A new arXiv preprint introduces two methods for boundary detection in high-dimensional point cloud data. The methods are explicitly intended for settings with non-uniform density, high ambient dimension, and substantial noise.
The abstract describes the first method as fitting a separating surface, but the description is cut off in the available text. Details of the second method, along with any experimental results or comparisons, are not included in the abstract.
Because only the abstract is available, the full evaluation and practical performance of these methods remain to be examined in the preprint itself.