Key points are not available for this paper at this time.
We present a simple variant of the k-d tree which automatically adapts to intrinsic low dimensional structure in data without having to explicitly learn this structure.
Dasgupta et al. (Sat,) studied this question.