A Research of Power Analysis Based on Multiple Neural Network Structures

In order to explore the performance difference of different deep neural networks used in power analysis attacks, DPA_Contest_V4 dataset is used to complete the experiment. After cracking the loop mask, first we compare the deep neural network with the traditional machine learning algorithm model such as SVM. Then we analyze the influence of the changes in the structure of the neural network model on the experimental results. Finally, combined with the recurrent neural network, different network models are compared comprehensively. The result shows that under the same experimental conditions, the neural network model is superior to the traditional machine learning model and the recurrentneural network model is superior to the deep neural network model, in which different layers of neural networks taking different activation functions lead to large changes.