Constrained submodular minimization for missing labels and class imbalance in multi-label learning

Computer Science

• Class Imbalance 100%
• Positive Instance 40%
• Training Instance 20%
• Approximation (Algorithm) 20%
• Experimental Result 20%
• Linear Program 20%
• Objective Function 20%
• Cardinality 20%
• Performance Degradation 20%
• Bound Constraint 20%

Keyphrases

• Label Imbalance 100%
• Negative Label 20%
• Lovász Extension 20%
• Cardinality Constraint 20%
• Class Cardinalities 20%