The mechanism is difficult to be make out by conventional weather analysis and statistical forecast , and could n ' t find the correspond criterions of these paroxysmal disaster weather forming . these puzzled the forecasters , and leaded to the lower accuracy rate of forecast 传统的天气学和统计方法对这样的突发事件的发生发展机制难以确定，找不到这些突发性灾害天气形成的相应判据，令预报员十分困惑，至今高原地区大到暴雨的预报准确率都很低。
The latter classification method combined variable precision rough set model with k - nn classification method , so we can control the classification accuracy rate by the most endurable given error classification rate , then we can make the classification result conform to what we expect , and also some examples are given 后一种分类方法是将可变精度粗集模型与k - nn分类结合起来，从而可通过给定最大容忍的错误分类率来控制分类的准确度，使分类结果达到所期望的目的。并且给出了一些例子。
Many research topics had focused on how human can input information into a computer as quickly as possible with high accuracy rate in an easy way . the recognition of handwritten word is one possible solution . but the recognition of nature word writings encountered lots of problems , such as , low writing speed and low recognition rate 手写体识别的研究为解决这个问题提供了一个必要的条件，但是由于传统文字的特点，使得识别的正确率很难取得突破性的进展，而且传统文字记录的缓慢速度也制约了它作为高速录入的解决方案的可能。
Then , the paper present a protocol misusage ids in application layer based on markov chain , and obtain the following results after test : the system can have good effection on detecting the probe packet , dos attack and novelty attack . the system can overcome part of disadvantages in traditional intrusion detection system , and adapt requirements with different network . the experiment shows that the system can improve the detection accuracy rate , reduce the false negatives probability 通过对检测系统的测试分析,系统对扫描探测报文、 dos攻击报文和新的攻击报文等有较好的检测率。基于马尔可夫链建立的网络异常检测系统可以克服传统入侵检测系统的部分缺陷,能够适应不同网络环境的要求,同时也有效提高了系统的检测能力和检测效率,降低了误报率。
The convergence of the pnn decision rule for the bayesian decision rule is proved with probability . an explicit formulation of pnn generalization ability is proposed . and the accuracy rate of a classification network is explained as the maximum likelihood estimation of the generalization ability 证明了pnn的决策函数依概率收敛于贝叶斯决策函数;给出了pnn的推广能力表达式;证明了一个分类网络的测试集正确率是该网络推广能力的极大似然估计;给出了分类网络中需要的测试集数目表达式;证明了pnn推广能力不大于由贝叶斯决策所带来的正确识别率
