Modern information security’s last line of defense depends on the confidentiality of devices. When the hardware implements encryption and decryption operations, it is easy to inadvertently leak sensitive information and affect security, such as execution time, power consumption, electromagnetic radiation, etc. This type of research was called Side-Channel Cryptanalysis more than 20 years ago, but it is now called Side-Channel Analysis (SCA) and has received widespread international attention. The speech will mention that compared with traditional analysis methods, such as Simple Power Analysis (SPA), Correlation Power Analysis (CPA), Template Attack, etc. The application of Machine Learning to enhance the side-channel analysis technology, regardless of whether the encryption and decryption algorithm is protected, the analysis results based on Deep Learning have been proven to have an overwhelming advantage in recent years.