基于最大熵模型的词位标注汉语分词

近年来基于字的词位标注汉语分词方法极大地提高了分词的性能,该方法将汉语分词转化为字的词位标注问题,借助于优秀的序列标注模型,词位标注汉语分词逐渐成为汉语分词的主要技术路线.该方法中特征模板集设定和词位标注集的选择至关重要,采用不同的词位标注集,使用最大熵模型进一步研究了词位标注汉语分词技术.在国际汉语分词评测Bakeoff2005的语料上进行了封闭测试,并对比了不同词位标注集对分词性能的影响.实验表明所采用的六词位标注集配合相应的特征模板集TMPT-6较其他词位标注集分词性能要好.     

模糊聚类与信息熵综合评价法在防空C3I系统中的应用

建立了防空C3I系统总体效能评估指标体系,将模糊聚类和信息熵综合到评判过程中,在充分分析已有算法的基础上,给出了对其系统效能进行评估的基于模糊聚类和信息熵的综合评价方法,并通过算例验证了方法的正确性和实用性。     

基于RBF神经网络的线性混叠盲分离算法

提出一种RBF神经网络算法应用于线性混叠信号的盲分离。所用的RBF神经网络算法是从输入信号的数据中训练出中心值和宽度值,再训练通过用最大熵值的代价函数推导的权值。所用的代价函数保证了网络的输出尽可能独立,使信号能正确地分离。仿真验证了所用的算法能减少分离时间和提高分离效率。对比ME算法,该算法更好。     

基于启发式信息熵的粗集数值属性离散化算法

在一致性假设前提下,以数据集的统计性质作为启发式知识,从候选离散点集中选择离散点,根据数据集的期望值和方差来确定搜索最优离散点的区域,提出一种新的基于信息熵粗集数值属性离散化算法,并采用UCI国际标准数据集来验证新算法.新算法与已报道的算法所得到的离散断点集完全一致,决策表的离散化结果也相同,但时间代价不同,新算法比其计算效率提高40%～50%.     

基于粗糙熵和K-均值聚类算法的图像分割

针对基于粗糙熵的图像分割算法不能满足复杂图像的多类目标提取的需要,本文先利用K-均值聚类算法对图像进行区域分割,再利用基于粗糙熵的方法对分割结果进行目标提取,从而达到多阈值分割的目的。通过对遥感图像进行分割处理,证明了改进后算法的有效性。     

一维离散平均曲率方程Neumann问题解的存在性

用紧向量场方程的解集连通理论给出一维离散平均曲率方程Neumann问题Δ[?(Δu(t-1))]=f(t,u(t)),t∈[1,T]?,Δu(0)=Δu(T)=0的上下解方法,并给出其解的存在性结果,其中:[1,T]?:={1,2,…,T-1,T},T≥2是正整数;?(s)=s/√1-s2,s∈(-1,1);非线性项f...     

粗糙Vague集的不确定性度量

将粗糙集理论和Vague集理论相结合,引入了粗糙Vague集的概念,讨论了粗糙Vague集的基本性质.并且基于描述模糊信息量的模糊熵及描述粗糙集的粗糙熵,给出了粗糙Vague集的模糊熵、粗糙度及贴近度的定义,讨论了粗糙Vague集的模糊熵、粗糙度及贴近度,并给出了相应的定理。     

The Entropy of System Decreases with the Deflection from the Equiprobability Distribution

1　IntroductionEntropy,asameasurementthatdescribestheuncertaintyofsubsystemsdistributioninasystem,hasdevelopedunceasingly.First,R.J.E.Clausiusintroduceditintothermodynamics,asastatefunction,todescribetheauto-evolutionabilityofthestate.Thedifferenceofentropybetweenthetwostatesequalstothequantityofheatabsorbedbythesysteminanisothermalprocess,thatisS2-S1=∫21dQT(1)　　L.E.Boltzmannconnecteditwiththeuncertaintyofsubsystemsdistributioninathermodynamicalsystem.Theentropyofathermodynamicalsystem…     

Helmholtz's Kant revisited (Once more). The all-pervasive nature of Helmholtz's struggle with Kant's Anschauung

In this analysis, the classical problem of Hermann von Helmholtz's (1821–1894) Kantianism is explored from a particular vantage point, that to my knowledge, has not received the attention it deserves notwithstanding its possible key role in disentangling Helmholtz's relation to Kant's critical project. More particularly, we will focus on Helmholtz's critical engagement with Kant's concept of intuition [Anschauung] and (the related issue of) his dissatisfaction with Kant's doctrinal dualism. In doing so, it soon becomes clear that both (i) crucially mediated Helmholtz's idiosyncratic appropriation and criticism of (certain aspects of) Kant's critical project, and (ii) can be considered as a common denominator in a variety of issues that are usually addressed separately under the general header of (the problem of) Helmholtz's Kantianism. The perspective offered in this analysis can not only shed interesting new light on some interpretive issues that have become commonplace in discussions on Helmholtz's Kantianism, but also offers a particular way of connecting seemingly unrelated dimensions of Helmholtz's engagement with Kant's critical project (e.g. Helmholtz's views on causality and space). Furthermore, it amounts to the rather surprising conclusion that Helmholtz's most drastic revision of Kant's project pertains to his assumption of free will as a formal condition of experience and knowledge.     

Huffman编码的优化

对图像压缩中常用的Huffman编码进行讨论,给出了在多种Huffman编码中寻找平均偏离方差最小的一种编码方法,以提高数据传输的准确率,同时在数据接收中减小数据缓冲器容量.