QuantAttack: Exploiting Quantization Techniques to Attack Vision Transformers

Computer Science

• Vision Transformer 100%
• Deep Neural Network 50%
• Matrix Multiplication 50%
• Energy Consumption 50%
• Computational Resource 50%
• Precision Performance 50%
• Art Performance 50%
• Dynamic Behavior 50%
• Adversarial Machine Learning 50%
• Memory Capacity 50%
• Graphics Processing Unit 50%

Keyphrases

• Contrast Dynamics 25%
• Hierarchical Vision Transformer 25%
• Low Bit Precision 25%