Definition of the classification probability threshold in 2D classification?


How exactly is the classification probability threshold defined as used in the “Select 2D classes” job?

I expected that it would be a threshold for the cumulativeprobability that a given particle is present in any of the selected classes, but it seems that it is actually a threshold for the maximum probability that a particle reaches in a single class (of the selected classes). I infer this from the fact that if all classes are selected, particles can still be rejected based on the classification probability threshold.

It seems like a more inclusive threshold might be a useful option - for example, I quite often observe redundant, but otherwise good, classes in 2D classification. A given particle might be of excellent quality, but still have a sub-50% inclusion probability for either of a pair of very similar classes. In this case, if I have both classes together, what I want to know is the probability that the particle belongs to any of the selected classes, not the max probability for an individual class. Does that make sense, or am I thinking about this the wrong way?