Influence of input parameters in Neural networks

G'day

Im trying to find out the influence of different inputs in neural networks

eg. if i have 6 inputs and 1 target output which of those 6 inputs has the most influence/weighting on the output.

is there an easy way to find this out??

Cheers

[Greg Heath]

Terminology: Input variables, not parameters. Parameters are constant during each use.

Answer 1

[Greg Heath]

 1. Minimize the number of hidden nodes that will yield the desired goal.

 2. Compare the performance of the 6 nets that result when the input from only one of the original variables is randomly scrambled. For example, 

 input(i,:) = input0( i , randperm(1:N));

 3. Repeat with different RNG states until the choice is clear.

Hope this helps.

Thank you for formally accepting my answer

Greg

[Greg Heath]

Compare with the results from the linear model obtained from stepwisefit in the backward mode.