Tag
This paper proposes CSDG, a method for offline reinforcement learning that expresses Bellman backups as in-sample targets plus a convex-hull-neighborhood local correction, controlling OOD action estimation errors and improving value stability.
This post discusses the use of SIMD to optimize collision detection for convex hulls in Box3D, particularly using wide SIMD and the Separating Axis Test to improve performance for hulls with many edges.
A web-based tool that computes and visualizes spherical Voronoi diagrams by using a randomized incremental algorithm for 3D convex hull, equivalent to the Delaunay triangulation on a sphere.