| Summary: | 碩士 === 國立中興大學 === 資訊管理學系所 === 103 === Classification problem is one of the applications of machine learning. Because of the numbers and dimensions of data increasing, the performance of classification algorithm is not well. Feature selection can not only remove those features which are not corresponding to the classification but find the best feature subset which can let classifier perform much better. In this study, we use bio-inspired algorithms to do feature selection, and we use SVM as a classifier.
In this study, we use lion’s algorithm, a novel bio-inspired algorithm developed in 2012, as our algorithm to do feature selection. In our lab, there was a modified version developed before. In this study, a new modified version based on the past paper in our lab will be introduced. Moreover, the mix version of two versions above will be developed. In the experiment part, we will compare the performance of those different modified versions of lion’s algorithm.
|
|---|
