Performance Analysis of Machine Learning Techniques for Microscopic Bacteria Image Classification

Abstract

Microscopic bacterial image classification is of great significance in medical science in order to diagnosis numerous fatal diseases caused by bacteria. But, professional individuals are required for manual identification process of bacteria which is time consuming and laborious. However, machine learning techniques provide a major breakthrough in identifying bacteria automatically with high accuracy and precision. Thus, this paper is presented three hybrid approach that are convolutional neural network with support vector machine (CNN-SVM), convolutional neural network with K-Nearest Neighbors (CNN-KNN) and convolutional neural network with Naïve Bayes (CNN-Naïve Bayes) for automatic bacteria identification from microscopic image. Experimental results shows among the three hybrid model, CNN-SVM achieved 98.7% accuracy which is higher compared to the other machine learning approach.