Are Your Sensitive Attributes Private? Novel Model Inversion Attribute Inference Attacks on Classification Models

Keyphrases

• Classification Model 100%
• Model Inversion 100%
• Sensitive Attribute 100%
• Attribute Inference Attack 100%
• Adversary 75%
• Training Data 75%
• Non-sensitive 50%
• Confidence Score 50%
• Machine Learning Technology 50%
• Vulnerability 25%
• Black Box 25%
• Decision Tree 25%
• Score Model 25%
• Score-based 25%
• Medical Diagnosis 25%
• Business Decisions 25%
• Deep Neural Network 25%
• Label-only 25%
• Specific Group 25%
• Target Classification 25%
• Machine Learning Models 25%
• Privacy Sensitive 25%
• Sensitive Domains 25%
• Attack Effectiveness 25%

Computer Science

• Classification Models 100%
• Sensitive Attribute 100%
• Training Data 50%
• Confidence Score 50%
• Machine Learning Technology 50%
• Training Dataset 25%
• Medical Diagnosis 25%
• Deep Neural Network 25%
• Business Decision 25%
• Decision Tree Model 25%
• Sensitive Domain 25%