CSI NN: Reverse Engineering Neural Network Structures via Electromagnetic Side Channels

CSI NN: Reverse Engineering Neural Network Structures via Electromagnetic Side Channels

Abstract Machine learning has become a mainstream solution applied across various industries. In this work, we investigate how to use side-channel information such as timing and electromagnetic (EM) emissions to perform reverse engineering of neural networks. To this end, we consider multilayer perceptrons and convolutional neural networks as the choices of machine learning architecture, assuming … Read more

Constructing an Oscilloscope with Fault Injection: Simple/Related Fault Attacks on Cryptographic Algorithms

Constructing an Oscilloscope with Fault Injection: Simple/Related Fault Attacks on Cryptographic Algorithms

Is there an inherent connection between passive side-channel analysis and active fault injection? Author | Shuyuan Editor | Pan Qizhi Background Introduction Side-channel analysis of cryptographic algorithms refers to the process of inferring intermediate values of cryptographic operations by collecting and analyzing the physical information leaked by cryptographic chips during computation, thereby recovering the keys. … Read more