Collective classification via discriminative matrix factorization on sparsely labeled networks

Keyphrases

• Classification Accuracy 20%
• Collective Classification 100%
• Compact Representation 20%
• Conjugate Gradient Method 20%
• Content Information 40%
• Content-centric Network 20%
• Discriminatory Power 20%
• Efficient Optimization 20%
• Empirical Loss Minimization 20%
• Existing Algorithms 20%
• Information Structure 20%
• Label Dependency 40%
• Labeled Networks 100%
• Latent Networks 20%
• Main Idea 20%
• Matrix Factorization 100%
• Neighbor Node 20%
• Network Representation 20%
• Network Representation Learning 20%
• Network Structure 20%
• New Matrix 20%
• Node Contents 40%
• Node Label 20%
• Optimization Algorithm 20%
• Performance Gain 20%
• Superior Performance 20%
• Topos 20%

Computer Science

• Classification Accuracy 20%
• Compact Representation 20%
• Conjugate Gradient Method 20%
• Content Information 40%
• Discriminative Power 20%
• Experimental Result 20%
• Information Structure 20%
• Loss Minimization 20%
• Matrix Factorization 100%
• Neighboring Node 20%
• Network Representation Learning 20%
• Network Structures 20%
• Objective Function 40%
• Optimization Algorithm 20%
• Performance Gain 20%
• Superior Performance 20%