LVQ神经网络在高校师职称评审中的应用 英文版.doc

Professional Titles Evaluation Forecasting Based on LVQ Neural Network Zhang Ji-Gang Liang Na Department of Mathematics,HuBei University Of Science And Technology, HuBei Xianning ,437100 ,china) E-mail:liangna7845@126.com Abstract：In this paper, associate professor promotion in Institution in 2012 was selected as the study’s samples and evaluation Forecasting model based on LVQ neural network was set up. After being trained and tested, the model has the excellent function of classifying. The results show that LVQ could be employed in the pattern-identification, while the model based on LVQ neural network could predict efficiently. Key words: professional titles evaluation; LVQ network Ⅰ.INTRONDUCTION University teachers title appraisal is mainly based on qualitative analysis presently in our country, beacause of complex actual operation and inevitable human subjectivity,the result was not accurate enough. the important problem universities facing is how to establish objective and highly effective prediction model to forecast the teachers title appraisal accurately. Because the problems of review prediction for teachers professional titles in Universities are complicated, moreover, the various indexes influence each other, so it presents a complicated nonlinear relationship. Based on the above difficulties, we introduce the neural network as a powerful tool to handle this kind of nonlinear problem of the evaluation prediction in recent years. BP network is a kind of network which has the most common application, but its weakness is adopting a nonlinear optimization strategy which is based on the gradient descent, so it is possible to fall into the local minimum, then can’t guarantee to find the global minimum. Some other optimization strategies such as genetic algorithm simulated annealing algorithm etc, although we can obtain the global minimum, it can lead to low efficiency at times for large amount of calculation. Fortunately, the advantages of LVQ neural ne