Lung Adenocarcinoma (LUAD) is one of the most fatal malignant tumors which has led to millions of deaths yearly all around the world. Arouse from new topic of selecting DNA methylation markers in cancer therapy diagnosis, we explore relationship between gene expression and DNA methylation, we applied machine learning unsupervised algorithms in narrowing the range of candidate DNA methylation probes and create a deep-learning-based diagnosis model which has high accuracy in predicting cells into malignant and benign.