Recognition of handwritten digits through the novel AI-IoT enabled approach

In this paper, the AI-IoT enabled approach is implemented from scratch without using any external machine learning libraries. We will create and train a simple neural network in python language. Neural networks basics are from the field of deep learning. The algorithm is inspired by the human brain. By using the AI and IoT enabled approach neural networks mimic the behavior of the human brain to handle the complicated data-driven difficulties. Neural networks take input data, prepare themselves to identify patterns within the data, and hence predict the outcome for newly provided data. The popular MNIST data set of handwritten digits is used to train and then test the network’s performance when introduced to real world problems. At each step, results are visualized using several plots and graphs using matplotlib library. Different activation functions are used, and the results are compared to obtain the best activation function for the given dataset.