You are here

BackPropagation

I have a question...I am using Encog framework to train a network using BP. I am training the network using images that has some object say a cat, telling the network which images are cats and which are not.

Bellow are my Neural Network results
actual=0.1545868370551181 ideal=0.0,
actual=0.797896306829758 ideal=1.0,
actual=0.1545868370551181 ideal=0.0,
actual=0.797896306829758 ideal=1.0

It is my understanding that since different pictures are presented to network some look like cat for example then they should have higher percentage...but i keep getting same percentage for any pic that has a cat which is 0.79 and for other pics i get 0.15...how can i solve this issue!, what is the intuition behind whats going on!

Many thanks

Sorry for repeating it in another comment!

Neural Network Forums: 

Theme by Danetsoft and Danang Probo Sayekti inspired by Maksimer