ICA(Independent Component Analysis)
Assumption: patterns are independent
Final map: regards as some patterns composite result
Training Step:
Step 1: Choose about 20 wafer maps from same pattern, applying ICA to each maps(suppose
have 2 or 3 sources)
Step 2: Averaging as the basis for the pattern
Classification Step:
For the unknown patterns map, applying ICA to get the basis
Using KNN to predicted
