netGlass - Analysis of Noisy Neurocontrollers
Project Description
ANNs trained in a simulator suffer from noise in real world applications, because there is always
noise in the real world. In theory, applaying noise to a ANN while it is evolving in a simulator,
should make it more noise resistant.
Therefore this Project is about finding and implementing an analyzer for noisy neurocontrollers.
We do so with the help of the Java Package NetGlass, which provides helpful procedures to extract
information of ANNs. We will also extend netGlass with additional analysation capabilities
in order for it to suit our purposes.