{"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/view-inter-prediction-gan-unsupervised","title":"View Inter-Prediction GAN: Unsupervised Representation Learning for 3D Shapes by Learning Global Shape Memories to Support Local View Predictions","arxiv_id":"1811.02744","date":"2018-11-07","proceeding":null,"authors":["Zhizhong Han","Mingyang Shang","Yu-Shen Liu","Matthias Zwicker"],"abstract":"In this paper we present a novel unsupervised representation learning\napproach for 3D shapes, which is an important research challenge as it avoids\nthe manual effort required for collecting supervised data. Our method trains an\nRNN-based neural network architecture to solve multiple view inter-prediction\ntasks for each shape. Given several nearby views of a shape, we define view\ninter-prediction as the task of predicting the center view between the input\nviews, and reconstructing the input views in a low-level feature space. The key\nidea of our approach is to implement the shape representation as a\nshape-specific global memory that is shared between all local view\ninter-predictions for each shape. Intuitively, this memory enables the system\nto aggregate information that is useful to better solve the view\ninter-prediction tasks for each shape, and to leverage the memory as a\nview-independent shape representation. Our approach obtains the best results\nusing a combination of L_2 and adversarial losses for the view inter-prediction\ntask. We show that VIP-GAN outperforms state-of-the-art methods in unsupervised\n3D feature learning on three large scale 3D shape benchmarks.","url_abs":"http://arxiv.org/abs/1811.02744v1","url_pdf":"http://arxiv.org/pdf/1811.02744v1.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":[],"tasks":[{"task_slug":"3d-point-cloud-linear-classification","task_name":"3D Point Cloud Linear Classification"},{"task_slug":"prediction","task_name":"Prediction"},{"task_slug":"representation-learning","task_name":"Representation Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/3d-point-cloud-linear-classification-on","task":"3D Point Cloud Linear Classification","dataset":"ModelNet40","model":"VIP-GAN","rank_in_archive_order":15,"of":20,"metrics":{"Overall Accuracy":"90.2"},"uses_additional_data":false}],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=1811.02744","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}