Trade-offs and guarantees of adversarial representation learning for information obfuscation

Keyphrases

• Information Leakage 100%
• Adversarial Learning 100%
• Information Obfuscation 100%
• Task Accuracy 66%
• Obfuscation 66%
• Formulation Optimization 33%
• Potential Information 33%
• Learning Approaches 33%
• Adversary 33%
• Information-theoretic Bounds 33%
• Crowdsourced Data 33%
• Sensitive Attribute 33%
• Minimax Optimization 33%
• Accuracy Maximization 33%
• Machine Learning Service 33%

Computer Science

• Representation Learning 100%
• Information Leakage 100%
• Obfuscation 100%
• Theoretical Framework 33%
• Learning Approach 33%
• Optimization Formulation 33%
• Sensitive Informations 33%
• Sensitive Attribute 33%
• Machine Learning 33%
• Learning System 33%