantages, limitations, and some of its applications. It is also intended to examine the impact and evolution this model has had from its introduction until today.File | Dimensione | Formato | |
---|---|---|---|
Califano_Stefano.pdf
accesso aperto
Dimensione
2.34 MB
Formato
Adobe PDF
|
2.34 MB | Adobe PDF | Visualizza/Apri |
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/20.500.12608/38999