Mitigating Membership Inference Attacks by Self-Distillation Through a Novel Ensemble Architecture

Computer Science

• Artificial Intelligence 100%
• Machine Learning 66%
• Training Data 66%
• Regularization 33%
• Differential Privacy 33%
• Attackers 33%
• Privacy Preserving 33%
• Classification Accuracy 33%
• Major Component 33%
• Training Dataset 33%
• Privacy Leakage 33%

Keyphrases

• Self-distillation 100%
• Ensemble Architecture 100%
• SELENA 60%
• Membership Privacy 40%
• Adversarial Regularization 20%
• AI Architecture 20%
• Privacy-preserving Model 20%