Fair-Net: A Network Architecture for Reducing Performance Disparity between Identifiable Sub-populations

Keyphrases

• Network Architecture 100%
• Class Imbalance 100%
• Performance Disparities 100%
• Class-imbalanced Datasets 50%
• Classification Accuracy 25%
• Overall Accuracy 25%
• Benchmark Dataset 25%
• Population Specificity 25%
• Underrepresentation 25%
• Machine Learning Classifiers 25%
• Complex Architecture 25%
• Neural Network Architecture 25%
• Error Function 25%
• Majority Class 25%
• Minority Class 25%
• Real-world Benchmark 25%
• Output Layer 25%
• Probability Calibration 25%
• Accuracy Calibration 25%
• Multi-task Neural Network 25%
• Naive Approach 25%
• Population Disparities 25%
• Classification Probability 25%
• Global Classifier 25%
• Output Error 25%

Computer Science

• Network Architecture 100%
• Class Imbalance 100%
• Classification Accuracy 33%
• Neural Network Architecture 33%
• Majority Class 33%
• Minority Class 33%
• Error Function 33%
• Naïve Approach 33%
• Layer Function 33%
• Straightforward Extension 33%
• Machine Learning 33%
• Learning System 33%