Binary thresholding defense against adversarial attacks

INIS

• images 100%
• defense 100%
• accuracy 75%
• speed 50%
• performance 50%
• calculation methods 50%
• architecture 25%
• layers 25%
• maps 25%
• neural networks 25%
• gain 25%
• trains 25%

Computer Science

• Adversarial Machine Learning 100%
• Gradient Descent 50%
• Computation Time 33%
• Convolutional Neural Network 16%
• Feature Map 16%
• Convolutional Layer 16%
• Adversarial Example 16%