Suppose you have a classifier $C^n$ which continuously outputs a stream of classification labels $K^n_i$ and corresponding timestamps $T^n_i$. Also, we know the prior probability $P(K^n) \forall n$. Now, suppose we have several classifiers obtaining classification labels from different sources of data but all attempting to recognize the same object.

How do I match the information from these streams? I could store a set of $k$ observations from all classifiers ({$K^n_i,T^n_i$} $\forall n$ and $i\in [1,k]$) and assume that observations correspond to each other based on the distance between their timestamps.

E.g. if we have 2 classifiers and $1\le a,b \le k$, {$K^1_a,T^1_a$} , {$K^2_b,T^2_b$} match if $|T^1_a-T^2_b|<\epsilon$.

Is there a more "intelligent" way to do this?

  • $\begingroup$ So let me get this straight: You have streams of objects with label and timestamp but no additional descriptive properties and certainly no object_id. Now you are looking for a way to identify the same object from different streams in order to make a combined prediction for it. Is this correct ? $\endgroup$ – steffen Dec 8 '11 at 15:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.