Precise Extraction of Deep Learning Models via Side-Channel Attacks on Edge/Endpoint Devices
The author argues that precise extraction of deep learning models can be achieved through side-channel attacks, emphasizing the importance of model information like ID and MA for successful attacks.