Re: Does a "symetric" bayes algoritm excist?
From: Rob J Meijer (rmeijer_at_xs4all.nl)
Date: 09/28/04
- Next message: Duncan Murdoch: "Re: Does a "symetric" bayes algoritm excist?"
- Previous message: Ian Jermyn: "Re: Does a "symetric" bayes algoritm excist?"
- In reply to: Ian Jermyn: "Re: Does a "symetric" bayes algoritm excist?"
- Next in thread: Duncan Murdoch: "Re: Does a "symetric" bayes algoritm excist?"
- Reply: Duncan Murdoch: "Re: Does a "symetric" bayes algoritm excist?"
- Messages sorted by: [ date ] [ thread ]
Date: 28 Sep 2004 09:19:49 GMT
"Ian Jermyn" <Ian.Jermyn@sophia.inria.fr> writes:
>It sounds to me like you are making an error in applying Bayes' theorem, but
>it is hard to tell because your post is vague. Can you be more precise?
>Ian.
Let me try to be more precise.
I have two vectors of what I think are unconditional probabilities:
* Su : A vector with the unconditional probability of the different possible state of my statemachine.
* Eu : A vector with the unconditional probability of the different possible events the incomming event could be.
Next to this I have two matrices:
* M : A matrix that holds the probabilities that given a particilar state, an incomming event will be of a specific type.
* N : A matrix that holds the probabilities that given a particular incomming event, the state machine would be in a
particular state.
What I am trying to do is to calculate the conditionl Sc and Ec.
The problem is that in order to validly calculate Sc I need Ec, and in order to validly calculate Ec I need Sc.
I have tried to solve this by :
a) first using Su as if it were Sc in order to calculate Ec, than using this result to calculate Sc.
b) first using Eu as if it were Ec in order to calculate Sc, than using this result to calculate Ec.
The ocational differences in the results of a and b have shown this aproach to be flawed, thus I am now
hoping that an algoritm may excist that could solve the two matrix bayes equations simultaniously.
>--
>--------------------------------------------------
>Ian Jermyn
>INRIA (Ariana)
>2004 route des Lucioles
>B. P. 93
>06902 Sophia Antipolis Cedex
>France
>T: +33 (0)4 9238 7683
>F: +33 (0)4 9238 7643
>E: Ian.Jermyn@sophia.inria.fr
>W: http://www-sop.inria.fr/ariana/personnel/Ian.Jermyn
>"Rob J Meijer" <rmeijer@xs4all.nl> a écrit dans le message de
>news:41592012$0$78753$e4fe514c@news.xs4all.nl...
>> I have been trying to build a probablistic state machine for some time
>now,
>> and the problem I seem to keep running into is the fact that the bayes
>theorem
>> is not quite usable for symetric partial information problems, only for
>asymetric
>> correction of patrial knowledge.
>>
>> The problem I have is that I have to equally valid sources of partial
>knowledge,
>> these sources are two probability vectors (one for state, an other for
>event),
>> I have two conditional probability matrices.
>>
>> I could randomly pick one, bayes correct the other one, and than with that
>result
>> bayes correct the first one, However the result A,B ocasionaly differs
>greatly
>> from B,A.
>> What I realy need however is a symetric way to simultaniously bayes
>correct both
>> sources of partial knowledge. Does there exist a algoritm for symetric
>bayes
>> correction of two interdependent sources of partial knowledge?
>>
>> T.I.A.
>>
>> Rob J Meije
- Next message: Duncan Murdoch: "Re: Does a "symetric" bayes algoritm excist?"
- Previous message: Ian Jermyn: "Re: Does a "symetric" bayes algoritm excist?"
- In reply to: Ian Jermyn: "Re: Does a "symetric" bayes algoritm excist?"
- Next in thread: Duncan Murdoch: "Re: Does a "symetric" bayes algoritm excist?"
- Reply: Duncan Murdoch: "Re: Does a "symetric" bayes algoritm excist?"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
|
