一种求解多峰函数优化问题的演化算法

针对演化计算产生新解无序的问题，提出了基于相似性的邻域搜索策略．利用邻域搜索，可以方便地建立自适应的新解产生机制．针对演化算法设计中存在的搜索效果和效率平衡问题，提出了利用适应值对个体进行分级的搜索策略．通过对个体的分级，可以区分个体在搜索过程中的职能：优秀的个体进行局部极小值的开采；其他的个体进行搜索空间的探索，以发现新的局部极小值．数值实验表明，新算法能有效处理低维多峰函数，能找到所有的全局最优解．对高维多峰函数．也能找到全局最优解．     

一种基于内容的数字视频典型帧提取方法

典型帧提取是多媒体视频检索的一个重要技术分支。本文提出一种基于内容的数字视频典型帧提取方法，并通过实验对算法的有效性作了验证。     

A continuous model of the dynamical systems capable to memorise multiple shapes

This paper proposes the novel approach to the mathematical synthesis of continuous self-organising systems capable to memorise and restore own multiple shapes defined by means of functions of single spatial variable or parametric models in two-dimensional space. The model is based on the certain universal form of the integral operator with the kernel representing the system memory. The technique for memorising shapes uses the composition of singular kernels of integral operators. The whole system is described by the potential function, whose minimisation leads to the non-linear dynamics of shape reconstruction by integro-differential non-linear equations with partial derivatives. The corresponding models are proposed and analysed for both parametric and non-parametric shape definitions. Main features of the proposed model are considered, and the results of numerical simulation are shown in case of three shapes memorising and retrieval. The proposed model can be used in theory of smart materials, artificial intelligence and some other branches of non-linear sciences where the effect of multiple shapes memorising and retrieval appears as the core feature.     

Using a similarity measure for credible classification

This paper concerns classification by Boolean functions. We investigate the classification accuracy obtained by standard classification techniques on unseen points (elements of the domain, {0,1}ⁿ, for some n) that are similar, in particular senses, to the points that have been observed as training observations. Explicitly, we use a new measure of how similar a point x∈{0,1}ⁿ is to a set of such points to restrict the domain of points on which we offer a classification. For points sufficiently dissimilar, no classification is given. We report on experimental results which indicate that the classification accuracies obtained on the resulting restricted domains are better than those obtained without restriction. These experiments involve a number of standard data-sets and classification techniques. We also compare the classification accuracies with those obtained by restricting the domain on which classification is given by using the Hamming distance.     

Molecular surface-volume and property matching to superpose flexible dissimilar molecules

Summary Steric complementarity is a prerequisite for ligand-receptor recognition; this implies that drugs with a common receptor binding site should possess sterically similar binding surfaces. This principle is used as the basis for an automatic and unbiased method that superposes molecules. One molecule is rotated and translated to maximize the overlap between the two molecular surface volumes. A fast grid-based method is used to determine the extent of this overlap, and this is optimized using simulated annealing. Matches with high steric similarity scores are then sorted on the basis of both hydrogen-bond and electrostatic similarity between the matched molecules. Flexible molecules are treated as a set of rigid representative conformers. The algorithm has correctly predicted superpositions between a number of pairs of molecules, according to crystallographic data from ligands that have been co-crystallized at common enzyme binding sites.     

A New Fourier Iterative Algorithm for the Design of Phase－only Diffractive Optical Element Used in Laser Beam Shaping

ＡＮｅｗＦｏｕｒｉｅｒＩｔｅｒａｔｉｖｅＡｌｇｏｒｉｔｈｍｆｏｒｔｈｅＤｅｓｉｇｎｏｆＰｈａｓｅ－ｏｎｌｙＤｉｆｆｒａｃｔｉｖｅＯｐｔｉｃａｌＥｌｅｍｅｎｔＵｓｅｄｉｎＬａｓｅｒＢｅａｍＳｈａｐｉｎｇＳＡＮＧＴａｏ；ＬＩＡＯＪｉａｎｇｈｏｎｇ；ＬＵＺ...     

基于多层次特征的跨场景服装检索

针对跨场景服装检索如何提取更有表述力的服装特征问题,提出了一种新的基于高层公共特征约束的相似性度量算法。首先,通过类别空间学习提取不同场景域的类别信息;然后,在场景域空间网络用类别信息约束传统对比损失函数,增大对类间负样本对的惩罚以减轻过拟合;最后,融合类别公共特征和域特定特征并通过类别判断进行辅助检索。分析和实验结果表明,新算法对跨场景服装检索的准确度要优于当前前沿的方法。     

Similarity Solution of the Flamant Problem by Means of a One-Parameter Group Transformation

The two-dimensional point load problem on a linear elastic half plane is solved using a one-parameter group transformation of the biharmonic equation. This solution technique allows the incorporation of the singular load directly through use of the Dirac delta function. The resulting fourth-order ordinary differential equation, while formidable in appearance, is exact and is readily solved through simple integration.