A link prediction algorithm based on low-rank matrix completion

Gao, Man; Chen, Ling; Li, Bin; Liu, Wei

doi:10.1007/s10489-018-1220-4

Cited by 15 publications

(4 citation statements)

References 53 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Network Big Data Compared with a missing data set, if the missing data are replaced by the inter-cell, the information entropy of the large data set will increase significantly [29][30][31][32][33][34][35][36].…”

Section: High Rank Matrix Filling Algorithm Formentioning

confidence: 99%

Algorithm for Filling High Rank Matrix of Network Big Data Based on Density Peak Clustering

Liu

2022

Mathematical Problems in Engineering

View full text Add to dashboard Cite

The traditional filling method for network big data matrix has poor filling effect and suffers from noise. Therefore, a filling algorithm for network big data high rank matrix based on density peak clustering is proposed. The missing data are replaced by small-interval data, the information entropy of the high rank matrix of network big data is calculated, the density peak clustering algorithm is optimized through the cluster center selection strategy, the block data set is obtained through the unknown block method, and the block filling is realized by the host filling algorithm. Experimental results show that the filling accuracy of the proposed algorithm is as high as 0.895, and the loss rate is between 2% and 12%.

show abstract

Section: High Rank Matrix Filling Algorithm Formentioning

confidence: 99%

Algorithm for Filling High Rank Matrix of Network Big Data Based on Density Peak Clustering

Liu

2022

Mathematical Problems in Engineering

View full text Add to dashboard Cite

show abstract

“…According to the definition of the nuclear norm inequality, we have ‖∑ (𝛼 𝒑 𝒒 ) ‖ * ≤ ∑ ‖𝛼 𝒑 𝒒 ‖ * . Therefore, we write the regularized part of (5) in the form of γ‖𝛼 𝒖 𝒗 ‖ * in (6). The values of 𝛼 , 𝒑 , and 𝒒 depend on 𝒫 (𝑹 ), which has values only within the range of Ω.…”

Section: B Low-rank Learningmentioning

confidence: 99%

“…Matrix completion is to recover an intact matrix via its partial observations, which has drawn much attention in the past several years for its numerous applications, such as image inpainting [1], [2], recommendation systems [3], [4], relational networks [5], [6], multi-task learning [7], [8] and so on. At present, most matrix completion methods are based on the critical premise assumption of a low-rank or approximately low-rank of observation matrix.…”

Section: Introductionmentioning

confidence: 99%

Matrix Completion Based on Low-Rank and Local Features Applied to Images Recovery and Recommendation Systems

Zhang

Liang

et al. 2022

IEEE Access

View full text Add to dashboard Cite

Recent advances have shown that the challenging problem of matrix completion arises from real-world applications, such as image recovery, and recommendation systems. Existing matrix completion methods utilize the low-rank property of sparse matrices to fill missing entries, which essentially exploits the row-rank relationship and ignores the local features of sparse matrices. In this paper, we propose a novel matrix completion method that takes the path of combining local features and low-rank information. First, the original sparse matrix is processed so that the neighboring rows and columns of the sparse matrix are most similar to each other, and then the missing data are filled by the iterative rank-one matrix completion method. Moreover, we use two-dimensional convolutional operations of sparse matrices to obtain the local features to dredge up the missing entries. Finally, the two filled results are integrated to obtain the final missing entries. We conduct extensive experiments on the real-world datasets. The experimental results demonstrate the significant outperforms of the proposed method on five image datasets of different sizes, and show a strong competitive advantage on the recommender system datasets.

show abstract

“…They build a model, e.g., hierarchical structure model [25], stochastic block model [19], and forward generative model [26], to fit the structure and hence to estimate the linkage likelihood of non-observed links. Matrix factorization methods regard link prediction as matrix completion problem, which can extract latent features and/or use additional features to perform prediction [27]- [29]. For example, Pech et al decomposed the adjacency matrix, by introducing the robust principle component analysis, into a low-rank matrix denoting the network backbone and a sparse matrix representing the spurious links in network [28].…”

Section: Tablementioning

confidence: 99%

Effective Link Prediction Based on Community Relationship Strength

Fang

Bai

et al. 2019

IEEE Access

View full text Add to dashboard Cite

Link prediction is one of the research hotspots in complex network analysis and has a wide range of applications in both theory and reality. To improve the prediction accuracy, this paper proposes a new link prediction framework by considering both node similarity and community information, which overcomes the weaknesses of existing community-based prediction methods. In the proposed framework, a reasonable measure, called community relationship strength (CRS), is defined to estimate the closeness between communities. In this paper, we hold the view that the connection likelihood between two target nodes rests upon not only their similarity but also the closeness of communities that they belong to. Therefore, to measure the connection likelihood, the proposed framework combines CRS with traditional similarity indexes. Three CRS-based methods are derived from the framework. The performance of the CRS-based methods is comprehensively studied on 12 real-world networks compared with several groups of baselines. The experimental results indicate that the CRS-based methods are more effective and robust than others. INDEX TERMS Link prediction, community structure, similarity index, complex networks. I. INTRODUCTION A network is a structure that is composed of nodes and their static or dynamic relations, called edges or links. This simple structure can model a large variety of complex systems such as social, biological and technological systems, in which nodes are entities in these systems and links indicate the relations between entities [1]-[3]. The proliferation of data has inspired the study of network mining and complex network analysis. A wealth of interesting problems related to network mining, such as community detection [4], [5], node rank [6], and network embedding [7], have been studied. Among these interesting network-related problems, link prediction has been receiving increasing interest from disparate disciplines because not only many available real-world networks are incomplete [8], [9], but also its wide range of applications in both theory and reality [10], [11]. For examples, link prediction can be used in exploring network evolution mechanisms [12], [13], finding the cooperation opportunities among scientists [14], [15], recommending friends in social The associate editor coordinating the review of this manuscript and approving it for publication was Lin Wang.

show abstract

A link prediction algorithm based on low-rank matrix completion

Cited by 15 publications

References 53 publications

Algorithm for Filling High Rank Matrix of Network Big Data Based on Density Peak Clustering

Algorithm for Filling High Rank Matrix of Network Big Data Based on Density Peak Clustering

Matrix Completion Based on Low-Rank and Local Features Applied to Images Recovery and Recommendation Systems

Effective Link Prediction Based on Community Relationship Strength

Contact Info

Product

Resources

About