Application Of Bt-SVM Multiclass Classification Algorithm In The Assessment Of Teaching

The quality of teaching is the core of school education, how to improve the teaching quality is the eternal theme for university educators. Since the establishment of five-year round assessment of teaching plan by Ministry Education, various colleges and universities try to find an effective teaching quality assurance system which can help them to do pre-assessment themselves.Support vector machine is one of the most widely used machine learning algorithms, which can efficiently perform statistical learning under limited samples. Considering the excellent learning performance of SVM and it is perfect for appraising system of teaching qualities, a modified binary tree SVM multi-classification algorithm has been proposed in this paper. The main work of this article in the following areas:1. The paper compares and reaearches the multiclass classification methods of SVM in many ways, and then a feasible optimization method was presented by analysis the disadvatages of the alorithms.2. In order to reduce adverse effect brought by the accumulation of error, this article refines the existing the multiclass classification algorithms of binary tree SVM by following the principle of the most easy to separate class first division. The modified algorithm is more scientific because it takes the distribution and center of samples.3. On the basis of researching the actual situation of the quality evaluation of teaching in domestic, this paper shows a new evaluation model of teaching which bases a modified binary tree SVM multi-class classification algorithms.