Cancer Detection based on DNA Methylation by Ensemble of Extreme Learning Machines

کد مقاله : 1011-CFIS (R4)

نویسندگان

Hedieh Sajedi *، Mina Alizadeh دانشگاه تهران

چکیده مقاله

Extensive researches have been done with DNA methylation background. Researches have interests in DNA methylation analysis because of its correlation with diseases such as cancers. One of the big challenges of buckling up with DNA data is working with its high dimensions. In this paper ensemble of Extreme Learning Machines(ELM) is proposed for classification. ELM is a feed forward neural networks for classification of DNA methylation records. Finally we use k-means method for clustering the results of ELM classifiers, which is going to assign each case with the closest centroid that can be cancerous or healthy label. The highest accuracy that have been reported so far is about 95.59 % for PCA + SOM method and 96.32% for NMF + SOM method. In this paper we achieved the accuracy about 98.18% by the first proposed method and 100% by the second method.

کلیدواژه ها

DNA mehtylation, Beta-value, Classification, ELM, Accuracy

وضعیت: پذیرفته شده برای ارائه شفاهی