Skip to main content
Karampistis Dimitrios's user avatar
Karampistis Dimitrios's user avatar
Karampistis Dimitrios's user avatar
Karampistis Dimitrios
  • Member for 5 years, 5 months
  • Last seen more than a month ago
  • Greece
awarded
awarded
awarded
accepted
awarded
asked
Loading…
awarded
comment
Neural network is getting partially trained
At the beginning I was like yeah this is right! I need the squared error. But then I figured out that I am actually using mean-squared-error. The error that is using in back-propagation is the derivative of the cost function, right? I am printing the error like (Σ(expected_error - actual_error)²)/n. So the error that I need to back propagate is the derivative of the cost function which is the function that I am using to find the average. This one (expected_error - actual_error)² which derivative is actually the one that I wrote to my question 2.0 * (expected_output - actual_output)right?
asked
Loading…