Handwritten Digit Recognition using Convolution Neural Networks
Shailesh Rajput, Yoonsuk Choi · 2022 IEEE 12th Annual Computing and Communication Workshop and Conference (CCWC) · 2022
Deep learning has witnessed a significant evolution recently with growth in high-performance devices and research in the neural network. The central aspect of this paper is to discuss the deep learning concept ideas and problems faced during training the model and come with a solution for better accuracy, illustrated by digit recognition and prediction using a convolution neural network. The first part of the research focused on the techniques used for training models in neural networks and enlightened the previous study in digit recognition. The second half of the paper detailed the concept of the proposed model for digit recognition with increased accuracy by preprocessing samples and ensemble of two or more models with different architectures and preprocessing. Preprocessing of samples and implementation of the model architecture is done using python.