A self coded ANN which can be trained on any data for classification or logistic regression.
Hi there, This is a self coded Neural-Net, made just by using standard libraries like: 1) Numpy : For matrix calculation. 2) Pandas : For handling the datasets. 3) Matplotlib : For graph plotting. 4) Time : For claculating computing time. The class neuralnet consists of 3 main functions: 1) __init__ : This is python constructor used for initializing weights and biases. 2) train : For training the net on the data. 3) test : For test the trained net. I have trained it using mnist dataset. The output displayed will be: 1) The acurracy of the model on the testing data. 2) The graph of cost function vs No. of iterations.