Deep Loss Convexification for Learning Iterative Models
원문 초록
Iterative methods such as iterative closest point (ICP) for point cloud registration often suffer from bad local optimality (e.g. saddle points), due to the nature of nonconvex optimization. To address this fundamental challenge, in this paper we propose learning to form the loss landscape of a deep iterative method w.r.t. predictions at test time into a convex-like shape locally around each ground truth given data, namely Deep Loss Convexification (DLC), thanks to the overparametrization in neural networks. To this end, we formulate our learning objective based on adversarial training by manipulating the ground-truth predictions, rather than input data. In particular, we propose using star-convexity, a family of structured nonconvex functions that are unimodal on all lines that pass through a global minimizer, as our geometric constraint for reshaping loss landscapes, leading to (1) extra novel hinge losses appended to the original loss and (2) near-optimal predictions. We demonstrate the state-of-the-art performance using DLC with existing network architectures for the tasks of training recurrent neural networks (RNNs), 3D point cloud registration, and multimodel image alignment.
함께 살펴볼 논문
공통 태그를 근거로 연결했습니다. 실제 인용 관계나 논문의 우수성을 뜻하지 않습니다.
Cross-modal feature fusion for robust point cloud registration with ambiguous geometry
Registration of 3D Point Sets Using Exponential-based Similarity Matrix
Matching Distance and Geometric Distribution Aided Learning Multiview Point Cloud Registration
ColabSfM: Collaborative Structure-from-Motion by Point Cloud Registration
Unlocking Generalization Power in LiDAR Point Cloud Registration
BUFFER-X: Towards Zero-Shot Point Cloud Registration in Diverse Scenes
토론
이 논문과 연결된 포럼 글이 아직 없습니다.