回归相关神经网络的分析与改进

本文分析了回归相关神经网络容量与动态范围之间的关系，指出在一定条件下，回归相关神经网络具有很高的存储容量。最后给出了几种改进网络性能的方案。     

Improvement on dynamic elastic interpolation technique for reconstructing 3-D objects from serial cross sections [biomedical application

An improved method for automatically reconstructing a three-dimensional object from serial cross sections is presented. The method improves the dynamic elastic contour interpolation technique. There are two major improvements: (1) in the case of branching that involves concave contours, instead of pairwise interpolation between the start contour and each goal contour, the goal image is considered globally and local constraints are imposed on the forces exerting upon the vertices; and (2) it takes the continuity of high-order derivatives into consideration and incorporates the schemes of spline theory, quadratic-variation-based surface interpolation, and finite-element-based multilevel surface interpolation to create smoother surfaces of the reconstructed object. Based on the output from the preliminary processing, two alternatives, a quadratic-variation-based surface interpolation algorithm and a finite-element-based multilevel surface interpolation algorithm, can be adopted to obtain the final surface representation     

Reconstruction of two-dimensional permittivity distribution using the distorted Born iterative method

The distorted Born iterative method (DBIM) is used to solve two-dimensional inverse scattering problems, thereby providing another general method to solve the two-dimensional imaging problem when the Born and the Rytov approximations break down. Numerical simulations are performed using the DBIM and the method proposed previously by the authors (Int. J. Imaging Syst. Technol., vol.1, no.1, p.100-8, 1989) called the Born iterative method (BIM) for several cases in which the conditions for the first-order Born approximation are not satisfied. The results show that each method has its advantages; the DBIM shows faster convergence rate compared to the BIM, while the BIM is more robust to noise contamination compared to the DBIM.     

石墨化钢石墨化过程的金相分析及其动力学方程

在650、680和710 ℃不同温度条件下对碳质量分数为0.66%的淬火高碳钢进行了石墨化处理,并利用场发射扫描电子显微镜、电子探针、X-射线衍射仪和透射电子显微镜对其石墨化过程的组织进行金相分析,以及利用组织转变动力学理论,绘制了其石墨化过程的动力学曲线,并建立了相应的动力学方程。研究结果显示:在石墨化过程中,淬火马氏体首先向析出碳化物的稳定状态转变,且在碳化物为渗碳体Fe₃C时,石墨粒子析出速度开始明显增加;基体组织中针叶状α-Fe发生再结晶,由等轴状铁素体逐步代替针叶状的α-Fe;铁素体中的碳含量随着石墨化时间的延长而逐步降低,即由过饱和状态转变为稳定态,碳含量在石墨粒子中突变增为峰值,而铁含量则突变降为谷值,由此表明,渗碳体分解的碳向石墨核心扩散,铁自石墨核心处扩散出来,而形成石墨粒子;石墨粒子面积分数随时间变化的曲线呈S形状,即该动力学过程符合动力学模型JMAK（Johnson-Mehl-Avrami-Kolmogorov）方程,且该方程中的n值为1.5～1.7。      

Atmospheric pressure chemical ionization mass spectrometry for analysis of lipids

Atmospheric pressure chemical ionization (APCI) mass spectrometry (MS) has proven to be a very valuable technique for analysis of lipids from a variety of classes. This instrumental method readily produces useful ions with gentle fragmentation from large neutral molecules such as triacylglycerols and carotenoids, which are often difficult to analyze using other techniques. Molecules that are easily ionized, such as phospholipids, produce molecular ions and diagnostically useful fragment ions that are complementary to those produced by methods such as electrospray ionization MS with collision-induced dissociation. The simplicity and versatility of APCI-MS make it an ideal tool for use in solving hitherto very difficult analytical problems.     

Improved activity and thermostability of Candida antarctica lipase B by DNA family shuffling

DNA family shuffling was used to create chimeric lipase B proteins with improved activity toward the hydrolysis of diethyl 3-(3',4'-dichlorophenyl)glutarate (DDG). Three homologous lipases from Candida antarctica ATCC 32657, Hyphozyma sp. CBS 648.91 and Crytococcus tsukubaensis ATCC 24555 were cloned and shuffled to generate a diverse gene library. A high-throughput screening assay was developed and used successfully to identify chimeric lipase B proteins having a 20-fold higher activity toward DDG than lipase B from C.antarctica ATCC 32657 and a 13-fold higher activity than the most active parent derived from C.tsukubaensis ATCC 24555. In addition, the stability characteristics of several highly active chimeric proteins were also improved as a result of family shuffling. For example, the half-life at 45 degrees C and melting point (T(m)) of one chimera exceeded those of lipase B from C.antarctica ATCC 32657 by 11-fold and 6.4 degrees C, respectively, which closely approached the stability characteristics of the most thermostable parent derived from Hyphozyma sp. CBS 648.91.     

弱监督深度语义分割网络的多源遥感影像水体检测

目的 深度语义分割网络的优良性能高度依赖于大规模和高质量的像素级标签数据。在现实任务中,收集大规模、高质量的像素级水体标签数据将耗费大量人力物力。为了减少标注工作量,本文提出使用已有的公开水体覆盖产品来创建遥感影像对应的水体标签,然而已有的公开水体覆盖产品的空间分辨率低且存在一定错误。对此,提出采用弱监督深度学习方法训练深度语义分割网络。方法 在训练阶段,将原始数据集划分为多个互不重叠的子数据集,分别训练深度语义分割网络,并将训练得到的多个深度语义分割网络协同更新标签,然后利用更新后的标签重复前述过程,重新训练深度语义分割网络,多次迭代后可以获得好的深度语义分割网络。在测试阶段,多源遥感影像经多个代表不同视角的深度语义分割网络分别预测,然后投票产生最后的水体检测结果。结果 为了验证本文方法的有效性,基于原始多源遥感影像数据创建了一个面向水体检测的多源遥感影像数据集,并与基于传统的水体指数阈值分割法和基于低质量水体标签直接学习的深度语义分割网络进行比较,交并比（intersection-over-union,IoU）分别提升了5.5%和7.2%。结论 实验结果表明,本文方法具有收敛性,并且光学影像和合成孔径雷达（synthetic aperture radar,SAR）影像的融合有助于提高水体检测性能。在使用分辨率低、噪声多的水体标签进行训练的情况下,训练所得多视角模型的水体检测精度明显优于基于传统的水体指数阈值分割法和基于低质量水体标签直接学习的深度语义分割网络。     

钒对430不锈钢内裂纹热愈合的影响

对添加不同含量钒的430不锈钢试样,采用钻孔压缩方法获得内裂纹,然后对其进行热愈合实验,研究了钒对试样内裂纹热愈合和形态演化的影响。扫描电镜分析结果表明,在加热温度为830℃时,试样的内裂纹产生了明显的愈合;当钒添加量为0.3%时,愈合效果最好。钒添加到基体材料后,部分钒固溶于基体,降低了基体金属Fe原子的扩散激活能,加剧了Fe的自扩散。分析认为,钢内裂纹愈合机理主要是扩散反应导致的热愈合及重结晶。