Clemen Zineng Deng
Age: 18
Portland, OR
There are many individual machine-learning methods used for breast cancer detection. Each has its strengths and weaknesses. However, none of them can predict satisfactory results for all database sizes. In Clemen's research, a novel machine learning method called Weighted Hierarchical Adaptive Voting Ensemble method, or the WHAVE method, was developed and implemented for breast cancer detection. This WHAVE method taught the computer to accurately predict breast cancer for all database sizes. WHAVE uses a novel mathematical weighting formula to combine individual machine learning methods and gives more weight to the more accurate individual methods. It also uses a hierarchical ensemble algorithm to combine individual machine learning methods and further improves accuracy and efficiency. It can adaptively search for the optimal weights and ensemble hierarchy to produce the highest accuracy for any breast cancer database sizes. Test results show WHAVE produces better accuracy for breast cancer detection for all database sizes than any existing individual machine learning methods. WHAVE achieves an accuracy of 99.8% for breast cancer detection, the highest ever reported in literacy.