In this seminar we present an alternative method for how entropy can be used to regulate the growth of a particular type of evolving connectionist system, namely the Evolving Fuzzy Neural Network (EFuNN). Previous work conducted showed promising results but there were areas identified where improvements could be made in terms of the eventual size of the resulting EFuNN as compared with its accuracy and increasing the role that entropy plays in governing how the EFuNN learns.
We present some preliminary results from experiments using two benchmark classification data sets. The results of this work attempt to provide further insights into approaches for incremental learning systems.
Last modified: Tuesday, 28-Jul-2015 16:18:57 NZST
This page is maintained by the seminar list administrator.