Section 2) Deciding on what kind of neural network configuration to use
Deciding on the setup of the network depends largely on how much data you have and how generalized you want
your network to be. Getting the right configuration can be challenging and is more art than science. Here are some general rules on configuration
General Rules
- A more generalized network is capable of finding patterns in your data based on "fuzzy" sample sets. A less generalized network is better at finding exact information.
- The more data you have the longer it will take to train your network. How much time are you willing to devote to training?
- The size of your sample sets will determine how generalized your network is.
There is no hard set rule for configuration, it is a trial and error process. Psychonet will assist you in this by providing a Wizard, allowing multiple networks, giving a customizable fitness function and
a training history for each network that you have trained.
more on neural network configuration
|
