Fig. 2From: Cancer type prediction based on copy number aberration and chromatin 3D structure with convolutional neural networksPerformances of our proposed method against three widely adopted data classifiers. a The comparison methods use raw CNA input data (without HiC). From left to right: Our method, SVM (polynomial kernel), KNN (number of neighbors = 5 and p = 2) and NB (multinomial distribution). Our method shows significant advantage against the comparison methods. b The comparison methods use both CNA and HiC as input data. From left to right: Our method, SVM (polynomial kernel), KNN (number of neighbors = 5 and p = 2) and NB (multinomial distribution). Our method shows even greater advantage against the comparison methodsBack to article page