Quote:
Originally Posted by LazyPiggy
I have exhausted great efforts to work out the PLA problem and there exists some cases which may never converge on all training data,will they happen or there is something wrong with my code? All comments are appreciated.
|
If you generated the data set as prescribed in the problem statement, the data will be linearly separable and PLA is guaranteed to converge. It may take a long time, but it should eventually converge.