期刊
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
卷 35, 期 8, 页码 8584-8595出版社
IEEE COMPUTER SOC
DOI: 10.1109/TKDE.2022.3206881
关键词
l(2,0)-norm; feature selection; row-sparse NMF and ONMF; non-convex optimization; scRNA-seq data clustering
Non-negative matrix factorization (NMF) is a powerful tool for dimensionality reduction and clustering. This paper introduces a row-sparse NMF with l(2,0)-norm constraint (NMF_l(20)), which incorporates feature selection by constraining the basis matrix W using the l(2,0)-norm constraint. The problem of solving the model is addressed by proving that the l(2,0)-norm constraint satisfies the Kurdyka-Lojasiewicz property and proposing a proximal alternating linearized minimization algorithm and its accelerated version.
Non-negativematrix factorization (NMF) is a powerful tool for dimensionality reduction and clustering. However, the interpretation of the clustering result from NMF is difficult, especially for the high-dimensional biological data without effective feature selection. To address this problem, we introduce a row-sparse NMF with l(2,0)-norm constraint (NMF_l(20)), where the basis matrix W is constrained by using the l(2,0)-norm constraint such that W has a row-sparsity pattern with feature selection. However, it is a challenge to solve the model, because the l(2,0)-norm constraint is a non-convex and non-smooth function. Fortunately, we prove that the l(2,0)-norm constraint satisfies the Kurdyka-Lojasiewicz property. Based on this finding, we present a proximal alternating linearized minimization algorithm and its monotone accelerated version to solve the NMF_l(20) model. In addition, we further present a orthogonal NMF with (2,0)-norm constraint (ONMF_l(20)) to enhance the clustering performance by using a non-negative orthogonal constraint. The ONMF_l(20) model is solved by transforming into a series of constrained and penalized matrix factorization problems. The convergence and guarantees for these proposed algorithms are proved and the computational complexity is well evaluated. The results on numerical and scRNA-seq datasets demonstrate the efficiency of our methods in comparison with existing methods.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据