Thanks for your interest in our work! The paper can be referred at http://ieeexplore.ieee.org/document/7888298/. 1. gendata.m is used to generate miner audio segments. 2. bkclassify.m is for softmax classificationas and generating the mat file. 3. mat2gz.py can transfer .mat to .gz, like the 3 given files at 1. 4. File 74-79 need .pkl.gz files to run, and will output the identification result. 5. plot_confusion_matrix.py is used to draw the confusion figure. 6. voting_tsne.py is used to draw the dimensionality reduction result, as well as generate the voting result. For any questions, please contact me at smart_qi@126.com.