HANDWRITTEN NUMERIC RECOGNITION USING SUPPORT VECTOR MACHINE TECHNIQUE IN MACHINE LEARNING

Handwritten Numeral recognition plays a vital role in postal automation services. This is an important but very hard practical problem. Digit recognition is used in post offices, in banks for reading cheques, for license plate recognition, for street number recognition. The digit recognition can be divided into two groups, printed digit recognition and handwritten digit recognition. Recognition of printed digits is easier compared to the handwritten digit recognition. On the other hand, there are numerous handwriting styles for the same digit; hence more effort is required to find the accurate handwritten digit. In this project, we propose using SVM for recognition of handwritten digit. SVM is Machine Learning Technique. Support Vector Machine (SVM) is one of the most successful classifiers. Many applications use SVM for solving the classification problem, especially those for handwritten digit recognition. The SVM is used to improve classification accuracy. Our proposed algorithm will be tested on standard MNIST dataset for handwritten digit recognition. Total dataset size is of 70,000 datapoints. Among 70,000 datapoints, 60,000 datapoints are for train dataset and 10,000 datapoints are for test dataset. Each image size in 28*28 pixels.