This program simulates a 3 or 4-layer Neural Network, and can be used to

simulate an arbitrary, complex, or non-linear function that would be

difficult to implement by traditional methods. The Back Propagation

method is used "teach" the network the desired function.

Adjacent layers of the net are fully interconnected that is, every

neuron in layer 1 is connected to every neuron in layer 2, and every

neuron in layer 2 is connected to every neuron in layer 3 (1->2, 2->3).

With a 4-layer net, there is further interconnection: 1->2, 1->3, 2->3,

2->4, 3->4.

-　 This program simulates a 3 or 4-layer Neural Network, and can be used to

　 simulate an arbitrary, complex, or non-linear function that would be

　 difficult to implement by traditional methods. The Back Propagation

　 method is used "teach" the network the desired function.

　 Adjacent layers of the net are fully interconnected　that is, every

　 neuron in layer 1 is connected to every neuron in layer 2, and every

　 neuron in layer 2 is connected to every neuron in layer 3 (1->2, 2->3).

　 With a 4-layer net, there is further interconnection: 1->2, 1->3, 2->3,

　 2->4, 3->4.