Deep learing is an active research area in machine learning community. Its main idea is to discover high-level abstract concepts in original datasets with huge computational power of the age of big data, by increasing the number of layers of the learners, so as to increase the size of channels and the quantity of parameters. It becomes a significant information source for decision support of application domains. We explore the methods of applying deep learning technologies in the data analysis tasks of information systems by presenting the main principles and implemetation details of two deep learning models, convolutionan neural network and stacked auto-encoders in emphasis, their application cases in the data analysis of information system, as well as the analysis on their application value.