Acetylene adsorption on defected MIL‐53

Acetylene (C₂H₂) is an important fuel for fuel cells, welding, and metal cutting. In this paper, a defected MIL‐53 is reported for C₂H₂ storage. The defects were simply generated by ball milling MIL‐53. Furthermore, it was found that the adsorption of C₂H₂ on the defected MIL‐53 is stronger than that on MIL‐53 without defection, reflected by adsorption heat increase from 19.3 to 25.5 kJ/mol. As a result, the specific reversible C₂H₂ capacity per surface area increased with increasing defects. Copyright © 2016 John Wiley & Sons, Ltd.     

Multiple kernel-based multi-instance learning algorithm for image classification

In this paper, a novel multi-instance learning (MIL) algorithm based on multiple-kernels (MK) framework has been proposed for image classification. This newly developed algorithm defines each image as a bag, and the low-level visual features extracted from its segmented regions as instances. This algorithm is started from constructing a “word-space” from instances based on a collection of “visual-words” generated by affinity propagation (AP) clustering method. After calculating the distance between a “visual-word” and the bag (image), a nonlinear mapping mechanism is introduced for registering each bag as a coordinate point in the “word-space”. In this case, the MIL problem is transformed into a standard supervised learning problem, which allows multiple-kernels support vector machine (MKSVM) classifiers to be trained for the image categorization. Compared with many popular MIL algorithms, the proposed method, named as MKSVM-MIL, shows its satisfactorily experimental results on the COREL dataset, which highlights the robustness and effectiveness for image classification applications.     

基于EMD的融合特征快速多示例人脸识别算法

在基于五官模版技术的人脸识别中,因光照、角度及缺少整体性考虑等情况,易产生异常示例,影响了算法精度;而适合多示例检索的EMD测度寻优路径较长,导致在人脸识别中应用受限。为此本文提出了一种基于EMD测度的快速融合特征多示例人脸识别算法（IIFEMD-MIL）。针对异常点的问题,通过引入结合整体特征的融合多示例技术以及距离阀值,从而减少异常示例的产生并对超过阈值的示例予以平滑处理;针对寻优路径长的问题,将人脸五官之三结合整体示例为模版构建四示例的一一匹配,并进一步提出了融合快速EMD-MIL框架,缩短了寻优遍历路径。在ORL和MIT图像集上进行的比对实验表明,该算法执行效率和分类准确性优于其他同类算法。     

Analysis of HBM ESD testers and specifications using a fourth-order lumped element model

This paper presents the general and analytical solution of a fourth-order lumped element model (LEM) to describe human body model (HBM) electrostatic discharge (ESD) testers including the main tester parasitic elements. The analytical fitting to the LEM of experimentally obtained HBM pulse data is a new scientifically justified tool to determine HBM tester parasitic elements. The impact of the test board capacitance on HBM testing is demonstrated and explained. Furthermore, the MIL 883C/3015.7¹ and EOS/ESD S5.1-1991² Standards on HBM testers are evaluated upon their selectivity to the test board capacitance. Finally, recommendations to improve HBM tester specifications regarding their selectivity to the parasitic test board capacitance are formulated     

Ru Species Supported on MOF‐Derived N‐Doped TiO2/C Hybrids as Efficient Electrocatalytic/Photocatalytic Hydrogen Evolution Reaction Catalysts

The development of efficient catalysts is of great importance for hydrogen evolution reaction (HER) of water splitting via electrocatalytic/photocatalytic processes to remediate the current severe environmental and energy problems. By aid of the stabilization effects of uncoordinated groups and inherent pore‐confinement of amine‐functionalized metal–organic frameworks (NH₂‐MIL‐125), two forms of Ru species including nanoparticles (NPs) and/or single atoms (SAs) can be firmly embedded in NH₂‐MIL‐125 derived N‐doped TiO₂/C support (N‐TC), and thus obtain two kinds of samples named Ru‐NPs/SAs@N‐TC and Ru‐SAs@N‐TC, respectively. In the synthetic process, the initial feeding amount of Ru³⁺ ions not only strongly determines the final size and dispersion states of Ru species but also the morphology and defective structures of N‐TC support. Impressively, Ru‐NPs/SAs@N‐TC exhibit superior catalytic activities to Ru‐SAs@N‐TC for either electrocatalytic or photocatalytic HER. This should be attributed to its larger specific surface area and benefiting from synergistic coupling of Ru NPs and Ru SAs. It is envisioned that the present work can provide a new avenue for development of high‐efficiency and multifunctional hybrid catalysts in sustainable energy conversion.     

融合SIFT和MIL的红外人脸识别方法

针对红外人脸识别问题,提出一种新的基于尺度不变特征转换(SIFT)与多示例学习(MIL)相结合的算法。该算法将图像当作多示例包,SIFT描述子当作包中的示例,利用聚类的方法对训练集中的所有SIFT描述子进行聚类,建立"视觉词汇表",再根据"视觉字"在多示例训练包中出现的频率,建立"词-文档"矩阵,采用潜在语义分析(LSA)的方法获得多示例包(图像)的潜在语义特征,将MIL问题转化成标准的有监督学习问题,即在潜在语义空间用支持向量机(SVM)求解MIL问题。基于OTCBVS标准数据集的对比实验结果表明,所提算法是可行的,且识别率明显高于其他方法。     

一种基于CRO的高阶神经网络多示例学习方法

在多示例学习(MIL)中,包是含有多个示例的集合,训练样本只给出包的标记,而没有给出单个示例的标记。提出一种基于示例标记强度的MIL方法(ILI-MIL),其允许示例标记强度为任何实数。考虑到基于梯度训练神经网络方法的计算复杂性和ILI-MIL目标函数的复杂性,利用基于化学反应优化的高阶神经网络来实现ILI-MIL,学习方法具有较强的非线性表达能力和较高的计算效率。实验结果表明,该算法比已有算法具有更加有效的分类能力,且适应范围更广。     

Metal Organic Framework Crystals in Mixed‐Matrix Membranes: Impact of the Filler Morphology on the Gas Separation Performance

Mixed‐matrix membranes comprising NH₂‐MIL‐53(Al) and Matrimid or 6FDA‐DAM have been investigated. The metal organic framework (MOF) loading has been varied between 5 and 20 wt%, while NH₂‐MIL‐53(Al) with three different morphologies, nanoparticles, nanorods, and microneedles has been dispersed in Matrimid. The synthesized membranes have been tested in the separation of CO₂ from CH₄ in an equimolar mixture. At 3 bar and 298 K for 8 wt% MOF loading, incorporation of NH₂‐MIL‐53(Al) nanoparticles leads to the largest improvement compared to nanorods and microneedles. The incorporation of the best performing filler, i.e., NH₂‐MIL‐53(Al) nanoparticles, into the highly permeable 6FDA‐DAM has a larger effect, and the CO₂ permeability increases up to 85% with slightly lower selectivities for 20 wt% MOF loading. Specifically, these membranes have a permeability of 660 Barrer with a CO₂/CH₄ separation factor of 28, leading to a performance very close to the Robeson limit of 2008. Furthermore, a new non‐destructive technique based on Raman spectroscopy mapping is introduced to assess the homogeneity of the filler dispersion in the polymer matrix. The MOF contribution can be calculated by modeling the spectra. The determined homogeneity of the MOF filler distribution in the polymer is confirmed by focused ion beam scanning electron microscopy analysis.     

典型CMOS器件总剂量加速试验方法验证

应用美军标试验方法1019．5和1019．4分别对两种典型（Complementary Metal-Oxide—Semiconductor Transistor,CMOS）器件进行试验验证,论述了试验原理,对试验现象进行了详细的分析,了解实验室条件下评估空间氧化物陷阱电荷和界面陷阱电荷的可行性和保守性。实验表明,两种不同工艺的CC4069器件没有通过1019．5所做出的试验验证,从实验现象观察认为是由于界面陷阱电荷大量建立所引起的。加固LC4007-RHA NMOS却通过了与1019．5相比过分保守的1019．4的试验验证。     

TLD目标跟踪算法研究

在TLD (Tracking Learning Detector)算法的基础上,提出了一种基于Online MIL(Online Multiple Instance Learning)的TLD目标跟踪算法.算法使用改进的MIL跟踪器进行目标跟踪,提高了目标在被遮挡情况下跟踪的鲁棒性.另外,在学习机制部分,目标位置的决策策略使用了跟踪结果优先的原则,使算法能适应出现类似目标时的跟踪.实验结果表明,该方法能够长时间准确地跟踪目标,并在出现类似目标时;跟踪效果较好.此外,改进后的算法在跟踪的稳定性和跟踪效率上较原算法提高了1倍.