A novel and powerful hybrid classifier method: Development and testing of heuristic k-nn algorithm with fuzzy distance metric

Kahraman, Hamdi Tolga

doi:10.1016/j.datak.2016.02.002

Cited by 21 publications

(7 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The method of the nearest neighbor has variations defined by the number of neighbors considered, for this reason, the number of neighbors is an evaluation parameter discussed in this paper. In the k-NN model, the k objects of the training set closest to x t are evaluated [37]. When k is greater than one, the neighboring k is obtained for each test point.…”

Section: Nearest Neighbors Methodsmentioning

confidence: 99%

Classification of Contaminated Insulators Using k-Nearest Neighbors Based on Computer Vision

Corso¹,

Perez²,

Stefenon³

et al. 2021

Preprint

View full text Add to dashboard Cite

The contamination on the insulators may increase its surface conductivity and, as a consequence, electrical discharges occur more frequently, which can lead to interruptions in the power supply. To maintain reliability in the electrical distribution power system, components that have lost their insulating properties must be replaced. Identifying the components that need maintenance, is a difficult task as there are several levels of contamination that are hardly noticed during inspections. To improve the quality of inspections, this paper proposes to use the k-nearest neighbours (k-NN) to classify the levels of insulator contamination, based on the image of insulators at various levels of contamination simulated in the laboratory. Using computer vision features such as mean, variance, asymmetry, kurtosis, energy, and entropy are used for training the k-NN. To assess the robustness of the proposed approach, statistical analysis and a comparative assessment with well-consolidated algorithms such as decision tree, ensemble subspace, and support vector machine models are presented. The k-NN showed results of up to 85.17 % accuracy using the k-fold cross-validation method, with an average accuracy higher than 82 % for multi-classification of the contamination of the insulators, being superior to the compared models.

show abstract

Section: Nearest Neighbors Methodsmentioning

confidence: 99%

Classification of Contaminated Insulators Using k-Nearest Neighbors Based on Computer Vision

Corso¹,

Perez²,

Stefenon³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Bunun yanında bulanık uzaklık metriği klasik metriklere kıyasla çok daha etkili sonuçlar vermektedir. Bulanık uzaklık metriğinin ortalama olarak k-nn algoritmasının sınıflandırma performansını klasik metriklere göre %10'dan fazla iyileştirdiği bilinmektedir (Kahraman, 2016;Kahraman vd., 2013;Yilmaz vd., 2018). Algoritma 1. k-nn Algoritmasının Sözde Kodu (Kahraman vd., 2013) (Pseudo Code of the k-nn Algorithm)…”

Section: K-en Yakın Komşu Algoritması (K-nearest Neigbour Algorithm)unclassified

“…MSA'lar sadece optimizasyon sürecinde arama algoritmaları olarak değil birçok çalışmada daha güçlü ve etkili melez yapay zekâ algoritmaları tasarlamak için de kullanılmaktadırlar. Bu amaçla tahmin (Kahraman vd., 2012), sınıflandırma (Kahraman, 2016;Kahraman vd., 2013;Yilmaz vd., 2018) ve kümeleme (Cerrada vd., 2019;Li vd., 2020;Yaday ve Prakash, 2020;Zhou vd., 2019;Aliarah vd., 2020) ve optimizasyon problemlerinin (Özkaraca ve Keçebaş., 2019;Özkaraca vd., 2018;Özkaraca vd. 2017) çözümlenmesi için geliştirilmiş MSA-tabanlı melez algoritmalar mevcuttur.…”

Section: Giriş (Introduction)unclassified

Development of Agde-Based Meta-Heuristic Dimension Reduction Algorithm for Classification Problems

Kahraman

Aras

YILDIZ

2020

Mühendislik Bilimleri Ve Tasarım Dergisi

View full text Add to dashboard Cite

Sınıflandırma problemleri yapay zekanın geniş bir uygulama alanını temsil etmektedirler. Sınıflandırma problemlerinin özelliklerine bağlı olarak algoritmaların performansları değişebilmektedir. Bir sınıflandırma probleminin boyutunun ya da nitelik sayısının değişmesi durumunda performansı bundan etkilenmeyen bir algoritma yoktur. Hemen tüm algoritmalar için problem boyutunun artması performansı olumsuz yönde etkileyen bir faktördür. Dolayısıyla sınıflandırma problemlerinde problem boyutu azaltma üzerinde yoğun bir şekilde çalışılan önemli bir konudur. Özellikle yüksek boyutlu nitelik uzaylarında problem için en uygun modeli yaratmak zor bir görevdir. Karmaşıklık düzeyi yüksek arama uzaylarında genel en iyi çözümü bulmak olanaksız olarak tanımlanmaktadır. Dolayısıyla bu tür arama problemleri için en ideal çözüm yolu meta-sezgisel arama algoritmalarından faydalanmaktır. Bu makale çalışmasında sınıflandırma problemlerinde boyut azaltmak için AGDE (adaptive guided differential evolution) algoritması temelli meta-sezgisel nitelik seçim yöntemi geliştirilmektedir. Geliştirilen yöntem niteliklerin ağırlıklandırılmasını ve eşik değerden küçük ağırlıklı niteliklerin tespit edilmesini sağlamaktadır. Deneysel çalışmalardan elde edilen sonuçlar AGDE-tabanlı nitelik seçim yönteminin sınıflandırma problemleri için etkili bir boyut azaltma aracı olduğunu göstermektedir.

show abstract

“…Accordingly, a chromosome representing a candidate for solution in the genetic algorithm has a structure consisting of eight genes, one for each dependent variable. Therefore, the mathematical expression of a population of solutions (population P consisting of m-number of solution candidates) to be created by the genetic algorithm for both cooling and heating load can then be represented as given in equation 1 [13,15]:…”

Section: Genetic K-nn Based Estimation and Weighting Modelmentioning

confidence: 99%

“…The estimation task is performed with k-nn [13], while the weighting task is performed with a genetic algorithm [14]. The weighting model represents the effect of the attributes of a problem on the target outputs with the optimal value in the range (0.1).…”

Section: Genetic K-nn Based Estimation and Weighting Modelmentioning

confidence: 99%

An Estimate of Energy Consumption for Housing Buildings in Hot Climatic Zones Through Artifical Intelligence Methods: Case of Antalya

Haydaraslan¹,

Haydaraslan²,

Kahraman³

et al. 2018

NWSA

View full text Add to dashboard Cite

Buildings use about one-third of total energy consumed in order to meet their heating and cooling needs. The building envelope that enables to protect it from physical factors in the outer environment is quite effective upon the amount of energy consumed. For the energy efficient solutions, it is necessary to enhance the heating and cooling performance of the building envelope. With this aim, in the study, the energy loads were calculated, which were consumed for heating and cooling by a building established as a reference through a simulation program in the province of Antalya, which respects a hot climatic zone, and the shifts in yearly heating and cooling loads of the alternative models were examined, which were developed by changing the thermal insulation thickness and the window-to-wall area ratio. In the study, the modern, effective artificial intelligence methods were used to enhance the energy performance of multi-dimensional buildings. Of the models for which heating and cooling load calculation had not been made before, the estimates for the thermal loads were made using an energy simulation program, and it has been reached that thermal insulation thickness and window-to-wall area ratio have effect on both loads.

show abstract

A novel and powerful hybrid classifier method: Development and testing of heuristic k-nn algorithm with fuzzy distance metric

Cited by 21 publications

References 31 publications

Classification of Contaminated Insulators Using k-Nearest Neighbors Based on Computer Vision

Classification of Contaminated Insulators Using k-Nearest Neighbors Based on Computer Vision

Development of Agde-Based Meta-Heuristic Dimension Reduction Algorithm for Classification Problems

An Estimate of Energy Consumption for Housing Buildings in Hot Climatic Zones Through Artifical Intelligence Methods: Case of Antalya

Contact Info

Product

Resources

About