Leveraging Noise and Aggressive Quantization of In-Memory Computing for Robust DNN Hardware Against Adversarial Input and Weight Attacks.
Sai Kiran CherupallyAdnan Siraj RakinShihui YinMingoo SeokDeliang FanJae-sun SeoPublished in: DAC (2021)
Keyphrases
- geometric attacks
- image noise
- digital image watermarking
- low cost
- image watermarking
- geometric distortions
- computing power
- quantization noise
- digital image watermarking scheme
- estimation error
- input data
- noise level
- real time
- noisy data
- noisy environments
- countermeasures
- noise reduction
- quantization error
- watermarking method
- virtual memory
- watermarking scheme
- label noise
- parallel hardware
- hardware implementation
- hardware and software
- massively parallel
- computer systems
- processing elements
- computational complexity
- neural network