基于1D CNN与XGBoost的恶意代码纹理检测

恶意代码数量已经呈现爆炸式增长,对于恶意代码的检测防护显得尤为重要.近几年,基于深度学习的恶意代码检测方法开始出现,基于此,提出一种新的检测方法,将恶意代码二进制文件转化为十进制数组,并利用一维卷积神经网络(1 Dimention Convolutional Neural Networks,1D CNN)对数组进行分类和识别.针对代码家族之间数量不平衡的现象,该算法选择在分类预测上表现良好的XGBoost,并对Vision Research Lab中的25个不同恶意软件家族的9458个恶意软件样本进行了实验.实验结果表明,所提的方法分类预测精度达到了97％.     

新型含吡啶基团二羧酸配体的合成及单晶培养

以1，4-二溴-2，5-二甲基苯和4-吡啶硼酸为原料，经两步合成了一种新颖的含吡啶基团的二羧酸配体。通过核磁共振氢谱及傅里叶变换红外光谱对配体结构进行表征，通过热重分析对配体的热稳定性进行了测试；通过溶剂热法对配体的单晶进行培养，并考察反应温度、时间、pH和溶剂等条件对单晶培养的影响。结果表明：在反应温度为80 ℃，反应时间为12 h、溶剂为蒸馏水、溶液pH=5的条件下可获得高质量单晶。该配体为单斜晶系，属于C2/c空间群，结构中含有4个氧原子和2个氮原子，可为金属离子的配位提供足够的位点，有望应用于配合物的设计或催化、吸附和医药载体等行业。     

Pentapnictogen heterocyclic monoanions: Structure,stability, and aromaticity

Inorganic planar ring-shape molecules with 4n + 2 π electrons are always the focus of experimental synthesis and theoretical research due to their potential aromaticity and stability. In this work, the whole series of five-membered heterocycle monoanions X _nY_5-n⁻ (X, Y = group 15 elements; n = 1-4) were thoroughly investigated by means of density functional theory calculations. They all have large formation energies and HOMO-LUMO gap energies, suggesting the potential thermodynamic and kinetic stability. Their aromaticities are comparable to that of typical aromatic hydrocarbons. Their thermal stabilities were firmly established by the ab initio molecular dynamics simulations. As most of them are predicted for the first time, their various spectra were simulated for experimental characterization. Furthermore, we demonstrate that these five-membered cyclic anions can be employed as η⁵-ligand to construct novel all-inorganic metallocenes, which may serve as the building blocks of low-dimensional nanomaterials.     

Synthesis and Self-Assembly of Mixed-Graft Block Copolymers

Mixed-graft block copolymers (mGBCPs) consist of two or more types of polymeric side chains grafted on a linear backbone in a random, alternating, or pseudo-alternating sequence. They can phase-separate with the backbone serving as the interface of the blocks, and the side chains dominate their self-assembly behavior. mGBCPs are an accessible polymer architecture for exploring the idea of encoding polymer properties through the macromolecular architecture, as there are two distinct structural components that can be tuned: the backbone and the side chains. In this Concept article, the current literature on the synthesis of mGBCPs is reviewed, and the advantages and disadvantages of each synthetic method are noted. The self-assembly of mGBCPs is also discussed where possible. Finally, directions for future research on mGBCP synthesis and self-assembly are suggested.     

一种单相串联混合电力有源滤波器的研究

文中对串联混合电力滤波器抑制谐波的原理进行了分析。为了检测单项系统的谐波电流,提出一种基于虚拟三相法和瞬时无功理论的谐波检测方法,并且对串联滤波逆变器的直流侧电压的控制进行了讨论。仿真结果证明了其有效性。     

Angular memory and frequency interferometry for mean heightprofiling of a rough surface

The polarimetric angular memory effect is applied to obtain the average topographic height of a rough surface. This novel effect improves the height sensitivity while maintaining a good degree of correlation between the sensors. By using a reference flat surface, the interferometric phase is linearly related to the mean topographic height. The combination of angular memory and wideband frequency interferometry (AMFI) is realized and offers a means to design a robust interferometric system. Extension of the technique to the pulse scattering problem is studied through the two-frequency mutual coherence function, and its time-domain transform provides an equivalent way to obtain the mean topographic height by combining both spatial and temporal diversity. Millimeter-wave (MMW) experiments are conducted with rough surfaces of different statistics and scattering media of different types (gravel, sand, and rough surfaces) to prove the effectiveness of the technique