This repo is about my practice implementation of Artificial Neural Networks with MNIST hand written digit dataset.
The first folder is about practice implementation of multi-class classification with ANN on keras built-in hand written MNIST dataset. The implementations are both with TensorFlow and Pytorch. In this multi-class classification of hand written digits, the predicted-labels VS true-labels are evaluated with Confusion Matrix table, and the predicted results are shown in True-classified Examples and Misclassified Examples.
The scond folder is about practice implementation of regression example with ANN on synthesis data- $$ y = \cos(2x_1) + cos(3x_1) $$ , and make same surface plot with this math equation(synthesis data) by using ANN.