Side channel attack method and system based on deep learning

The invention discloses a side channel attack method and system based on deep learning, relates to the technical field of side channel attacks, and aims to solve the problem that the efficiency and precision of side channel attacks are affected due to the fact that an existing deep learning network often excessively pays attention to non-key parts around an information area. Comprising the following steps: S1, acquiring a data set of attack tracks of a side channel; s2, preprocessing the collected data; s3, a side channel attack network model is established, the network uses a plurality of convolution kernels of different sizes, different convolution operations and pooling operations are carried out during feature extraction, feature extraction of different scales is carried out on data to obtain global information, attention to non-key parts around an informative region is reduced, and the network performance is improved. Therefore, the data can be better represented; s4, training and testing the model; and