Locality Constrained-ℓ_p Sparse Subspace Clustering for Image Clustering

详细信息    查看全文

• 作者：Wenlong Cheng ^{cheng.python@gmail.com" class="auth_mail" title="E-mail the corresponding author}Author Vitae ; Tommy W.S. Chow ; ^{eetchow@cityu.edu.hk" class="auth_mail" title="E-mail the corresponding author}Author Vitae ; Mingbo Zhao ^{mzhao4@cityu.edu.hk" class="auth_mail" title="E-mail the corresponding author}Author Vitae
• 关键词：Subspace clustering ; Sparse coding ; ℓ1-norm minimization ; ℓp-norm minimization
• 刊名：Neurocomputing
• 出版年：2016
• 出版时间：12 September 2016
• 年：2016
• 卷：205
• 期：Complete
• 页码：22-31
• 全文大小：2322 K

文摘

In most sparse coding based image restoration and image classification problems, using the non-convex ℓ_p-norm minimization (0≤p<1) can often deliver better results than using the convex ℓ₁-norm minimization. Also, the high computational costs of ℓ₁-graph in Sparse Subspace Clustering prevent ℓ₁-graph from being used in large scale high-dimensional datasets. To address these problems, we in this paper propose an algorithm called Locality Constrained-ℓ_p Sparse Subspace Clustering (kNN-ℓ_p). The sparse graph constructed by locality constrained ℓ_p-norm minimization can remove most of the semantically unrelated links among data at lower computational cost. As a result, the discriminative performance is improved compared with the ℓ₁-graph. We also apply the k nearest neighbors to accelerate the sparse graph construction without losing its effectiveness. To demonstrate the improved performance of the proposed Locality Constrained-ℓ_p Sparse Subspace Clustering algorithm, comparative study was performed on benchmark problems of image clustering. Thoroughly experimental studies on real world datasets show that the Locality Constrained-ℓ_p Sparse Subspace Clustering algorithm can significantly outperform other state-of-the-art methods.