Neural Net MSE trough

Asked Nov 01 '16 at 14:46

Active Nov 02 '16 at 07:48

Viewed 58 times

I am trying to train a BackProp Neural Net with 21 inputs and 1 output.

The input data is S&P 500 stock data (indicators) and the output is if a buy should occur.

I have tried a number of neural net configurations but as the epochs progress the MSE gets larger after a decent as seen in the picture below.

enter image description here

Does this behavior denote something about my data or that something is wrong?

edited Nov 02 '16 at 07:48

DanielBarbarian

asked Nov 01 '16 at 14:46

Mike Smith

1

Is that the MSE on the training data or the test data? – Aaron Nov 01 '16 at 22:17
It is on the training data. – Mike Smith Nov 02 '16 at 00:49
1

Can you give more information about the neural net work you are training and could you shows an exact plot of the MSE? – João Almeida Nov 02 '16 at 07:53
I have the same issue: https://stackoverflow.com/questions/43855720/c-c-fann-error-barely-drops-and-then-goes-back-up-again – andersfylling May 08 '17 at 19:28

0 Answers0