Feature selection based on buzzard optimization algorithm for potato surface defects detection
- PDF / 868,765 Bytes
- 19 Pages / 439.37 x 666.142 pts Page_size
- 61 Downloads / 197 Views
Feature selection based on buzzard optimization algorithm for potato surface defects detection Ali Arshaghi 1 & Mohsen Ashourian 2
& Leila Ghabeli
1
Received: 15 January 2020 / Revised: 12 June 2020 / Accepted: 24 June 2020 # Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract
Different methods of feature selection find the best subdivision from the candidate subset. In all methods, based on the application and the type of the definition, a subset is selected as the answer; which can optimize the value of an evaluation function. The large number of features, high spatial and temporal complexity, and even reduced accuracy are common problems in such systems. Therefore, research needs to be performed to optimize these systems. In this paper, for increasing the classification accuracy and reducing their complexity; feature selection techniques are used. In addition, a new feature selection method by using the buzzard optimization algorithm (BUOZA) is proposed. These features would be used in segmentation, feature extraction, and classification steps in related applications; to improve the system performance. The results of the performed experiment on the developed method have shown a high performance while optimizing the system’s working parameters. Keywords Buzzard optimization algorithm . Global optimization . Potato defect detection . Feature selection . Image processing
1 Introduction The feature selection is one of the major issues in machine learning and statistical pattern recognition. In various applications including classification, there are many features that are
* Mohsen Ashourian [email protected] Ali Arshaghi [email protected] Leila Ghabeli [email protected]
1
Department of Electrical Engineering, Islamic Azad University, Central Tehran Branch, Tehran, Iran
2
Department of Electrical Engineering, Islamic Azad University, Majlesi Branch, Isfahan, Iran
Multimedia Tools and Applications
not important or have a low amount of useful data. Therefore, removing information about such unnecessary features would not cause any problem in analyzing data. The unnecessary features increases the computational load and memory for the intended application [14, 24]. One of the main problems in feature selection algorithms is their high computational complexity [18]. Therefore, research has been performed for generating faster algorithms. Different methods of feature selection have been applied to discover the best subset from the candidate subset; in all methods, based on the application and the type of the definition, a subset is selected as the answer which can optimize the value of an evaluation function [16]. In general, based on the type of search and analyses, feature selection methods could be divided into different categories [28]. In some solutions, all possible space is searched (for example in image security [22, 23]). In other solutions, which could be discoverable or random search, the search space becomes smaller if accepting lower efficiency. Various opti
Data Loading...
