Theodore John De Jesus
2007-08-09 10:24:22 UTC
hi im experimenting the xor_train.c, i used spam data instead of xor data.
when i run it, the error is too big, 0.5xxxx. I then changed the activation
function from sigmoid to fann_elliot. It goes down from 0.5xx to 0.1xxx with
100 hidden layes and 1000 epochs. Do you have other reference regarding
fann_elliot because im clueless why the error goes down like that? How
elliot works? Thank you for your time.
when i run it, the error is too big, 0.5xxxx. I then changed the activation
function from sigmoid to fann_elliot. It goes down from 0.5xx to 0.1xxx with
100 hidden layes and 1000 epochs. Do you have other reference regarding
fann_elliot because im clueless why the error goes down like that? How
elliot works? Thank you for your time.