Pocket Algorithm

[mddsangster]

I'm working on coding the pocket algorithm, however I feel I have hit an impasse. I'm not sure if it is a bug in my code or a flaw in the inherent structure of the pocket algorithm, but when testing the code on a known linearly separable data set, the algorithm often 'settles' on a non-optimal weight vector (i.e. one where Ein != 0). I believe this is due to the fact that the algorithm does not allow for weights producing higher values of Ein even if those values may later lead to an even lower Ein.

So, what I'm asking is, should the pocket algorithm always be able to find a linear classifier that separates separable data?

[yaser]

Originally Posted by mddsangster

I believe this is due to the fact that the algorithm does not allow for weights producing higher values of Ein even if those values may later lead to an even lower Ein.

Pocket in fact does not interfere at all with what is allowed. All it does is record the best PLA solution so far, but otherwise let PLA run as it would otherwise.

[elyoum]

Originally Posted by yaser

Pocket in fact does not interfere at all with what is allowed. All it does is record the best PLA solution so far, but otherwise let PLA run as it would otherwise.

really you help me yaser, good job

[habertezi.com]

Türkiye'nin en büyük haber sitesi Son dakika haberler