Integration of the dendritic cell algorithm with K-means clustering

Mohamad Farhan Mohamad Mohsin, Azuraliza Abu Bakar, Abdul Razak Hamdan

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

The dendritic cell algorithm is an effective technique to detect anomalies in time series applications. However, the algorithm is less effective when it mines a general classification dataset because the items are not organized in an orderly event-driven manner. Ideally, for they need to be arranged in sequence by sorting them according to decision class. However, it is not practicable to apply this step because the decision classes for real datasets is unknown. Therefore, an integrated model that combines the dendritic cell algorithm and the k-means algorithm is proposed as an alternative to the existing sorting function based on decision class. The proposed model is evaluated by applying it to eight universal classification datasets and assessing its performance according to four evaluation metrics: detection rate, specificity, false detection rate, and accuracy. The results show that the proposed clustered dendritic cell algorithm is more effective than the non-clustered version. When applied to abenchmark dataset, the clustered dendritic cell algorithm demonstrates significant improvement in performance on the unordered version of the dataset and generates a comparable result to that of its competitor. For the other seven datasets, the proposed algorithm generates better specificity, false detection rate, and accuracy. The findings indicate that item-centroid distance within a cluster can be adopted to transform an unordered dataset into a sequential dataset, thus fulfilling the dendritic cell algorithm requirement for ordered data.

Original languageEnglish
Pages (from-to)60-77
Number of pages18
JournalInternational Journal of Advances in Soft Computing and its Applications
Volume6
Issue number3
Publication statusPublished - 2014

Fingerprint

Sorting
Dendritic Cells
Time series

Keywords

  • Artificial immune system
  • Clustering
  • Dendritic cell algorithm
  • Kmeans

ASJC Scopus subject areas

  • Computer Science Applications

Cite this

Integration of the dendritic cell algorithm with K-means clustering. / Mohamad Mohsin, Mohamad Farhan; Abu Bakar, Azuraliza; Hamdan, Abdul Razak.

In: International Journal of Advances in Soft Computing and its Applications, Vol. 6, No. 3, 2014, p. 60-77.

Research output: Contribution to journalArticle

@article{1ec1ae489867404e9afe4926deee7272,
title = "Integration of the dendritic cell algorithm with K-means clustering",
abstract = "The dendritic cell algorithm is an effective technique to detect anomalies in time series applications. However, the algorithm is less effective when it mines a general classification dataset because the items are not organized in an orderly event-driven manner. Ideally, for they need to be arranged in sequence by sorting them according to decision class. However, it is not practicable to apply this step because the decision classes for real datasets is unknown. Therefore, an integrated model that combines the dendritic cell algorithm and the k-means algorithm is proposed as an alternative to the existing sorting function based on decision class. The proposed model is evaluated by applying it to eight universal classification datasets and assessing its performance according to four evaluation metrics: detection rate, specificity, false detection rate, and accuracy. The results show that the proposed clustered dendritic cell algorithm is more effective than the non-clustered version. When applied to abenchmark dataset, the clustered dendritic cell algorithm demonstrates significant improvement in performance on the unordered version of the dataset and generates a comparable result to that of its competitor. For the other seven datasets, the proposed algorithm generates better specificity, false detection rate, and accuracy. The findings indicate that item-centroid distance within a cluster can be adopted to transform an unordered dataset into a sequential dataset, thus fulfilling the dendritic cell algorithm requirement for ordered data.",
keywords = "Artificial immune system, Clustering, Dendritic cell algorithm, Kmeans",
author = "{Mohamad Mohsin}, {Mohamad Farhan} and {Abu Bakar}, Azuraliza and Hamdan, {Abdul Razak}",
year = "2014",
language = "English",
volume = "6",
pages = "60--77",
journal = "International Journal of Advances in Soft Computing and its Applications",
issn = "2074-8523",
publisher = "International Center for Scientific Research and Studies (ICSRS)",
number = "3",

}

TY - JOUR

T1 - Integration of the dendritic cell algorithm with K-means clustering

AU - Mohamad Mohsin, Mohamad Farhan

AU - Abu Bakar, Azuraliza

AU - Hamdan, Abdul Razak

PY - 2014

Y1 - 2014

N2 - The dendritic cell algorithm is an effective technique to detect anomalies in time series applications. However, the algorithm is less effective when it mines a general classification dataset because the items are not organized in an orderly event-driven manner. Ideally, for they need to be arranged in sequence by sorting them according to decision class. However, it is not practicable to apply this step because the decision classes for real datasets is unknown. Therefore, an integrated model that combines the dendritic cell algorithm and the k-means algorithm is proposed as an alternative to the existing sorting function based on decision class. The proposed model is evaluated by applying it to eight universal classification datasets and assessing its performance according to four evaluation metrics: detection rate, specificity, false detection rate, and accuracy. The results show that the proposed clustered dendritic cell algorithm is more effective than the non-clustered version. When applied to abenchmark dataset, the clustered dendritic cell algorithm demonstrates significant improvement in performance on the unordered version of the dataset and generates a comparable result to that of its competitor. For the other seven datasets, the proposed algorithm generates better specificity, false detection rate, and accuracy. The findings indicate that item-centroid distance within a cluster can be adopted to transform an unordered dataset into a sequential dataset, thus fulfilling the dendritic cell algorithm requirement for ordered data.

AB - The dendritic cell algorithm is an effective technique to detect anomalies in time series applications. However, the algorithm is less effective when it mines a general classification dataset because the items are not organized in an orderly event-driven manner. Ideally, for they need to be arranged in sequence by sorting them according to decision class. However, it is not practicable to apply this step because the decision classes for real datasets is unknown. Therefore, an integrated model that combines the dendritic cell algorithm and the k-means algorithm is proposed as an alternative to the existing sorting function based on decision class. The proposed model is evaluated by applying it to eight universal classification datasets and assessing its performance according to four evaluation metrics: detection rate, specificity, false detection rate, and accuracy. The results show that the proposed clustered dendritic cell algorithm is more effective than the non-clustered version. When applied to abenchmark dataset, the clustered dendritic cell algorithm demonstrates significant improvement in performance on the unordered version of the dataset and generates a comparable result to that of its competitor. For the other seven datasets, the proposed algorithm generates better specificity, false detection rate, and accuracy. The findings indicate that item-centroid distance within a cluster can be adopted to transform an unordered dataset into a sequential dataset, thus fulfilling the dendritic cell algorithm requirement for ordered data.

KW - Artificial immune system

KW - Clustering

KW - Dendritic cell algorithm

KW - Kmeans

UR - http://www.scopus.com/inward/record.url?scp=84922843879&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84922843879&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:84922843879

VL - 6

SP - 60

EP - 77

JO - International Journal of Advances in Soft Computing and its Applications

JF - International Journal of Advances in Soft Computing and its Applications

SN - 2074-8523

IS - 3

ER -