what does Validation Patience mean in training Option?

68 visualizaciones (últimos 30 días)
NGR MNFD
NGR MNFD el 8 de Ag. de 2021
Respondida: Esen Ozbay el 10 de Ag. de 2021
hi all
what dose ValidationPatience mean in trainingOption? and how I should use the number of it? i do not understand help of mathwork in this field. thanks

Respuestas (1)

Esen Ozbay
Esen Ozbay el 10 de Ag. de 2021
Validation patience is the number of epochs that the algorithm tries to improve the performance before giving up (if the error is not decreasing).
For example:
Epoch: ... 191 192 193 194 195 196 197 198 199 200 201 202 203 204
Performance ... 10 9 2 1.2 1.2 1.2 1.2 0.8 0.7 0.7 0.7 0.7 0.7 0.7
In this case, if validation patience were 2, the algorithm would stop at epoch 196. If it were 5, it would stop at epoch 204.
Normally, it is assumed that if the performance does not decrease for a few epochs in a row, then a minimum has been found. However, in rare cases where the minimum can be surpassed with some 'patience', a high number for validation patience can help the network attain lower performances.
There is a trade off between waiting a long time for nothing and missing out on better performances.
I personally leave ValidationPatience at 6.

Productos


Versión

R2019a

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by