Reduced row echelon form and non-linear approximation for subspace segmentation and high-dimensional data clustering
نویسندگان
چکیده
Article history: Received 10 May 2012 Received in revised form 9 December 2013 Accepted 15 December 2013 Available online 17 December 2013 Communicated by Jared Tanner
منابع مشابه
Reduction and Null Space Algorithms for the Subspace Clustering Problem
This paper presents two algorithms for clustering high-dimensional data points that are drawn from a union of lower dimensional subspaces. The first algorithm is based on binary reduced row echelon form of a data matrix. It can solve the subspace segmentation problem perfectly for noise free data, however, it is not reliable for noisy cases. The second algorithm is based on Null Space represent...
متن کاملBinary Reduced Row Echelon Form Approach for Subspace Segmentation
This paper introduces a subspace segmentation and data clustering method for a set of data drawn from a union of subspaces. The proposed method works perfectly in absence of noise, i.e., it can find the number of subspaces, their dimensions, and an orthonormal basis for each subspace. The effect of noise on this approach depends on the noise level and relative positions of subspaces. We provide...
متن کاملA Constrained Clustering Algorithm for Spare Parts Segmentation
This paper explores the application of constrained clustering algorithms to inventory grouping based problems. Our research in this area is based upon a segmentation methodology that considers operationally relevant part attributes. In this paper, we present a brief review of literature on multi-echelon inventory models, and clustering applications to inventory segmentation. Finally, we discuss...
متن کاملSubspace Clustering with Applications to Dynamical Vision ( CS 229 Final Project )
Data that arises from engineering applications often contains some type of low dimensional structure that enables intelligent representation and processing. This leads to a very challenging problem: discovering compact representations of high-dimensional data. A very common approach to address this problem is modeling data as a mixture of multiple linear (or affine) subspaces. Given a set of da...
متن کاملSubspace Segmentation by Successive Approximations: A Method for Low-Rank and High-Rank Data with Missing Entries
We propose a method to reconstruct and cluster incomplete high-dimensional data lying in a union of low-dimensional subspaces. Exploring the sparse representation model, we jointly estimate the missing data while imposing the intrinsic subspace structure. Since we have a non-convex problem, we propose an iterative method to reconstruct the data and provide a sparse similarity affinity matrix. T...
متن کامل