{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/ursa-a-neural-network-for-unordered-point","title":"URSA: A Neural Network for Unordered Point Clouds Using Constellations","arxiv_id":"1808.04848","date":"2018-08-14","proceeding":null,"authors":["Mark B. Skouson","Brett J. Borghetti","Robert C. Leishman"],"abstract":"This paper describes a neural network layer, named Ursa, that uses a\nconstellation of points to learn classification information from point cloud\ndata. Unlike other machine learning classification problems where the task is\nto classify an individual high-dimensional observation, in a point-cloud\nclassification problem the goal is to classify a set of d-dimensional\nobservations. Because a point cloud is a set, there is no ordering to the\ncollection of points in a point-cloud classification problem. Thus, the\nchallenge of classifying point clouds inputs is in building a classifier which\nis agnostic to the ordering of the observations, yet preserves the\nd-dimensional information of each point in the set. This research presents\nUrsa, a new layer type for an artificial neural network which achieves these\ntwo properties. Similar to new methods for this task, this architecture works\ndirectly on d-dimensional points rather than first converting the points to a\nd-dimensional volume. The Ursa layer is followed by a series of dense layers to\nclassify 2D and 3D objects from point clouds. Experiments on ModelNet40 and\nMNIST data show classification results comparable with current methods, while\nreducing the training parameters by over 50 percent.","url_abs":"http://arxiv.org/abs/1808.04848v2","url_pdf":"http://arxiv.org/pdf/1808.04848v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"ursa-a-neural-network-for-unordered-point","repo_url":"https://github.com/RadicalAcronym/URSA","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"point-cloud-classification","task_name":"Point Cloud Classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}