QUANTUM ANT LION OPTIMIZATION AND SUPPORT VECTOR MACHINE FOR THE FEATURE SELECTION AND GENE CLASSIFICATION

Gene selection for cancer prediction is a required model for the medical domain to effectively treat cancer patients. The presence of large information related to genes makes the existing model difficult to analyze the relationship between the features for gene classification. The existing models have the limitations of local optima trap, lower convergence and overfitting. To improve gene classification performance this research proposes the Quantum Ant Lion (QAL) optimization for feature selection. The Quantum search process is applied in the Ant Lion method to increase the search efficiency which helps to increase the exploration and overcome the local optima trap. The Archimedes spiral search is applied in the QAL method to increase exploitation in the feature selection based on the fitness function. The QAL method increases exploration and exploitation which helps to improve the convergence rate of the QAL method. The QAL method has 97.4 % accuracy and DNN-CNN model has 93.5 % accuracy for gene classification.