Abstract
A new evolutionary learning model that has a different architechture from the traditional evolutionary learning model is developed. We call it as the interactive evolutionary learning model. Both evolutionary learning models are similar in several components. In the interactive evolutionary learning algorithm, some key mechanisms, the solution set, the partial fitness and the reward backpropagation, are roposed. We describe how both the traditional evolutionary leaning model and the interactive evolutionary learning model can be applied into playing 9*9 Go games. Finally we analyze the learning curves in both learning processes and we show many interesting Go game concepts that are discovered by the learning system.