On the transferability of adversarial examples between convex and 01 loss models

Keyphrases

• 0-1 Loss 100%
• Activation Loss 13%
• Activation Network 6%
• Adversarial Examples 100%
• Adversary 20%
• Black Hole Attack 6%
• CIFAR-10 13%
• Convex Model 26%
• Convolutional Networks 6%
• Decision Boundary 6%
• Different Boundaries 6%
• Discretize 6%
• Dual-layer 20%
• Hinge Loss 6%
• ImageNet 6%
• Logistic Loss 6%
• Loss Model 100%
• MNIST 13%
• Neural Network 13%
• Non-continuous 6%
• Non-convex Cost Function 100%
• Non-transferability 40%
• Non-uniqueness 6%
• Sigmoid Activation Function 6%
• Substitute Model 13%
• Transferability 100%
• White-box 6%

Computer Science

• Adversarial Example 100%
• Attack Model 25%
• Black-Box Attack 25%
• Convolutional Network 25%
• Decision Boundary 25%
• Layer Neural Network 50%

Mathematics

• Black Box 14%
• Loss Model 100%
• Neural Network 28%
• Nonuniqueness 14%