Non-Linear Machine Learning Techniques for Multi-Label Image Data Classification

Authors

D. Senthilkumar, Department of Computer Science and Engineering, University College of Engineering (BIT Campus), Anna University, Tiruchirappalli, Tamil Nadu, IndiaFollow
A. K. Reshmy, Department of Computer Science and Engineering, Saveetha School of Engineering, SIMATS Tamil Nadu, IndiaFollow
M. G. Kavitha, Department of Computer Science and Engineering, University College of Engineering, Pattukkottai, Tamil Nadu, IndiaFollow

Author Country (or Countries)

India

Abstract

In this paper, we propose non-linear Machine Learning Techniques (MLT) for Multi-label Image Classification (MLIC) problems. Multi-label Learning requires MLT to identify the complex non-linear relationship between the features and class labels. Also, Multi-label data degrades the performance of the classifiers and processing of this data with a large number of features is too complex while using traditional methods. Therefore, we propose two approaches namely ensemble Deep Learning Network (DLN) and Multivariate Adaptive Regression Splines (MARS) for MLIC. The experimental results show that the proposed (DLN and MARS) algorithms achieves a superior predictive performance rate of 94.77% and 81.68% respectively, compared to the existing methods.

Digital Object Identifier (DOI)

http://dx.doi.org/10.18576/amis/120608

Recommended Citation

Senthilkumar, D.; K. Reshmy, A.; and G. Kavitha, M. (2018) "Non-Linear Machine Learning Techniques for Multi-Label Image Data Classification," Applied Mathematics & Information Sciences: Vol. 12: Iss. 6, Article 8.
DOI: http://dx.doi.org/10.18576/amis/120608
Available at: https://digitalcommons.aaru.edu.jo/amis/vol12/iss6/8