AN EFFICIENT MODEL FOR CLASSIFICATION OF CANCER TYPES ON GENE EXPRESSION DATA
European Journal of Molecular & Clinical Medicine,
2020, Volume 7, Issue 4, Pages 1269-1285
AbstractIn recent times, cancer subtype classification has been regarded for its capability of considerably advancing the disease prognosis and progressing the customized administration of patients. Cancer subtypes identification is considered to be the most complicated process, as it faces the challenge of inadequate methods with regards to the accurate determination of gene expressions. Previously, the classification process of cancer subtypes preferred a Deep Flexible Neural Forest (DFN Forest) strategy which is an incorporation of Flexible Neural Tree (FNT) approach. However, it possesses the minimum accuracy due to the inability of choosing the appropriate features and consumption of extended time for classification process. This problem has surpassed by proposing the approach that has developed as Artificial Bee Colony (ABC) with Deep Fuzzy Flexible Neural Forest (DFFN Forest) approach. The proposed research intended to measure the cancer subtypes through introducing novel strategies of feature selection and classifier, as these two processes play a significant role during the cancer diagnosis. During the feature selection process, the Artificial Bee Colony (ABC) algorithm has been used to diminish the miss rate of the classifier, besides the chosen feature has performed in the Deep Fuzzy Flexible Neural Forest (DFFN Forest), in which the Fuzzy function has presented, concerning the advancement of DFN Forest classifier’s outputs. The DFFN Forest method included the fuzzy logic for updating the classifier’s weight values during the prediction of cancer subtype. The evaluation of proposed algorithm has processed by considering the factors, such as accuracy, precision, recall, f-measure, and error ratio of the classifier.
- Article View: 155
- PDF Download: 245