|
|
|
Instructions:
|
|
Train
the BAM network to remeber the selected pairs of patterns.Here training
means adjusting the connection weight so that the network remembers
the chosen pairs. In addition to predefined patterns, you may define
also your own patterns.
The network has 100 neurons in the v layer and 64 neurons in the x layer since the pattern pairs defined on 10x10 and 8x8 grids respectively are to be stored. After you have trained the network, load one of the pattern either for v or x layer. You may distort the loaded pattern by clicking the elements of the pattern on the grid. Then let the network converge to recall the stored pattern pair. Observe what happens as the amount of noise is increased, and also as the number of pattern pairs are increased. |