查询词典 conditional entropy

Statistical characteristics include Tectonic Density, Degree of Central Symmetry, Skewness Index, Degree of Tectonic Optimal Direction, Power Spectrum Density, Cubic Entropy and so on. The former four indexes are implemented in this paper; Structural characteristics show the characteristics of spatial distribution of geological bodies. In chapter four, a quantitative method for describing geological structure by using the structure "tree" is studied and implemented, including its formalized definition and the measurement of the similarity of different "trees".

统计特征有构造密度、中心对称度、失真指数、优益度、密度功率谱、方位熵等等，本文实现了前四个参数值的提取；结构特征是反映地质体空间分布的结构特点，第四章研究和实现了用"树结构"描述地质结构的定量方法，包括结构的形式化描述和结构的相似性度量。

At first, a model of a pair of excitable neurons coupled via gap junction is built by Chay model, the effect of the strength of gap junction on the synchronous oscillation is studied;then, the synchronous oscillations of a 2-D network are also studied;at last, the Lyapunov exponent and phase portrait are utilized in the analysis of nonlinear, the approximate entropy is utilized to measure the complexity.

首先以Chay模型为基础建立了两个神经元电突触耦合模型，研究了不同电突触耦合强度对两个神经元同步振荡的影响；其次，研究了可兴奋神经元组成的二维网络的同步振荡；最后，应用李雅普诺夫指数和相图分析了同步振荡的非线性，应用近似熵分析了同步振荡的复杂性。

Given a finite training sample set, according to the proposal of establishing a Local Most Entropy probability density function to estimate the number of cluster a method for estimating 〓 is proposed, where a complicated method based on Maximum Likelihood Cross-Validation is replaced by Chebyshev inequality.

为改善BP算法的收敛速度慢、易陷入局部最小的问题，提出了带一类非线性特性动量项的变步长BP算法，构造了一个能自适应升降温的动量项非线性强度函数，在编程中又采用了有条件引入非线性特性动量项的措施，大大提高了搜索过程跳出局部极小、快速稳定收敛的能力。

In the first approach the metadata extraction is viewed as a classification problem. Three statistic classification models, including Support Vector Machines, Perceptron with Uneven Margins, and Maximum Entropy are studied. The approach suffers from the problems of ignoring the variance in lengths of documents and ignoring the dependency of text units within documents. To solve the problems, our second approach takes the metadata extraction as a sequence data labeling problem and makes full use of the sequential information.

论文首先研究了基于分类的元数据抽取方法，考察了支持向量机、非对称间隔感知机和最大熵模型在ADME中的应用；然后针对分类方法中存在的忽略文档之间的差异、不考虑文档上下文等问题，提出了基于序列数据标注模型的抽取方法，并结合最大熵马尔可夫模型、投票感知机模型和条件随机场模型进行了研究；最后本文根据前两种方法的设计思想和优缺点，研究提出一种新的结构预测模型—— CDASP（Coordinate Descent Algorithm for Structure Prediction），并将其应用于元数据的自动抽取中。

For registration of unknown-inner geometry connections, Nonlinear Correlation Information Entropy is proposed as the registration metric, which can use a value in the closed interval [0, 1] to estimate the general relationship among multi-variables, and may not be affected by the order of images.

对于具有未知内部关系的序列刚体图像配准，提出使用非线性相关信息熵作为配准测度。

Industry;water pollution;improved entropy method;variable cluster

工业；水污染；改进熵权法；变量聚类

Evaluation index of industrial water pollution are defined based on industrial wastewater emission standards and the emphasis of water control, according to industrial water pollution emissions data of 31 provinces in China impact of water pollution is calculated by the improved entropy method, lead, cyanide and ammonia nitrogen are in the top three on the impact of water pollution, water pollutant index are divided into heavy metals index and composite index by variable cluster.

根据我国工业各行业的排放标准和水污染控制重点，确定了我国工业水污染状况评价指标，采用改进熵权法对我国31个省区工业水污染物的排放数据进行计算，得到工业废水中铅、氰化物、氮氨对水污染影响居于前三位，应用变量聚类将水污染物分成重金属指标和复合指标，通过熵值对全国各地区工业水污染总体状况、重金属指标、复合指标进行排序，得到东部、中部地区是我国工业水污染较严重地区，但两个地区工业水污染原因有较大区别。

The main topics include entropy and mutual information, data compression, noiseless coding, Shannon theorem, noiseless coding, Shannon Theorem, linear codes, cyclic codes.

本课程宗旨是简介基础的资讯理论，主要课程含嫡及相互信息、资料压缩、无杂讯编码及Shannon定理、线性码、循环码等。

The topics covered include the basic concepts of information theory--entropy, mutual information, channel capacity, information rate, Shannon's noiseless coding theorem and Shannon's fundamental coding theorem; modeling of information sources--zero-memory and Markov models; modeling of information channels--BSC and BEC channels, additivity of information and cascaded channels; construction of compact source codes--Kraft inequality, compact codes, Huffman and LZW compression codes; and analysis and design of error-control channel codes--Hamming distance, binary linear codes and the parity-check matrix, Hamming codes, checksum codes, cyclic codes and the generator polynomial and CRC codes.

课程的内容包括：信息论的基本概念（熵、交互信息、信道容量、信息率、 Shannon无噪声编码定理和Shannon基本编码定理）、信源模型（无记忆模型和Markov模型）、信息信道模型（BSC 和 BEC模型、信息的可加性和级联信道），紧致信源编码设计（Kraft不等式，紧致编码，Huffman 和LZW压缩编码）以及差错控制信道码的设计与分析（海明距离，二元线性码，奇偶校验矩阵，海明码，校验和，循环码，生成多项式和CRC码）。

On the basis of entropy, the objective coefficient can be gotten.

并在对产品性能模糊评价的基础上，得出了产品的性能评价结果排序。

There are different types of personal loans available.

有许多不同类型的个人贷款优惠。

The above mentioned teaching model of spiral rising can be used to avoid rigidness in teaching.

上述螺旋式上升的教学模式，可以避免教学形成模式後的僵化现象。