Exploring Machine Learning Privacy/Utility Trade-Off from a Hyperparameters Lens

Computer Science

• Machine Learning 100%
• Gradient Descent 100%
• Privacy Preserving 100%
• Activation Function 100%
• Design-Space Exploration 33%
• Data Privacy 33%
• User Data 33%
• Learning Rate 33%
• New-State 33%
• Preserving Property 33%
• Batch Normalization 33%

Keyphrases

• Machine Learning Privacy 100%
• Differentially Private Stochastic Gradient Descent 100%
• Private Model 33%
• Model Privacy 33%