融合累积变异比和集成超限学习机的高光谱图像分类

目的 高光谱图像具有高维度的光谱结构,而且邻近波段之间往往存在大量冗余信息,导致在随机样本选择策略和图像分类过程中出现选择波段算法复杂度较高和不适合小样本的现象。针对该问题,在集成学习算法的基础上,考虑不同波段在高光谱图像分类过程中的作用不同,提出一种融合累积变异比和超限学习机的高光谱图像分类算法。方法 定义波段的累积变异比函数来确定各波段在分类算法的贡献程度。基于累积变异比函数剔除低效波段,并结合空谱特征进行平均分组加权随机选择策略进行数据降维。为了进一步提高算法的泛化能力,对降维后提取的空谱特征进行多次样本重采样,训练得到多个超限学习机弱分类器,再将多个弱分类器的结果通过投票表决法得到最后的分类结果。结果 实验使用Indian Pines、Pavia University scene和Salinas这3种典型的高光谱图像作为实验标准数据集,采用支持向量机（support vector machine,SVM）,超限学习机（extreme learning machine,ELM）,基于二进制多层Gabor超限学习机（ELM with Gabor,GELM）,核函数超限学习机（ELM with kernel,KELM）,GELM-CK（GELM with composite kernel）,KELM-CK（KELM with composite kernel）和SS-EELM（spatial-spectral and ensemble ELM）为标准检测算法验证本文算法的有效性,在样本比例较小的实验中,本文算法的总体分类精度在3种数据集中分别为98.0%、98.9%和97.9%,比其他算法平均分别高出9.6%和4.7%和4.1%。本文算法耗时在3种数据集中分别为15.2 s、60.4 s和169.4 s。在同类目标空谱特性差异较大的情况下,相比于分类精度较高的KELM-CK和SS-EELM算法减少了算法耗时,提高了总体分类精度;在同类目标空谱特性相近的情况下,相比于其他算法,样本数量的增加对本文算法的耗时影响较小。结论 本文算法通过波段的累积变异比函数优化了平均分组波段选择策略,针对各类地物目标分布较广泛并且同类目标空谱特性差异较大的高光谱数据集,能够有效提取特征光谱维度的差异性,确定参数较少,总体分类效果较好。     

基于遗传优化的自适应凸松弛人体姿势估计*

针对凸松弛方法在解决三维人体姿势估计的问题时存在迭代次数较多、准确度不高的不足,提出一种基于遗传优化的自适应凸松弛人体姿势估计算法。该算法首先对关键参数的更新方式进行自适应处理,然后利用遗传优化算法对该关键参数的初始值进行寻优,最后利用寻优结果对凸松弛方法中闭式解的公式进行调整。实验结果表明,提出的算法迭代次数更少,准确度更高,更有利于实际应用。     

基于改进单类支持向量机的工业控制网络入侵检测方法

针对单类支持向量机（OCSVM）入侵检测方法无法检测内部异常点和离群点导致决策函数偏离训练样本的问题,提出了一种结合具有噪声的密度聚类（DBSCAN）方法和K-means方法的OCSVM异常入侵检测算法。首先通过DBSCAN算法,剔除训练数据中的离群点,消除离群点的影响;然后利用K-means划分数据类簇的方法筛选出内部异常点;最后利用OCSVM算法为每一个类簇建立单分类器用于检测异常数据。工控网络数据集上的实验结果表明,该组合分类器能够利用无异常数据样本检测出工控网络入侵,并且提高了OCSVM方法的检测效果。在气体管道网络数据集入侵检测实验中,所提方法的总体检测率为91.81%;而原始OCSVM算法则为80.77%。     

Electrically Reconfigurable 3D Spin-Orbitronics

The explosive demands of storage capacity and the von Neumann bottleneck of modern computer architectures trigger many innovations in information technology. Amongst them, nonvolatile spintronics attract considerable attentions for which can embed the computation capability into memory, enable neuromorphic, and probabilistic computing. These exciting progresses typically rely on the manipulation of the relative magnetization orientations of two magnetic layers. By extending to 3D spintronic architectures made of multiple magnetic layers (n), the exponentially increased 2ⁿ magnetic states can provide ample opportunities for implementing novel spintronic functionalities. Here, through building perpendicularly magnetized 3D spin-orbitronic architectures – [Pt/Fe_1−xTb_x/Si₃N₄]_n multilayers, it is demonstrated the electrical programing of 2ⁿ memory states via current-induced spin–orbit torques (SOTs), and the accompanied reconfigurable multifunction in-memory logic features in a single four-terminal Hall device. Further, an electrical readout of these 2ⁿ states, together with the implementation of Boolean logic gates and digital circuitry such as 2–4 and 3–8 decoders, are successfully conducted. More complex logic circuits are also envisioned. The experiments thus substantiate 3D spin-orbitronic structures as a promising platform for exponentially boosting the storage capacity and accommodating in-memory computing that can be important for promoting the emerging 3D nanospintronics.     

Thermoresponsive Hydrogels as Microniches for Growth and Controlled Release of Induced Pluripotent Stem Cells

The recently emerging stem-cell artificial niche engineering in induced pluripotent stem cell (iPSCs) 3D cultures has provided enormous opportunities to fully utilize the potential of these cells in biomedical applications. Although a fully chemically defined niche environment can supply cells with desirable safety for clinical use, establishing an artificial degradable niche environment for the controlled release of proliferated cells under mild conditions is still a big challenge. Here, an advanced controlled releasable iPSC 3D artificial niche is reported based on dendritic polyglycerol and poly(N-isopropylacrylamide)-co-polyethylene glycol polymers via a physical–chemical cogelation strategy. Benefiting from the chemically defined synthetic materials and their precise cooperation by covalent cross-linking and physical phase transition, the cogelation-based artificial niche system can be adjusted with optimal parameters and owns high cell biocompatibility to support the robust production of high quality iPSCs with an excellent expansion efficiency. Moreover, the expanded cells can be released out of their niche environment controllably only by adjusting the temperature. Overall, this controlled release hydrogel scaffold shows great promise in iPSC 3D culture for downstream applications.     

基于频谱空间域特征注意的音乐流派分类算法

为了提升深度卷积神经网络对音乐频谱流派特征的提取效果,提出一种基于频谱空间域特征注意的音乐流派分类算法模型DCNN-SSA。DCNN-SSA模型通过对不同音乐梅尔谱图的流派特征在空间域上进行有效标注,并且改变网络结构,从而在提升特征提取效果的同时确保模型的有效性,进而提升音乐流派分类的准确率。首先,将原始音频信号进行梅尔滤波,以模拟人耳的滤波操作对音乐的音强及节奏变化进行有效过滤,所生成的梅尔谱图进行切割后输入网络;然后,通过深化网络层数、改变卷积结构及增加空间注意力机制对模型在流派特征提取上进行增强;最后,通过在数据集上进行多批次的训练与验证来有效提取并学习音乐流派特征,从而得到可以对音乐流派进行有效分类的模型。在GTZAN数据集上的实验结果表明,基于空间注意的音乐流派分类算法与其他深度学习模型相比,在音乐流派分类准确率和模型收敛效果上有所提高,准确率提升了5.36个百分点～10.44个百分点。     

Thermal and physical properties of allyl PPO and its composite

The thermal behavior of allyl PPO and its cured resin were investigated. In the allyl PPO curing process, the specific temperatures were T_gel = 173.6°C, T_cure = 225.4°C, and T_treat = 237.7°C, and the activation energy (E_a) was 122 kJ/mol. The average number of PPO molecular units between two crosslinking points was about 20. In the degradation process of cured allyl PPO resin, the temperature at which mass loss equaled 1% was much higher than 300°C. The E_a for degradation was calculated as 125 kJ/mol, with a degradation fraction (α) in the range of 0.15–0.65, or 117 kJ/mol with an α of 0.10–0.90. The most probable mechanism function of decomposition of the cured allyl PPO resin was f(α) = 2(1 ? α)^3/2 or g(α) = (1 ? α)^?1/2 ? 1. The thermocompressed laminate of the allyl PPO blending with an additive resin (made from BDM and DP) exhibited the desired properties. ©2006 Wiley Periodicals, Inc. J Appl Polym Sci 102: 4111–4115, 2006     

基于超扭曲优化算法的风机最大功率跟踪控制

针对风力发电系统在低风速区采用传统控制方法具有风能转换效率较低、风轮转速跟踪实时风速的性能较差、发电机转矩波动范围较大等问题,提出了一种将超扭曲算法与最佳转矩法相结合的最大功率跟踪改进控制策略。为进一步改善控制性能,采用粒子群算法对控制器参数进行优化。最后以风轮角速度、发电机输出功率、发电机转矩、功率系数等为评价指标,通过Matlab/Simulink平台验证所提方法的可行性与有效性。仿真结果表明,所提控制策略在提高最大风能捕捉能力的同时可有效地抑制发电机转矩的抖振。     

低周反复荷载下新型组合框架结构受力性能的试验

按照现行规范的有关规定设计制作了1榀2跨单层新型外包钢-混凝土组合梁-普通钢混凝土柱的组合框架结构模型，并通过施加恒定竖向荷载和低周反复水平荷载，对模型框架进行了抗震性能试验研究。分析了新型组合框架滞回性能、变形能力、承载能力、延性、耗能能力、刚度退化等受力性能。试验表明，新型组合框架结构具有良好的抗震能力。     

Preparation and characterization of nanosized Zn-Co spinel oxide by solid state reaction method

ZnCo₂(C₂O₄)₃·6H₂O nanocrystal was prepared by a one-step solid state reaction method at room temperature under ambient conditions using oxalic acid as a chelating agent. Adding oxalic acid to the mixture of zinc acetate dihydrate and cobalt acetate tetrahydrate, grinding and subsequently calcining the precursor at different temperatures, the pink precursor was attained. Thermogravimetric analysis and differential scanning calorimetry have been used to study the thermal behaviors of the precursor, which reveal that the precursor is a compound. Powder X-ray diffraction data show that the sample obtained by heating the precursor between 350 °C and 550 °C for 3 h is a single-phase cubic material having the spinel-type structure, which can be confirmed by infrared spectra. Transmission electron microscopy images show that the grain size of the precursor and the products are about 50 nm and 20 nm, respectively.