Equifinality versus evolutionary fragility: the role of collective weak interactions
L. Gwenn Volkert, Michael Conrad · 2002
Dual dynamics networks consist of simple pattern recognizing elements connected to each other by strong influences and subject to weak collective interactions. The networks are assigned input and output nodes and trained to perform pattern recognition tasks with an evolutionary programming algorithm. Different network types are investigated, each utilizing a different coupling of strong and weak influences. The results demonstrate that the manner of coupling significantly affects the response to mutation and the consequent learning characteristics.