Matthew Toseland wrote:
> How valuable are negative samples? If you are directly connected to the 
> originator, you will *never* get a negative sample from it (assuming you can 
> unambiguously identify the requests of interest).

The target location and nearest location of a negative sample could 
reveal a lot. If nearestLoc==prevLoc then dist(originator,target) >= 
dist(prevLoc,target). Each sample gives you a range of possible 
originator locations, and if you get a series of linkable samples you 
can intersect the ranges.

Example: a peer at location 0.3 sends you an interesting request with 
target==0.7 and nearestLoc==prevLoc, so dist(originator,0.7) >= 
dist(0.3,0.7), so the originator's between 0.1 and 0.3. Then the same 
peer sends you another interesting request with target==0.9 and 
nearestLoc==prevLoc, so dist(originator,0.9) >= dist(0.3,0.9), so the 
originator's between 0.3 and 0.5. Your peer must be the originator.

Cheers,
Michael

Reply via email to