Showing 1–1 of 1 results for author: Pieloch, A

Supervised Learning of Labeled Pointcloud Differences via Cover-Tree Entropy Reduction

Authors: Abraham Smith, Paul Bendich, John Harer, Alex Pieloch, Jay Hineman

Abstract: We introduce a new algorithm, called CDER, for supervised machine learning that merges the multi-scale geometric properties of Cover Trees with the information-theoretic properties of entropy. CDER applies to a training set of labeled pointclouds embedded in a common Euclidean space. If typical pointclouds corresponding to distinct labels tend to differ at any scale in any sub-region, CDER can ide… ▽ More We introduce a new algorithm, called CDER, for supervised machine learning that merges the multi-scale geometric properties of Cover Trees with the information-theoretic properties of entropy. CDER applies to a training set of labeled pointclouds embedded in a common Euclidean space. If typical pointclouds corresponding to distinct labels tend to differ at any scale in any sub-region, CDER can identify these differences in (typically) linear time, creating a set of distributional coordinates which act as a feature extraction mechanism for supervised learning. We describe theoretical properties and implementation details of CDER, and illustrate its benefits on several synthetic examples. △ Less

Submitted 19 January, 2018; v1 submitted 25 February, 2017; originally announced February 2017.

Comments: Distribution Statement A - Approved for public release, distribution is unlimited. Version 2: added link to code, and some minor improvements. Version 3: updated authors and thanks

MSC Class: 62H30; 60G55