Closed Tobimaru closed 1 year ago
Yes, that is a good interpretation. Technically it should be after relative confidence threshold like you say, but in practice we use relative confidence threshold = 0, so it doesn't matter.
Well it does matter I think if the behavior can be changed by the client 😅? I don't think the code will do what you expect with a relative confidence threshold that is non-zero, given that the confidence is already flattened above the conf_thresh
. In any case, thank you for the feedback.
You're right :) The code we provide here is not without errors. I'll fix this issue in an upcoming release.
Hello. Not completely sure if this is an issue, but I am a bit confused about what the
conf_thresh
parameter should be used for. I'm referring to this line in dkm.py. Shouldn't this line come after the filtering withrelative_confidence_threshold
? Is it correct to say thatconf_thresh
filters out outliers and the inliers are given the same probability?