用三层网络逼近任何连续函数

在神经网络基础理论中，有一个这样的问题：怎样用三层网络实现对任意连续函数的映射，本文利用Ｋｏｈｏｎｅｎ提出的Ｓｅｌｆ－ＯｒｇａｎｉｚｉｎｇＦｅａｔｕｒｅＭａｐ竞争型网络以及区域的２Ｍ－分割概念，解决了用具有有限个隐层单元的三层网络以任意精度逼近一个连续函数的问题，具体的网络结构和学习算法已给出．     

基于分形理论的图象编码改进算法

基于分形理论的图象编码算法可以获得比经典图象编码算法高得多的压缩比，是很有潜力的一种算法。基本的分形图象编码技术虽然有许多优点，但其细码时间长，匹配复杂，从实用考虑，还须从多方面进行改进以提高性能。本文给出一种自适应的改进算法，利用混合编码将块截短编码（BTC）与分形编码技术相结合，同时采用三级分决技术和块分类技术，使压缩比可在20～256之间随图象特点而取值，此算法在信噪比（PSNR）方面亦有所提高。     

On the course content of nonlinear circuits for graduate programs

The author lists the contents of the nonlinear circuits course taught as part of a graduate engineering programme. The course includes the results of new research as well as an emphasis on the application of theories for solving practical problems.<>     

工业电解镍沉积层的微观组织演变特征

本文研究了工业电解镍沉积层的微观组织随着生产时间变化的特征。利用XRD、SEM、EBSD等分析手段对电解镍板择优取向,微观形貌,特征晶界分布等随着生产时间变化的特征进行了研究。结果表明：电沉积镍板在表面上晶体主要以垂直于(200)面的方向生长,在截面上表现为(111)、(200)双择优取向,晶体生长方式为侧向生长。表面与截面表现出不同的微观形貌,表面上由棱锥状转变为胞状,生长机制由螺旋位错驱动生长变为原子聚集堆积长大;各阶段镍板截面形貌始终为层片状组织。沉积层中截面主要为大角度晶界,并产生大量的∑3孪晶界。稳定生长过程中,随着沉积进行大角度晶界比例逐渐下降,∑3晶界相对频率逐渐升高,但是它们都会受到电解槽环境变化的影响。在截面上沿着生长方向,电解镍板主要表现为<001>纤维织构,大量∑3孪晶界对沉积层取向产生影响。通过提高∑3孪晶界比例或选择∑3孪晶界较高频率所对应时间的电解镍,以此获得强度和塑性更优的电解镍。电解镍沉积层演变规律得以明确意味着在后期的研究或者指导工业生产中能够对电解镍的组织进行微观调控,实现工业条件下电解镍的订制生产。     

P(MMA-MA)/OMMT纳米复合材料对PVC的改性研究

把聚(甲基丙烯酸甲酯-丙烯酸甲酯)/有机化蒙脱土(P(MMA-MA)/OMMD纳米复合材料对聚氯乙烯OVC)进行共混改性.通过扫描电镜(SEM)、热失重(TGA)、拉伸、冲击和动态力学测试(DMA)分别表征了共混物的两相相容性、热性能及力学性能.结果发现:共混物的两相共混效果较好且有很好的相容性;纳米复合材料的加入改善了共混物的热稳定性.当添加量达到30%时,共混物的拉伸强度、挠曲模量和冲击强度比纯PVC分别提高了34.6%、75.4%和21.1%.     

一个新的基于度约束的组播路由算法

论文根据网络节点在通信中复制信息的能力,研究了节点的度约束以及费用最小的组播树问题,并提出了一种新的算法,当动态加入节点时,不用完全重新计算就可以得到新的组播树,以适用于某些特殊的场合。通过实验表明,算法的费用接近于以往算法,但是建立初始路由的时间要少得多,实验表明该算法具有较强的实用性。     

模拟电路故障诊断的神经网络方法

本文讨论了在模拟电路故障诊断中应用神经网络的若干方法,涉及了模拟电路故障诊断中的SBT法,SAT法及专家系统方法等,给出了上属方法中应用神经网络的基本原则。     

解决陷入局部极小的一种新方法

本文提出一种能有效。解决Ｈｏｐｆｉｅｌｄ网在能量变。过程之中陷入局部极小问题的方法，这种方法通过有选择性地改变权值矩阵Ｗ的对角元以及有某种特定顺序的串行工作方式来使得网络跳出局部极小值（点）向能量最小点逼近，模拟结果显示；这种方法不仅速度比较快（与模拟退火法ＳｉｍｕｌａｔｅｄＡｎｎｅａｌｉｎｇＭｅｔｈｏｄ比较），而且在每次实验中达到全局最小点的成功率也非常高．在所进行的众多的实验中，迄今为止，尚未出现不成功的情况．     

Pb/S/1,2-ethanedithiol composite thin films for efficient solid-state quantum-dot sensitized TiO<Subscript>2</Subscript> nanorod array solar cells

The Pb/S/1,2-ethanedithiol composite thin films were successfully deposited on TiO₂ nanorod arrays by spin-coating step-by-step 5 mmol dm^?3 Pb(NO₃)₂, Na₂S and 1% 1,2-ethanedithiol solution and their chemical compositions can be easily adjusted by changing the concentration of Na₂S solution from 5 to 3.5 mmol dm^?3 and 2 mmol dm^?3. The average crystal sizes of Pb/S/1,2-ethanedithiol quantum-dots decreased from 7.9 to 7.1 nm and 6.5 nm with the decrease of the concentration of Na₂S solution and the chemical bonding of Pb²⁺ and S in EDT was chelation of the penta-heterocycle in Pb/S/1,2-ethanedithiol composite thin films. All solid-state Pb/S/1,2-ethanedithiol composite thin film sensitized TiO₂ nanorod array solar cells using 5, 3.5, 2 mmol dm^?3 Na₂S solution exhibited the photoelectric conversion efficiency of 2.68, 3.41 and 4.51% under the illumination of simulated AM 1.5 sunlight (100 mA cm^?2).     

Top-k queries on temporal data

The database community has devoted extensive amount of efforts to indexing and querying temporal data in the past decades. However, insufficient amount of attention has been paid to temporal ranking queries. More precisely, given any time instance t, the query asks for the top-k objects at time t with respect to some score attribute. Some generic indexing structures based on R-trees do support ranking queries on temporal data, but as they are not tailored for such queries, the performance is far from satisfactory. We present the Seb-tree, a simple indexing scheme that supports temporal ranking queries much more efficiently. The Seb-tree answers a top-k query for any time instance t in the optimal number of I/Os in expectation, namely, O(log_B \fracNB+\frackB){O\left({\rm log}_B\,\frac{N}{B}+\frac{k}{B}\right)} I/Os, where N is the size of the data set and B is the disk block size. The index has near-linear size (for constant and reasonable k _max values, where k _max is the maximum value for the possible values of the query parameter k), can be constructed in near-linear time, and also supports insertions and deletions without affecting its query performance guarantee. Most of all, the Seb-tree is especially appealing in practice due to its simplicity as it uses the B-tree as the only building block. Extensive experiments on a number of large data sets, show that the Seb-tree is more than an order of magnitude faster than the R-tree based indexes for temporal ranking queries.