特征向量

Firstly the sampling convolutive mixture signals are reseted for matching instantaneous mixture model, then considering non-stationarity of the sources, space whitening and joint block-diagonalization method is exploited to obtain the original signals.

该算法先将采集到的卷积混合信号进行重排，使之满足重新定义向量后的瞬时混合模型特征，然后考虑到信号的非平稳特性，采用空间白化和联合近似块对角化方法分离出源信号。

This model uses nonlinear mapping to map samples into high dimensional feature space, and compares the distribution of each type of samples in the high dimensional space. So that a binary decision tree is constructed through clustering, which can make the more separable classes classified at the upper node of the binary decision tree. The model overcomes the accumulation of misclassification and avoids the condition of impartibility efficiently. At the same time, each node of decision tree uses binary least squares wavelet support vector machine to obtain higher generalization ability.

该模型利用非线性映射将样本投影到高维特征空间，比较每类样本在高维空间的分布情况，进行聚类构造出一个二叉决策树，使容易区分的类别从根节点逐层分类出来，有效克服了错分积累和避免不可分情况；同时，各个节点采用二值最小二乘小波支持向量机，以获得较高的泛化能力。

A frame image of a video frequency sequence is read into a postpositional memory zone, and staring position and size information of a tracked object is obtained in the frame, then distribution statistical information of a target signature is extracted to build a gauss mixed model as an object template, a mean vector and a covariance matrix in gauss mixed distribution are used to describe the position and the size of the object, then the next frame image of the video frequency sequence is read into the postpositional memory zone. In a new frame video frequency image, a parameter estimation method is used to obtain object gauss mized model parameters in the current frame in iterative computation and find candidate template similar to the object template, and the final model parameters obtained in iteration are used to update the tracking window for realizing self-adaption of the tracked window.

将视频序列的一帧图像读入缓存区，获得该帧中被跟踪目标的初始位置和尺寸信息，然后提取目标特征的分布统计信息建立高斯混合模型作为目标模板，用高斯混合分布的均值向量和协方差矩阵描述目标的位置和尺寸，再将视频序列下一帧图像读入缓存区，在新一帧视频图像中利用参数估计方法迭代计算当前帧中目标高斯混合模型参数找到与目标模板最相似的候选模板，利用迭代得到的最终模型参数更新跟踪窗口实现跟踪窗宽的自适应。

After analysis and study on email, by disperseing and charactering email, my paper use vector to express email. And then bring forward a determination tree classifying model base on information entropy. At last I do a series of experiment and testing.

本文通过对电子邮件的分析和研究，提出对邮件结构字段信息和邮件正文信息加以离散和特征化处理，用向量的方式表示电子邮件；接着针对向量化的邮件表示，建立了一种基于信息熵的决策树邮件分类识别模型；论文最后通过编程对该模型作了一系列的实验和测试。

This paper designs the automatic text categorization system of public security information based on Support Vector Machine,according to the requirements of public security information.This system not only distinguishes the text classification automatically,but also retrains the features information at different stages of the categorization process.It will support the further information procession.Meanwhile,this system reduces the coupling degree of the subcomponents,provides the flexible system structure,and improves the adaptability of system.

根据公安情报分类系统的需求，设计了基于支持向量机的公安情报分类系统，该系统不仅能够实现情报文本的自动分类，而且能够保留在情报文本分类的不同阶段语料的特征信息，为情报信息的进一步加工处理提供支持，同时系统中各模块间采用松耦合的方式衔接，提高了系统的适应性和灵活性。

In order to improve the performance of speaker recognition, the key techniques are feature extraction and recognition algorithm.

4基于支持向量机的话者识别是本文的另一个研究内容，提取表征说话人个性的特征和寻找高效的识别算法是话者识别取得良好性能的关键。

An intelligent fault detection and diagnosis method based on space vector is presented, fault diagnosis knowledge self-acquisition and self-tuning and symptom optimum algorithm are infered. 4、An early fault detection and diagnosis method based on ANN for process industry is presented, which has perfect self-learning and self-tuning ability and can establish diagnosis model automatically with the process of fault diagnosis in the absence of sufficient prior knowledge.

提出一种基于特征空间向量的智能故障诊断方法，推导了其故障诊断知识的自动获取算法和故障诊断的自适应算法以及征兆信号优化选取算法； 4、提出一种基于人工神经网络的生产过程早期故障检测与诊断方法，该故障诊断方法知识获取容易，具有良好的自学习和自适应能力，能够在故障先验知识不完备的情况下，通过与用户的交互，逐步自动建立起完善的故障模型知识。

Firstly, 19 seismic characteristic parameter vectors are extracted in about five species such as grey, fuzzy, fractal, amplitude , self correlation and power spectra from data sequence of the target stratum.

首先对目的层地震数字序列提取灰色、模糊、分维、振幅、自相关5类19个地震特征参数向量，然后在地质建模的基础上，进行灰色模式训练与识别，获得整个评价区的油气检测结果。

Based on the characteristic of the vegetation scatter plots in the redband and near infrared band, this paper mainly aims to probe a new method ofextracting vegetation changes by building the coordinate plane of brightindex-perpendicular vegetation index and then getting the variation vectors.

基于植被在红光波段与近红外波段坐标系中的散点图特征，本文着重探讨了一种通过建立多时相&亮度-垂直植被指数&坐标平面，利用变化向量提取矿区植被变化的方法。

According to the features of natural language, conduct a deep study and comparison in many fields such as the algorithm optimization, features'representation and selection, multi-class problem, parameter selection and probabilistic output of support vector machine such that the general machine learning technology can be applied into natural language processing tasks. 2. According to the feature of lexical tagging , propose a novel disambiguate algorithm that improves the performance and efficiency of this task and provides a good fundament for deeper analysis.

针对自然语言处理任务的特点，对于算法优化、特征的表示和选择、多分类问题、参数选择和支持向量机的概率输出等问题进行深入的研究和比较，使自然语言处理任务可以应用通用的高精度机器学习方法得以解决。

In the negative and interrogative forms, of course, this is identical to the non-emphatic forms.

。但是，在否定句或疑问句里，这种带有"do"的方法表达的效果却没有什么强调的意思。

Go down on one's knees;kneel down

屈膝跪下。。。下跪祈祷