Towards Defending against Adversarial Examples via Attack-Invariant Features

Keyphrases

• Invariant Features 100%
• Adversarial Examples 100%
• Unseen 75%
• Adversarial Noise 75%
• Attack Strategy 50%
• Adversarial Learning 25%
• Deep Neural Network 25%
• Learning Mechanisms 25%
• Adversarial Robustness 25%
• Attack Model 25%
• Semantic Classification 25%
• Classification Information 25%
• Adaptive Attack 25%

Computer Science

• Adversarial Example 100%
• Representation Learning 25%
• Invariant 25%
• Deep Neural Network 25%
• Learning Mechanism 25%
• Attack Model 25%
• Normalization Term 25%