What would this look like if we used some form of K-Means clustering in the set partition, instead of splitting objects recursively?
catherinecang
do people generally use some heuristic for which axis / spatial dimension to partition over? i.e. if all the objects were approximately in the same y-level but had largely different x-coords, it would probably make sense to split on the x-axis -- would algorithms take this into account?
What would this look like if we used some form of K-Means clustering in the set partition, instead of splitting objects recursively?
do people generally use some heuristic for which axis / spatial dimension to partition over? i.e. if all the objects were approximately in the same y-level but had largely different x-coords, it would probably make sense to split on the x-axis -- would algorithms take this into account?