vioft2nntf2t|tblJournal|Abstract_paper|0xf4ff9b5e27000000886e010001000300 Different clustering algorithms produce distinct sub-divisions as they apply disparate partition on the data. Hence, no single clustering algorithm is said to be optimal and therefore resulting in different partitions. To utilize the complementary nature of different partitions, ensemble clustering is used. The work in this paper focuses on producing ensembles through several clustering algorithms that perform feature extraction using deep learning and malicious tumor detection through ensemble cluster. In this study, to improve the performance and reduce the complexity involved in the malicious tumor detection process, Deep Learning Feature Extraction (DLFE) technique is presented. Furthermore, to improve the quality of results obtained, ensemble clusters namely, Normalized Spectral Cluster and Gaussian Mixture technique has been applied to the extracted features. The experimental results of the proposed technique have been evaluated and validated for performance and quality analysis on three datasets based on accuracy, sensitivity, specificity. The experimental results achieved 85.28% accuracy, 70.43% specificity, and 97.19% sensitivity, demonstrating the effectiveness of the proposed technique for identifying normal and abnormal tissues from various test images. The simulation results prove the significance in terms of quality parameters and accuracy in comparison to the state-of-the-art techniques.
S. Subash Chandra Bose1, T. Christopher2
Government Arts College, Udumalpet, India1, Government Arts College, Coimbatore, India2
Clustering Algorithm, Deep Learning, Feature Extraction, Normalized Spectral Cluster, Gaussian Mixture
| January | February | March | April | May | June | July | August | September | October | November | December |
| 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Published By :
ICTACT
Published In :
ICTACT Journal on Soft Computing
( Volume: 8 , Issue: 4 , Pages: 1750-1757 )
Date of Publication :
July 2018
Page Views :
186
Full Text Views :
2
|