2
$\begingroup$

Most classification methods typically train and predict over all instances, possibly giving a less efficient answer. To give an example, imagine two randomly populated regions on a 2D grid (both populated with the same two binary classes). One well defined space (let's say top half) has 80% of positive classes, and the other only 50%. Further, assume these probabilities are stable.

I envision some type of learner like a classification tree could identify the regions using entropy based splitting. But rather than merging more noisy sets, I would want to literally prune or recode the more noisy regions to a 3rd class as a method to pre-process for the prediction step. So that rather than getting a low average prediction across all of the exemplars, my learner could focus more on high regions of probability. Does anyone know of such an existing method?

Improve this question
$\endgroup$

1 Answer 1

Reset to default
1
$\begingroup$

This is exactly the focus of active learning. Do watch this video by Sanjoy Dasgupta, John Langford: http://videolectures.net/icml09_dasgupta_langford_actl/ to understand the framework. In a way, it combines unsupervised learning(understanding the locality structure) with semi-supervised learning and label propagation.

Improve this answer
$\endgroup$
2
  • $\begingroup$ Thanks for sharing, but I don't really see that as solving the problem. As I understand it, he is still classifying two to n distinct classes over the original region of interest (by sampling over smaller subsets and inferring discrimination boundaries), but never actually identifying then discarding or recoding previously classified low probability regions. $\endgroup$
    – pat
    Commented Aug 23, 2012 at 3:52
  • $\begingroup$ @pat I haven't watched the video. But as Praneeth says this is the point of active learning. You may want to google ' Uncertainty sampling'. I think that is what you want to achieve. P.S: and I just realised the question is three years old, not of much help now I suppose :) $\endgroup$
    – Zhubarb
    Commented May 21, 2015 at 7:14

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.