Handwritten Digit Recognition using SVM Algorithms to Improve the Accuracy Compared with Gaussian Naive Bayes.

Abstract

Aim: The main objective of this paper is to detect handwritten digits with the help of Machine

Learning algorithms such as Novel Support Vector Machine and Gaussian Naive Bayes algorithms. Materials and Methods: The datasets were extracted from the SKLEARN module of python which has around 70000 sample examples to solve the detection of handwritten digits. Novel Support Vector Machine predicts the output for dependent variable and independent variable. Sample count for group 1 Novel Support Vector Machine is 20 and sample count for group 2 Gaussian Naive Bayes is 20. Total sample size count is 20 for both groups using Gpower as 80%. Results: Novel Support Vector Machine (SVM) comes up with mean accuracy when contrasted with the Gaussian Naive Bayes (GNB) algorithm. Ultimately the Novel Support vector machine pops up with a better accuracy rate when compared with the Gaussian Naive Bayes algorithm.The two algorithms SVM and Gaussian Naive Bayes are statistically satisfied with the independent sample T-Test value (p=0.001) with a confidence level of 95%. Image Processing definitely happened to convert the handwritten digit into digital image of 8 * 8 pixel size. In the Image Processing, Onces the Image is converted into digital format then the digital format image is detected using Image Detection.  Conclusion: Within the limits of the study the Support Vector Machine algorithm has better accuracy compared with the Gaussian Naive Bayes algorithm.