2
$\begingroup$

Suppose that a classifier A classifies a data-point $x$ in class LA1 with probability PA1 and class LA2 with probability PA2 (with LA1 != LA2 and PA1>PA2); and that a classifier B classifies a data-point $x$ in class LB1 with probability PB1 and class LB2 with probability PB2 (with LB1 != LB2 and PB1>PB2).

Note that LA1 may be equal to LB1 or to LB2, and that LB1 may be equal to LA1 or to LA2.

Question: given these results, how can one predict the class of $x$ ?

Example: - A classifies $x$ in class 1 (i.e. LA1) with probability 0.8, and in class 2 (i.e. LA2) with probability 0.1. - B classifies $x$ in class 1 (i.e. LB1) with probability 0.6 and in class 3 (i.e. LB2) with probability 0.2.

Improve this question
$\endgroup$
8
  • $\begingroup$ I dont understand why LA1 and LA2 and LB1 and LB2 MAY be respectively equal. Why don't you know? $\endgroup$
    – jerad
    Commented Nov 29, 2012 at 17:48
  • $\begingroup$ @jerad LA and LB are just notations, that means Label given by classifier A, or Label given by classifier B. I'm just saying that for a given point x, it will be classified in a given class by classifier A, and in another or same class by classifier B. Example: A classify x in class 1 (i.e. LA1) with probability 0.8, and in class 2 (i.e. LA2) with probability 0.1. B classify x in class 1 (i.e. LB1) with probability 0.6 and in class 3 (i.e. LB2) with probability 0.2. Here LA1 = LB1 = class 1. $\endgroup$
    – shn
    Commented Nov 29, 2012 at 18:09
  • 1
    $\begingroup$ The question is ill stated. Without any measure of the accuracy/reliability of the classifiers (such as true positive fraction), we can't be sure that even if A classified x as LA1 100% of the time, x is ever an element of LA1. You should also specify whether these classes are mutually exclusive (like died/living) $\endgroup$
    – AdamO
    Commented Jan 28, 2013 at 20:37
  • $\begingroup$ @AdamO what do you mean by "whether these classes are mutually exclusive (like died/living)" ? If you make a detailed answer about that would be awesome. $\endgroup$
    – shn
    Commented Jan 29, 2013 at 10:20
  • $\begingroup$ Is it possible for x to simultaneously fall in both LA1 and LA2? $\endgroup$
    – AdamO
    Commented Jan 29, 2013 at 15:50

1 Answer 1

Reset to default
1
$\begingroup$

If you dont know whether or not LA1 = LB1 and LA2 = LB2 then you have no way of knowing if your classifiers are commensurate. If however you do know that the two classes are the same for both classifiers, then there's a broad class of methods known as Ensemble Learning available for combining the their outputs to improve classification performance.

Improve this answer
$\endgroup$
2
  • $\begingroup$ We know if LA1 = LB1 or not by just comparing LA1 with LB1 ! I've never said that we dont know whether or not two labels are equals ! $\endgroup$
    – shn
    Commented Nov 29, 2012 at 18:30
  • $\begingroup$ Okay, I guess misinterpreted this comment: "Note that LA1 may be equal to LB1 or to LB2, and that LB1 may be equal to LA1 or to LA2." $\endgroup$
    – jerad
    Commented Nov 29, 2012 at 18:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.