RAB: Provable Robustness Against Backdoor Attacks

Keyphrases

• Provable Robustness 100%
• Backdoor Attack 100%
• Machine Learning Models 66%
• NK Model 33%
• Deep Neural Network 33%
• Robustness Bound 33%
• Evasion Attack 33%
• Tight 16%
• Smoothing Methods 16%
• Comprehensive Experiment 16%
• Unified Framework 16%
• Learning Strategies 16%
• Training Process 16%
• Noise Distribution 16%
• K-nearest 16%
• Work Support 16%
• Exact Algorithm 16%
• Comprehensive Evaluation 16%
• Theoretic Analysis 16%
• Training Model 16%
• Support Vector Machine 16%
• Model Robustness 16%
• Threat Model 16%
• Training Algorithm 16%
• Adversarial Attack 16%
• Robust Learning 16%
• Machine Learning Datasets 16%
• Tabular Data 16%
• Randomized Smoothing 16%
• MNIST 16%
• CIFAR-10 16%
• Robust Training 16%
• Certified Robustness 16%
• Spambase 16%
• Training Time Attacks 16%
• General Training 16%
• Backdoor Poisoning Attack 16%

Computer Science

• Backdoors 100%
• Machine Learning 57%
• Unified Framework 14%
• Training Process 14%
• Theoretic Analysis 14%
• Support Vector Machine 14%
• Threat Model 14%
• Exact Algorithm 14%
• training algorithm 14%
• Adversarial Machine Learning 14%
• Nearest Neighbors Classifier 14%