Por favor, use este identificador para citar o enlazar este ítem:
http://cimat.repositorioinstitucional.mx/jspui/handle/1008/1027| ANALYSIS OF SOME STATISTICAL PROPERTIES OF NEURAL NETWORKS | |
| JUAN FRANCISCO MANDUJANO REYES | |
| Acceso Abierto | |
| Atribución-NoComercial | |
| PROBABILIDAD Extreme Learning Machine | |
| his thesis provides a theoretical study of the statistical properties of some neural network models by means of random matrix theory and model selection tools. We discuss the concentration inequalities approach, established in Louart, Liao and Couillet in 2018 to study the performance of extreme learning machines. Some limiting spectral results of certain matrices, presented in Pennington and Worah in 2017 and Benigni y Peche in 2019, are studied, as well as two novel applications of a model selection approach to select a hyperparameter in extreme learning machines. Moreover, some original and applicable results about new and useful activation functions are presented, as well as conjectures related to the speed of the training of deep neural networks. | |
| 13-09-2019 | |
| Tesis de maestría | |
| OTRAS | |
| Versión aceptada | |
| acceptedVersion - Versión aceptada | |
| Aparece en las colecciones: | Tesis del CIMAT |
Cargar archivos:
| Fichero | Descripción | Tamaño | Formato | |
|---|---|---|---|---|
| TE 752.pdf | 1.04 MB | Adobe PDF | Visualizar/Abrir |