News for hackers. Stuff that matters.
Re:Top Algorithms in Data Mining (2008) [pdf] by mikkom on Monday March 24, 2014 @02:20AM (#7456999)
Now you are talking about supervised vs unsupervised learning and it seems that you are talking about clustering, not classification. Classification problems are problems where you have number of classes and the classificator tries to estimate ẃith given parameters what group(s) the given sample belongs to. It doesn't 'invent' new classes.
Re:Top Algorithms in Data Mining (2008) [pdf] by graycat on Monday March 24, 2014 @10:40AM (#7458439)
Okay, then, a critique of the OP is that in classification the number of classes is an input to the algorithm, not an output. My view is that being clear about such details is crucial.