We found a match
Your institution may have rights to this item. Sign in to continue.
- Title
SINIFLANDIRMA PROBLEMLERİ İÇİN AGDE-TABANLI META-SEZGİSEL BOYUT İNDİRGEME ALGORİTMASININ GELİŞTİRİLMESİ.
- Authors
KAHRAMAN, Hamdi Tolga; ARAS, Büşra; YILDIZ, Orhun
- Abstract
Classification problems represent a wide range of applications of artificial intelligence. Depending on the properties of classification problems, algorithms' performances may vary. If the number of attributes/features of a problem changes, the performance of the algorithm used to solve this problem also changes. For almost all algorithms, increasing problem dimension is a factor that negatively affects performance. Therefore, dimensionality reduction is an important issue that is studied extensively. It is a difficult task to create the most suitable model for the problem, especially in high-dimensional search spaces (with a high number of independent variables). It is impossible to find the best overall solution in highly complex search spaces. Therefore, the ideal solution for such search problems is to use meta-heuristic search algorithms. In this article, meta-heuristic feature selection method based on AGDE (adaptive guided differential evolution) algorithm is developed for dimensionality reduction in classification problems. The developed method enables the weighting of the attributes and the determination of the weighted features less than the threshold value. The results obtained from the experimental studies show that AGDE-based dimensionality reduction method has very effective performance for classification problems.
- Subjects
DIFFERENTIAL evolution; ARTIFICIAL intelligence; PROBLEM solving; SEARCH algorithms; FEATURE selection
- Publication
SDU Journal of Engineering Sciences & Design / Mühendislik Bilimleri ve Tasarım Dergisi, 2020, Vol 8, Issue 5, p206
- ISSN
1308-6693
- Publication type
Article
- DOI
10.21923/jesd.828518