Eventos

mércores, maio 11, 2022

mércores, 11 maio, 2022 -
11:00 - 13:00

The extreme learning machine (ELM) is a popular neural network that has two major drawbacks for large-scale data sets: the need of tuning of the number of hidden neurons, and the pseudo-inversion of the hidden activation matrix. This thesis proposes algorithms that keep the simplicity and speed of the ELM network and: 1) avoid tuning and bound the size of hidden activation matrix; 2) raises the ELM performance with a suitable choice of the random bias values; 3) speeds up the hyper-parameter tuning by reducing the number of training executions required.

 

 

mércores, 11 maio, 2022 -
17:00 - 19:00
mércores, 25 maio, 2022 -
17:00 - 19:00
mércores, 1 xuño, 2022 -
17:00 - 19:00
mércores, 8 xuño, 2022 -
17:00 - 19:00
mércores, 22 xuño, 2022 -
17:00 - 19:00

Objetivo