-1

I am using neural network to train the data. The problem definition is something like: the neural network training provides me the prediction output and error in prediction which the difference between prediction and target. Since I have the error samples, I can derive a probability density function.

Now when I use the network for prediction using different sets of inputs, I want to know the error in this prediction. Is there a method to do this based on the pdf I derived earlier? Something to indicate the probability of error in my prediction.

Thanks. Kashyap

user6848479
  • 1

2 Answers2

0

neural network only gives the predicted output which is supposed to be the task of NN. To get error for each different set of inputs, you have to get it as you quoted.

abhi
  • 87
  • 7
0

From what I understand in regards to your question - is that you are curious in finding the error of your Neural network. You can't find the error of each prediction unless you know the output. If you want to know the average error of your neural network (in regards to your trainingset) iterate through the training set and average the absolute value of output minus the prediction.

Jjoseph
  • 206
  • 2
  • 9