In the previous section, we prepared the training and test sets. However, we need to put some extra efforts into making them consumable by DL4J. To be more specific, DL4J expects the training data as numeric and the last column to be the label column, and the remaining are features.
We will now try to prepare our training and test sets like that. First, we show the files where we saved the training and test sets:
String trainPath = "data/TCGA_train.csv"; // training setString testPath = "data/TCGA_test.csv"; // test set
Then, we define the required parameters, such as the number of features, number of classes, and batch size. Here, I use 128 as the batchSize but adjust it accordingly:
int labelIndex = 20531;// number of ...