Learning credible DNNs via incorporating prior knowledge and model local explanation

Keyphrases

• Adversarial Attack 33%
• Adversarial Vulnerability 33%
• Algorithmic Discrimination 33%
• Attention Model 33%
• Classification Datasets 33%
• Data Dependency 33%
• Demographic Groups 33%
• DNN Model 100%
• DNN Training 33%
• Equality of Opportunity 33%
• Holdout 33%
• Local Explanations 100%
• LSTM Model 33%
• Meaningful Improvement 33%
• Model Discrimination 33%
• Network Architecture 33%
• Overfitting 33%
• Text Classification 33%
• Training Process 33%
• Unseen 33%

Computer Science

• Adversarial Machine Learning 10%
• Convolutional Neural Network 10%
• Deep Neural Network 100%
• Demographic Group 10%
• Experimental Result 20%
• Long Short-Term Memory Network 10%
• Model Discrimination 10%
• Network Architecture 10%
• Text Classification 10%
• Training Process 10%

Psychology

• Discrimination 100%