Ex2.py defines a Multi-layer percepteron in Keras, compiles and trains the network. EX2 Documentation
Ex3.py defines a convolutional neural network in Keras, compiles and trains the network. EX3 Documentation
ROC.py calculates and plots the ROC curve of true positive rate (recall) vs false positive rate with 1-precision, from the “detector groundtruth.dat” and “detector output.dat”. ROC analysis provides a systematic tool for quantifying the impact of variability among individuals' decision thresholds. ROC stands for ”Receiver Operating Characteristic”. Wikipedia
Mnist.py is an auto-encoder that trains the neural network to encode and decode noisy images and to report different values from our network.