MTL-Leak: Privacy Risk Assessment in Multi-Task Learning

Keyphrases

• Multi-task Learning 100%
• Privacy Risk Assessment 100%
• Privacy Concerns 30%
• Training Data 10%
• Multi-task 10%
• Generalization Ability 10%
• Learning Support 10%
• Model Driven Architecture 10%
• Shared Representations 10%
• Membership Inference Attack 10%
• Model Extraction Attack 10%
• Simultaneous Learning 10%
• Popular 10%
• Attack Performance 10%
• Multi-task Learning Framework 10%

Computer Science

• Multitask Learning 100%
• Model Architecture 11%
• Generalization Ability 11%
• Sensitive Informations 11%
• Training Data 11%
• Membership Inference Attack 11%