A Novel Approach to Classification of Gene Expression Datasets Using Computational Intelligence Techniques

摘要

The main focus of the proposed system is to handle gene expression datasets of the cancer patients. As per the demography study, the maximum females case suffering with cancer which could be traced by the genes available in their blood, samples. An algorithm is proposed in such a way that which infers in directly predicting the probe set value which indicates at which gene levels, the human is suffering from the research entitled above. The proposed system obtained good result when there is relevant data in the knowledge base, but if failed when there is out loss in data for avoiding this problem, the proposed algorithm is a modified and extended approach of Particle Swarm Optimization (PSO) to find out the exact optimistic gene from which the patient is been lead to cancer at particular levels. An introduced concept namely SIFTS parameter where identification of gene levels are traced at 65-95%. The subsequent outputs which are obtained are giving better results compared with previous research. For this analysis we had considered Poor Differentiated normal cancer (PD) from CPDR datasets supported by IRC and authorized WAMPR, US. The data is consistent and provided better results.