混合模式识别系统研究

讨论基于多种分类方法的模块组合实现的混合模式识别系统，它不同于利用多分类器输出结果表决的集成系统．提出两个系统：一个面向印刷体汉字文本识别，另一个面向自由手写体数字识别．利用多种特征和多种分类方法的组合、部分识别信息控制混淆字判别策略以及提出的动态模板库匹配后处理方法，使系统的性能与传统单一分类器系统比较，获得明显改善．实验表明：多方法多策略混合是解决复杂和增强系统鲁棒性的一条途径

从第四轮“关于科研单位的组织与业绩的国际比较研究”（ICSOPRU）看满意度对科研单位业绩的影响

What role satisfaction plays in the factors contributing to performance continues to be a major area of interest in the study of industrial and organization psychology, but there is a lack of quantative study dealing with this question in research units. The author has a try in this paper to answer this question using the data from China, Ghana, Hungary and Mexico of the Fourth Round International Comparative Study on the Organization and Performance of Research Units (ICSOPRU). The data-analysis include the principle component factor analysis of the performance and the satisfaction items in the Fourth Round ICSOPRU Questionnaires, and the multiple classification analysis, the multivariate nominal analysis of the performance and the satisfaction factors. The main findings show that a certain facet of the satisfaction explains the largest proportion of variances of a certain dimention of the performance and has a higher relative importance in contributing to the understanding of the performance. There also a comparison between the results from the four countries and that from China.