分类机

This paper brings forward a new detection of steganography method based on genetic algorithms and one-class SVM.

针对二类支持向量机分类器在隐秘图像检测中训练步骤复杂与推广性弱的缺点，提出了一种新的基于遗传算法和一类支持向量机的隐秘图像检测方案。

A weighted posterior probability method is presented to calculate the probability outputs of support vector machinesfor multi class cases.

中文摘要：为解决传统支持向量机方法不提供概率输出的问题，在支持向量机多类分类问题输出概率建模中，提出了加权后验概率建模方法。

This paper expoles an algorithm about dearing surface defects detection by Support Vector Machines and principal component analysis that is the new branch of machine learning, in which the defective area and non-defective area are treated as two different textures and are sampled respectively to be learned, in order to reduce dimension, the image data can be processed by PCA.

提出了一种基于支持支持向量机和主成分分析的轴承表面缺陷检测算法，该算法把轴承中的非缺陷区域和缺陷区域分别看作两种不同的纹理模式，先利用主成分分析法对图像进行降维处理，然后用支持向量机方法对降维后的样本采样学习，然后进行分类判断。

When processing the restrained conditions, we convert the restrained conditions into punish function and add it to fitness function which profits from the conventional method. We introduce the support vector machine to the selection operator, which increases the multiplicity of selection and increases the probability of the algorithm getting the overall optimal solution. Also, we use the adaptive probabilities of crossover and mutation in mixed GAs which enhances the efficiency of the algorithm. The result of tests about the system indicates that the improved genetic algorithm has more high convergence speed and better solution quality.

在生成初始种群时，不是简单的随机生成，而是用支持向量机对历史数据进行分类，用较优的一类历史数据作为初始种群；在约束条件的处理上，本文借鉴常规方法中的罚函数法，采用将约束条件作为罚函数包含到适应度评价中去的方法；在选择算子上，本文引入了支持向量机的方法，增加了选择的多样性，增加算法搜索到全局最优解的概率；同时针对交叉概率和变异概率在算法的前后期不同的情况，本文引入了Srinvivas等提出的动态自适应的方法，使得交叉概率和变异概率能够随着算法中个体的质量动态改变，提高算法的效率。

Cooperating the advantages of SVM solve classification problems about small sample set, non-linear and multi-dimension model identification, I use SVM to train the sample set of different frequency signal and look for a super plane to divide the data into two parts. This method gets rid of complex computation which is a flaw of traditional decoding method.This paper introduces the theory and algorithm of SVM at first. It proposes to use the SVM to decode FSK signal based on the classifiable character of SVM.

结合支持向量机在解决小样本、非线性及高维模式识别问题中表现出的许多特有优势，运用支持向量机对不同频率信号的数据样本集进行训练，通过寻找最优超平面来对不同频率信号的数据进行快速分类，克服传统频移键控解码算法中需要进行大量运算的缺点，从而实现频移键控的解码。

In the third part of this paper, it introduces detailedly the hardware and software information to you. It focuses on the hardware which relative to DSP in hardware introduction. In software introduction, relative time decision method and C language program approach of decision-function are the emphases. In the last of this part, it gives a method that uses decoded data to configurate right calling numbers based on the deeply research on Calling Information Data protocol.

本文第三部分详细介绍应用支持向量机进行解码的实际系统硬件和软件设计，其中硬件部分重点介绍了与DSP处理信号相关的硬件原理，软件部分则重点介绍正弦信号采样点相对时间确定的方法和支持向量机分类决策函数的C语言实现方法；在深入分析CID信号协议的基础上，按照解码系统的要求，给出了如何将解码得到的二进制数据串恢复成调制之前信息的方法。

And the essay particular dissertate structure and calculation principle of support vector machines.

本文介绍了支持向量机方法和模糊积分，论述了支持向量机分类方法的结构及算法原理。

As a leading Japanese Pachinko, a metal pin ball, gaming machines recycling company with plants in China and Hong Kong, this Company focus on importing Pachinko machines to Hong Kong for disassembling, sorting and processing into resalable ICs including CPU and memory ICs, metal and plastic raw materials for resale.

作为一家日式弹球盘游乐机回收公司，其在中国和香港地区处于行业领先地位，该公司专注于从事将日式弹球盘游乐机进口到香港进行拆解、分类、处理成可转售的芯片、金属和塑胶原料，用于转售。

NC bender, tube ends former, tube expanding and shrinking machine, tube cutting machine, Pressing machine, tube boring machine, tube swaging machine,...

后生制造与研发的机种大致分类：； 1管成型系列：管末成型机、多段式管端成型机、缩管机、扩管机、弯管。。。

An automatic update system based on semantic web techniques to implement the drug-law checking automatically is proposed in this thesis. The design concept is to use OWL as the knowledge model. To represent the drug law knowledge, we view a drug law as a class of medical status, and we can use classification to implement the decision of drug use legality. This approach simplifies the design of inference engine. The OWL supports the syntax of class inherences, so it can reach the cognitive economy and shorten the drug laws encoding length. The OWL syntax can represent most drug laws style, so the system can just use one document and one inference engine to handle all drug laws rules.

本文导入语意网技术提出一法规检核同步化系统，来实现药品检核规则自动更新的功能，其主要设计理念为以语意网的OWL模型来实作法规检核系统，在法规的知识表徵上，系统将所有的药品规则视为医疗状态的类别，因此判别用药的合法性，便可使用分类运算达成，这简化了推理机的设计；利用OWL支援类别继承的语意，故能形成认知经济性效应，缩小医疗法规机器编码长度；最后OWL的语法能表示大部份规则型态，使得系统可以使用一个文件档，一种推理机模型便能表示所有的法规。

For a big chunk of credit-card losses; the number of filings (and thus charge-off rates) would be rising again, whether

年美国个人破产法的一个改动使得破产登记急速下降，而后引起了信用卡大规模的亏损。

Eph. 4:23 And that you be renewed in the spirit of your mind

弗四23 而在你们心思的灵里得以更新