I am using an embedded 3-layer network with a rbf hidden layer and linear input and output layers. When providing sample input to the network without prior training, the network outputs show a random behavior. I searched the source code but could not find the reason for this.
Can someone help me?
Thanks in advance for your help.
I'm out of the office and far away from joone at present, but remember that a network will initialise its weights to random numbers when it is used the first time. Since you are dabbing in the source, you can check on this and see if it is not perhaps the problem. I think if you force a weight randomisation, then it will not do so automatically and the issue may be resolved.
Sign up for the SourceForge newsletter:
You seem to have CSS turned off.
Please don't fill out this field.