Comparison of Logistic regression, Convolution Neural Network, and Kernel Approaches for Classifying the Caenorhabditis Elegans Motion

Abstract

: Time series data are widely used in many fields including microbiology data. It is necessary to know how to classify the category to which observation belongs by using statistical classification methods and machine learning and deep learning algorithms. The study of the movement of some types of nematodes as one of the types of microorganisms including Caenorhabditis elegans (CE) is important to determine the actions and their impact on the life of the worms. In this study the CE motion time series data were represented by its wave motion angles which would be the study case. the non-linearity and uncertainty will be among the most common problems in this type of data that may lead to classifications that are not accurate. Convolutional Neural Network (CNN) will be used as one of the deep learning techniques and it is a non-linear method used to classify CE movement as a dependent variable in binary cases based on images of wave motion angles as an independent variable and its use will lead to accurate results because it is a suitable non-linear method to deal with Study data to solve nonlinearity and uncertainty problems through digital data visualization. Logistic regression (LR) and kernel method were also used to classify CE angles of movement. The AR(p) rank was used to determine the structure of the used methods. And by comparing the results between the methods used, it was found that the CNN method is superior to the other methods used. Therefore, it is possible to conclude that the use of the CNN method, which is based on pictorial classification, leads to accurate classification results compared to other methods based on numerical classification.