PrathameshDeshpande/Mnist_kaggle

This is a deep learning model with accuracy of 99.5 for classifying hand-written number with softmax activation function

This was my first Kaggle competition. The data consists of train and test CSV files. We need to build a model to classify grayscale images using softmax activation function.I used DNN with 3 fully connected layers. Data augmentation was also applied in order to lower variance. Overall I got an accuracy of 99.45% with a rank under top 15% of. This was a fun model to start with.