Reply
Mon 20 Feb, 2012 06:01 am
The MLP network in the picture (http://www.alpha.cc.tut.fi/mallinnus/kurssit/soft04/soft04-demo.pdf, ex. 2) is taught to classify 2-dimensional input vectors to two separate category. The text says: "Draw corresponding class boundaries to (x1,x2)-plane assuming that the activation function of neurons is sign."
Some questions:
Does activation function mean those triangles in the picture?
What are class boundaries?
How would you define activation function sign?