October 2018
Intermediate to advanced
246 pages
6h 26m
English
The weights of all the layers are initiated with random values while maintaining a standard deviation value of 0.01. The network is trained with the previously mentioned training dataset from scratch. The resultant values for the training, represented as binary vectors, correspond to truth classes. This result carries the proper label for the age group classification, as well as the gender truth class associated with it.
Read now
Unlock full access