Quote:
|
Originally Posted by meglamaniac
I've just never heard of any learning techniques in weather forecasting, not that that means there aren't any.
|
Pretty much every single classifier you'll encounter uses learning techniques - thats how you generalise from a finite training set to being able to accurately classify new observations. Its not unique to Bayes methods (or even statistical inference). If my classifier notices that out of 10000 emails, every single one marked as spam contains the word 's3x', and this word isnt in any non-spam emails, then it will (if its decent) learn the rule "mark email as spam <=> email contains the word 's3x'". Theres nothing Bayesian or statistical about this though.
Obviously you wouldnt ever find such a simple "necessary and sufficient" criteria for spam in the real world though, which is why statistical methods are useful. But the idea of learning rules from a training set is fairly universal.