Handwriting-Recognition

This is a Neural Network model written in python for classification of handwritten digits.

Introduction

The data is from Mnist database. The MNIST database (Modified National Institute of Standards and Technology database) is a large collection of handwritten digits. This is a dataset of 60,000 28x28 grayscale images of the 10 digits, along with a test set of 10,000 images.

Overalls

This repository is application of classification task on Mnist dataset with several ML models like: Neural Network from scratch, ANN via keras, and Convolutional Neural Network. The tests demonstrate that the CNN model with LeNet architecture performs most efficiently on one channeled small images.