Parameter estimation of support vector machine with radial basis function kernel using grid search with leave-p-out cross validation for classification of motion patterns of subviral particles

Ricardo M Schuhmann,Thomas Schanze,Andreas Rausch

doi:10.1515/cdbme-2021-2031

Ricardo M Schuhmann, Thomas Schanze + Show 1 more

Open Access

https://doi.org/10.1515/cdbme-2021-2031

Copy DOI

Abstract

Abstract The classification of subviral particle motion in fluorescence microscopy video sequences is relevant to drug development. This work introduces a method for estimating parameters for support vector machines (SVMs) with radial basis function (RBF) kernels using grid search with leave-pout cross-validation for classification of subviral particle motion patterns. RBF-SVM was trained and tested with a large number of combinations of expert-evaluated training and test data sets for different RBF-SVM parameters using grid search. For each subtest, the mean and standard deviation of the accuracy of the RBF-SVM were calculated. The RBF-SVM parameters are selected according to the optimal accuracy. For the optimal parameters, the accuracy is 89% +- 13% for N = 100. Using the introduced computer intensive machine learning parameter adjustment method, an RBF-SVM has been successfully trained to classify the motion patterns of subviral particles into chaotic, moderate and linear movements.

Highlights

Despite many positive aspects, advancing globalization holds risks
Parameter estimation of support vector machine with radial basis function kernel using grid search with leave-p-out cross validation for classification of motion patterns of subviral particles estimated parameters shown in a previous publication [5]
Resulting from the radial basis function (RBF)-support vector machines (SVMs), the support vectors are denoted by circles and the decision functions are shown as dashed lines

Summary

Introduction

Despite many positive aspects, advancing globalization holds risks. The current Covid-19 pandemic demonstrates the risk of fast globally spreading viral diseases. All the more reason for virus research to come to the fore Another thread to the human population are pathogens of hemorrhagic fevers, like the Ebola virus that has a mortality rate of about 50% [1]. To advance the development of new drugs, a profound knowledge of the pathogens' subviral particles is mandatory. The cells analyzed for this work were infected with the Marburg virus, whose properties are similar compared to those of the Ebola virus. For this purpose, image sequences have been provided by scientists from the Institute of Virology at the Philipps-University, Marburg, which originate from an innovative fluorescence microscopic livecell imaging method to visualize intracellular processes [2]. Algorithms to automatically track subviral particles in fluorescence images were introduced by Kienzle et al [3] and further developed by Rausch et al [4]

Methods

Results

Conclusion