Improved Generalization Bounds for Adversarially Robust Learning

Keyphrases

• Generalization Bounds 100%
• Robust Learning 100%
• Adversary 75%
• Binary Classification 75%
• Corruption 50%
• Multi-class 50%
• Adversarial Examples 50%
• Class Function 25%
• Minimization Algorithm 25%
• Zero-sum Game 25%
• Training Data 25%
• Oracle 25%
• Black Box 25%
• Sample Complexity 25%
• Training Samples 25%
• Multi-class Classification 25%
• Near-optimal Policy 25%
• Regret Minimization 25%
• Adversarial Environment 25%
• Robust Classifier 25%
• Dimension Complexity 25%
• K-fold 25%
• Rademacher Complexity 25%
• Fat-shattering Dimension 25%

Mathematics

• Binary Classification 100%
• Main Result 50%
• Optimal Policy 50%
• Training Sample 50%
• Black Box 50%
• Training Data 50%

Computer Science

• Binary Classification 100%
• Adversarial Example 66%
• Training Sample 33%
• Training Data 33%
• Multiclass Classification 33%