Analysis of Multiobjective Algorithms for the Classification of Multi-Label Video Datasets

Gizem Nur Karagoz,Adnan Yazici,Ahmet Cosar,Tansel Dokeroglu

doi:10.1109/access.2020.3022317

Gizem Nur Karagoz, Adnan Yazici + Show 2 more

Open Access

https://doi.org/10.1109/access.2020.3022317

Copy DOI

Abstract

It is of great importance to extract and validate an optimal subset of non-dominated features for effective multi-label classification. However, deciding on the best subset of features is an NP-Hard problem and plays a key role in improving the prediction accuracy and the processing time of video datasets. In this study, we propose autoencoders for dimensionality reduction of video data sets and ensemble the features extracted by the multi-objective evolutionary Non-dominated Sorting Genetic Algorithm and the autoencoder. We explore the performance of well-known multi-label classification algorithms for video datasets in terms of prediction accuracy and the number of features used. More specifically, we evaluate Non-dominated Sorting Genetic Algorithm-II, autoencoders, ensemble learning algorithms, Principal Component Analysis, Information Gain, and Correlation Based Feature Selection. Some of these algorithms use feature selection techniques to improve the accuracy of the classification. Experiments are carried out with local feature descriptors extracted from two multi-label datasets, the MIR-Flickr dataset which consists of images and the Wireless Multimedia Sensor dataset that we have generated from our video recordings. Significant improvements in the accuracy performance of the algorithms are observed while the number of features is being reduced.

Highlights

Multi-label classification has been applied to many problems in various fields of application, including the diagnosis of diseases based on many signs and symptoms [1] and used in many tools developed for the classification of social media resources, images, bioinformatics [2], videos [3], patient classification [4], text [5], and audio that may need to be assigned with more than one label [6]
We analyze the performance of multi-label video data classification algorithms through feature selection techniques
The multi-objective evolutionary NSGA-II is used for the feature selection process and autoencoders with regularizations as denoising autoencoder and drop-out regularization are implemented

Summary

INTRODUCTION

Multi-label classification has been applied to many problems in various fields of application, including the diagnosis of diseases based on many signs and symptoms [1] and used in many tools developed for the classification of social media resources, images, bioinformatics [2], videos [3], patient classification [4], text [5], and audio that may need to be assigned with more than one label [6]. Karagoz et al.: Analysis of Multiobjective Algorithms for the Classification of Multi-Label Video Datasets data processing. Embedded methods combine feature selection methods with a model construction process (wrapper), so that they have an ability to stop the attribute filtering process when the performance achieved by the classification/learning algorithm reaches a sufficient level [10]. Most descriptive features that are selected by two different methods are combined and ensemble feature selection results are achieved with NSGA-II and multi-label classification algorithms. We analyze the performance of multi-label classification algorithms, Non-dominated Sorting Genetic Algorithm (NSGA-II) [11], autoencoders, ensemble learning algorithms, Principal Component Analysis (PCA), Information Gain (IG), and Correlation Based Feature Selection (CBFS). A parallel multi-objective NSGA-II algorithm is used to select the best subset of features and the resulting set is combined with the feature set of the autoencoder. Our final remarks and future studies are presented in the last section

RELATED WORK

PERFORMANCE EVALUATION OF THE ALGORITHMS

Findings

CONCLUSION