0

Can a Basic probability assignment for an event A, m(A)=1 or 0. m(A) could be a decision output for any classifier.

rudresh dwivedi
  • 141
  • what is your opinion? – Seyhmus Güngören Jan 03 '18 at 11:05
  • @SeyhmusGüngören I am trying to combine 4 binary decision outputs (such as 1,0,0,1 or any combination 0 and 1 out of total 16) using D-S theory. According to me, BPA lies in between 0 and 1. The doubt is that does it include 0 and 1 also or not. – rudresh dwivedi Jan 03 '18 at 11:19
  • 1
    according to wiki it does :https://en.wikipedia.org/wiki/Dempster%E2%80%93Shafer_theory – Seyhmus Güngören Jan 03 '18 at 11:22
  • @SeyhmusGüngören okay. Do you know how to apply it over any possible output such as 0,1,0,1 i.e. two out of four classifier accepts and two rejects. The rocedural steps to perform : Fused decision = DS-theory(0,1,0,1) – rudresh dwivedi Jan 03 '18 at 11:26
  • 1
    yes, there is no single way to combine the masses but there is one again by the same guys, which you can again find in the link I gave you. So, first you need to define the masses, or beliefs or plausibilities. Note that from one you can obtain the other. I mean if you have beliefs then you can obtain the masses as in the wiki article. After you obtain the masses for each event $A$, you just use the given fusion rule there. See this section "Dempster's rule of combination" and you will repeat the same rule if you have to combine more masses. – Seyhmus Güngören Jan 03 '18 at 11:31
  • @SeyhmusGüngören Thanks – rudresh dwivedi Jan 03 '18 at 11:32
  • You are welcome. Next time please state the problem more clearly else you will probably not get any response.. good luck.. – Seyhmus Güngören Jan 03 '18 at 11:33

0 Answers0