Databases
StackOverflow2013
Postsdbo
PONeural network training. Only few outcomes
Body
 

Note that there are some explanatory texts on larger screens.

plurals
  1. PONeural network training. Only few outcomes
    text
    copied!<p>I have a network which has 3 inputs, 2 hidden layers (6 neurons each, Sigmoid activation function) and one neuron as output. I expect my network to be continuous, as I'm not looking at a classification network (hope that makes sense).</p> <p>My inputs represent days in a year (0-365 range). I actually normalize them to 0-1 range (because of sigmoid).</p> <p>My problem is the following: however small the training error gets, the actual values when reusing the training set are not correct. Depending on the number of epochs I run I get different outcomes. </p> <p>If I train my network more than a few thousand times, I only get two possible outcomes. If I train it less, I get more possible outcomes, but the values are nowhere near what I expect.</p> <p>I've read that for a continuous network, it's better too use two hidden layers. </p> <p>I'm not sure what I'm doing wrong. If you can be of any help, that would be great. Let me know if you need more details.</p> <p>Thanks</p> <p><strong>UPDATE 1</strong></p> <p>I reduced the number of elements in the training set. This time the network converged in a small number of epochs. Below are the training errors:</p> <hr> <p>Training network</p> <hr> <p>Iteration #1. Error: 0.0011177179783950614</p> <p>Iteration #2. Error: 0.14650660686728395</p> <p>Iteration #3. Error: 0.0011177179783950614</p> <p>Iteration #4. Error: 0.023927628368006597</p> <p>Iteration #5. Error: 0.0011177179783950614</p> <p>Iteration #6. Error: 0.0034446569367911364</p> <p>Iteration #7. Error: 0.0011177179783950614</p> <p>Iteration #8. Error: 8.800816244191594E-4</p> <hr> <p>Final Error: 0.0011177179783950614</p> <hr>
 

Querying!

 
Guidance
An individual column

Larger individual text columns get their own page to allow for proper reading.

SQuiL has stopped working due to an internal error.

If you are curious you may find further information in the browser console, which is accessible through the devtools (F12).

Reload