Abstract: Machine learning and template attack in the traditional side channel attack techniques have similar procedures, they all consist of two phases:learning and testing. Template attack can be considered as a classification technique for supervised learning, and there are many such classification algorithms in the machine learning field. In order to explore the application of machine learning algorithms in side channel attack, using the data movement operation in actual cryptographic chip as the attack target, the forecasting effect of some machine learning algorithms is investigated. These algorithms make use of power traces with known value of the moved data, then predict the value of the moved data for some new power traces. The results show that, when employing only one power trace in the testing stage, some machine learning algorithms have higher correctness rate than template attack.