In the field of pattern recognition, researchers are trying to figure out how to make a machine that can accurately recognize and predict handwritten digits. The problem falls into the category of object detection and multi-class classification. Several machine learning (ML) algorithms have been used and optimized to achieve effective prediction results for digit recognition. A generic research question that comes up in this context is the usage of specific ML algorithms for performing this task. The purpose of this work is to build efficient deep learning (DL) algorithms to recognize digits and compare their performance with that of conventional ML algorithms. Two of the most common DL algorithms, convolutional neural network (CNN) and multilayer perceptron (MLP) or artificial neural network (ANN), are used here. A widely used conventional ML algorithm, Support Vector Machine (SVM), which usually provides robust performance in general classification tasks, is also used. The performance of these algorithms is compared and analyzed based on accuracy and test results. The MNIST dataset from Kaggle's Digit Recognizer competition is used here for training and testing the model. A graphical user interface (GUI) is constructed, in which the implemented ML model can be used to predict real-time user input of handwritten digits. |
*** Title, author list and abstract as seen in the Camera-Ready version of the paper that was provided to Conference Committee. Small changes that may have occurred during processing by Springer may not appear in this window.