Comparison of Different Classification Techniques Using Knowledge Discovery to Detect Malaria-infected Red Blood Cells

EasyChair Preprint 1657

Abstract

Malaria is an infectious disease which poses a major threat to the global health field. The objective of this research paper is to present an analysis  on the main machine-learning algorithms for the classification techniques used for malaria-infected red blood cells (MRBCs) and determine the best techniques by comparing classification accuracy. This study uses knowledge discovery to analyse the accuracy achieved by each algorithm and their ability to forecast and predict classification results. The system that demonstrates the computerised methods of image analysis generally involves three main phases. Firstly, data collection, pre-processing and feature extraction are conducted based on the characteristics of normal and MRBCs. Secondly, artificial neural network (ANN) and support vector machine (SVM) classification algorithm are used to classify the data set of 1,000 MRBCs. We use ten-fold cross-validation methods to estimate three prediction models to compare their performance fairly. Thirdly, we investigate the main parameters of numeric and graphical performance to evaluate the classification algorithms. The results indicate that ANN is the best predictor with 94% accuracy in the holdout sample.  Where, the model evaluated prediction with a 92.9% ability to distinguish positive and negative classification. Additionally, the model has high reliability at 93%.

Keyphrases: Machine Learning Algorithms, Malaria, knowledge discovery

@booklet{EasyChair:1657,
  author    = {Jameela Ali Alkrimi and Sherna Aziz Tome and Loay E. George and Azizah Bet. Suliman},
  title     = {Comparison of Different Classification Techniques Using Knowledge Discovery to Detect Malaria-infected Red Blood Cells},
  howpublished = {EasyChair Preprint 1657},
  year      = {EasyChair, 2019}}