Elastic nonnegative matrix factorization
作者：
Highlights：
• ENMF introduces an elastic loss that takes advantage of both Frobenius norm and ℓ2, 1 norm when noise distribution is uncertain, therefore ENMF is far more insensitive to noise and outliers.
• ENMF takes the geometric information of the projected data points in the low dimensional manifold as feedback to construct the affinity graph, hence ENMF can handle the situation that a few exceptional data pairs are close in the original space but far away from each other in the manifold.
• ENMF utilizes the exclusive LASSO to enhance the intra-cluster competition and therefore the “winner” is more likely to stand out while the “loser” tends to be out in a sparse manner.
• ENMF provides consistently better clustering results on several well-known data sets as compared to standard NMF and several other variants of the NMF algorithm.
摘要
•ENMF introduces an elastic loss that takes advantage of both Frobenius norm and ℓ2, 1 norm when noise distribution is uncertain, therefore ENMF is far more insensitive to noise and outliers.•ENMF takes the geometric information of the projected data points in the low dimensional manifold as feedback to construct the affinity graph, hence ENMF can handle the situation that a few exceptional data pairs are close in the original space but far away from each other in the manifold.•ENMF utilizes the exclusive LASSO to enhance the intra-cluster competition and therefore the “winner” is more likely to stand out while the “loser” tends to be out in a sparse manner.•ENMF provides consistently better clustering results on several well-known data sets as compared to standard NMF and several other variants of the NMF algorithm.
论文关键词：NMF,Elastic,Robust,Manifold,Clustering,Exclusive LASSO
论文评审过程：Received 21 October 2017, Revised 24 May 2018, Accepted 2 July 2018, Available online 17 July 2018, Version of Record 1 March 2019.
论文官网地址：https://doi.org/10.1016/j.patcog.2018.07.007
