Comment by rocqua

Is over fitting different from 'too much optimization'? Optimization still needs a value that is optimized. Over fitting is the result of too much optimization for not quite the right value (i.e. training error when you want to reduce prediction error)