Motivation: Previous studies have demonstrated that machine learning based molecular cancer classification usinggene expression profiling (GEP) data is promising for the clinic diagnosis and treatment of cancer. Novelclassification methods with high efficiency and prediction accuracy are sti