K-means clustering algorithm based on improved Cuckoo search algorithm and its application

Because the K-Means algorithm is easy to fall into the local optimum and the Cuckoo search (CS) algorithm is affected by the step size, this paper proposes a K-Means clustering algorithm based on improved cuckoo search (ICS-Kmeans). The algorithm is compared with the original K-means, the Kmeans alg...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:2018 IEEE 3rd International Conference on Big Data Analysis (ICBDA) s. 422 - 426
Hlavní autoři: Ye, Shuce, Huang, Xiaoli, Teng, Yinyin, Li, Yuxia
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: IEEE 01.03.2018
Témata:
Birds
Classification algorithms
Clustering algorithms
Convergence
cuckoo search
heart sound characteristics
Integrated circuits
K-Means
Optimization
Partitioning algorithms
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Because the K-Means algorithm is easy to fall into the local optimum and the Cuckoo search (CS) algorithm is affected by the step size, this paper proposes a K-Means clustering algorithm based on improved cuckoo search (ICS-Kmeans). The algorithm is compared with the original K-means, the Kmeans algorithm based on particle swarm optimization (PSO-Kmeans) and the K-Means algorithm based on the cuckoo search (CS-Kmeans). The experimental results show that the proposed algorithm can obtain better clustering effect, faster convergence rate and better accuracy rate through the experimental test on the UCI standard data set. The algorithm is also applied to the clustering of the characteristic parameters of the heart sound MFCC. The results show that a better clustering center can be obtained, the algorithm converges fast.
DOI:10.1109/ICBDA.2018.8367720