Small-Variance Nonparametric Clustering on the Hypersphere
Straub, Campbell, How, Fisher · cs.CV,math.ST,stat.AP,stat.TH · 2016-07-21 · 原文
Structural regularities in man-made environments reflect in the distribution of their surface normals. Describing these surface normal distributions is important in many computer vision applications, such as scene understanding, plane segmentation, and regularization of 3D reconstructions. Based on the small-variance limit of Bayesian nonparametric von-Mises-Fisher (vMF) mixture distributions, we propose two new flexible and efficient k-means-like clustering algorithms for directional data such as surface normals. The first, DP-vMF-means, is a batch clustering algorithm derived from the Dirichlet process (DP) vMF mixture. Recognizing the sequential nature of data collection in many applications, we extend this algorithm to DDP-vMF-means, which infers temporally evolving cluster structure from streaming data. Both algorithms naturally respect the geometry of directional data, which lies on the unit sphere. We demonstrate their performance on synthetic directional data and real 3D surface normals from RGB-D sensors. While our experiments focus on 3D data, both algorithms generalize to high dimensional directional data such as protein backbone configurations and semantic word vectors.
讲义
讲义·推断 依据「原文」自动生成的结构化摘要（推断），非原文表述；以原文为准。
1. 人话版
Structural regularities in man-made environments reflect in the distribution of their surface normals.
Describing these surface normal distributions is important in many computer vision applications, such as scene understanding, plane segmentation, and regularization of 3D reconstructions.
2. 领域脉络
本文类目：cs.CV、math.ST、stat.AP、stat.TH，属于其所在研究脉络的最新进展。
3. 机制拆解
The first, DP-vMF-means, is a batch clustering algorithm derived from the Dirichlet process (DP) vMF mixture.
Recognizing the sequential nature of data collection in many applications, we extend this algorithm to DDP-vMF-means, which infers temporally evolving cluster structure from streaming data.
4. 证据与数字
We demonstrate their performance on synthetic directional data and real 3D surface normals from RGB-D sensors.
While our experiments focus on 3D data, both algorithms generalize to high dimensional directional data such as protein backbone configurations and semantic word vectors.
5. 反例与边界
Based on the small-variance limit of Bayesian nonparametric von-Mises-Fisher (vMF) mixture distributions, we propose two new flexible and efficient k-means-like clustering algorithms for directional data such as surface normals.
6. 跨领域连接与意外收获
横跨 4 个类目（cs.CV、math.ST、stat.AP、stat.TH），关注其在你兴趣板块间的迁移面。
7. 可复用方法
把本文机制与你手头项目对照，找一个两周内能验证的最小实验。
8. 术语表
精读时把不熟的术语记入此处，作为下次回忆的锚点。