基于改进BiGRU-CNN的中文文本分类方法

传统的自注意力机制可以在保留原始特征的基础上突出文本的关键特征,得到更准确的文本特征向量表示,但忽视了输入序列中各位置的文本向量对输出结果的贡献度不同,导致在权重分配上存在偏离实际的情况,而双向门控循环单元（BiGRU）网络在对全局信息的捕捉上具有优势,但未考虑到文本间存在的局部依赖关系。针对上述问题,提出一种基于改进自注意力机制的BiGRU和多通道卷积神经网络（CNN）文本分类模型SAttBiGRU-MCNN。通过BiGRU对文本序列的全局信息进行捕捉,得到文本的上下文语义信息,利用优化的多通道CNN提取局部特征,弥补BiGRU忽视局部特征的不足,在此基础上对传统的自注意力机制进行改进,引入位置权重参数,根据文本向量训练的位置,对计算得到的自注意力权重概率值进行重新分配,并采用softmax得到样本标签的分类结果。在两个标准数据集上的实验结果表明,该模型准确率分别达到98.95%和88.1%,相比FastText、CNN、RCNN等分类模型,最高提升了8.99、7.31个百分点,同时精确率、召回率和F1值都有较好表现,取得了更好的文本分类效果。     

基于自适应方向导数滤波器的彩色边缘检测

传统彩色边缘检测算法在提高边缘检测准确性时可能将噪声检测为边缘,而在提高噪声鲁棒性时会将部分边缘当作噪声进行抑制,导致部分边缘信息丢失。为解决传统彩色边缘检测算法在边缘检测准确性与噪声鲁棒性之间的矛盾问题,提出一种基于自适应各向异性高斯方向导数（ANDD）的彩色边缘检测算法。通过彩色图像的微分自相关矩阵构建反映边缘类型的度量准则,以自适应地确定每个像素处ANDD滤波器的形状,从而准确提取不同类型的边缘特征,采用ANDD滤波器组对图像进行平滑处理,提取在三个通道上的ANDD特征。在此基础上,利用奇异值分解得到最优融合权值,并融合三个通道的ANDD特征,以增强彩色边缘强度。实验结果表明,该算法在无噪声和含噪声环境下的Pratt品质因子分别为0.849 6和0.791 4,与彩色Canny、RCMG-MM和FRPOS算法相比,在保持较高边缘检测准确率的同时具有较优的噪声鲁棒性。     

融合多通道CNN与BiGRU的字词级文本错误检测模型

文本校对是自然语言处理领域的重要分支。深度学习技术因强大的特征提取与学习能力被广泛应用于中文文本校对任务。针对现有中文文本错误检测模型忽略句子连续词间的局部信息、对于长文本的上下文语义信息提取不充分等问题,提出一种基于多通道卷积神经网络（CNN）与双向门控循环单元（BiGRU）的字词级文本错误检测模型。利用Word2vec向量化待检错文本,采用CNN挖掘待检错文本的局部特征,使用BiGRU学习待检错文本的上下文语义信息及长时依赖关系,并通过Softmax处理后输出文本分类结果以判断文本中是否含有字词错误,同时采取L2正则化和dropout策略防止模型过拟合。在SIGHAN2014和SIGHAN2015中文拼写检查任务数据集上的实验结果表明,与基于长短时记忆网络的文本错误检测模型相比,该模型的检错F₁值提升了3.01个百分点,具有更优的字词级文本错误检测效果。     

Parametric optimization of multiwalled carbon nanotube-assisted electric discharge machining of Al-10%SiCp metal matrix composite by response surface methodology

The demand for miniaturized products having a glossy surface or nano-level surface is increasing exponentially in automobile, aerospace, biomedical, and semiconductor industries. The mirror-like surface finish has generated a need to develop advanced machining processes. The addition of powder particle into electric discharge machining (EDM) oil is considered a promising technique to achieve surface integrity at the miniaturization level. In this research, the Al–10%SiC_p metal matrix composite (MMC) has been machined after mixing the appropriate amount of multiwalled carbon nanotubes (MWCNTs) into the EDM dielectric fluid. An advanced experimental setup has been designed and fabricated in the laboratory for conducting the experiments. This proposed technology is called nano powder mixed electric discharge machining (NPMEDM). The input parameters of NPMEDM are also optimized using central composite rotatable design (CCRD) based on response surface methodology (RSM) in order to obtain the best surface finish and material removal rate (MRR). The MRR has been increased by 38.22% and surface finish has been improved by 46.06% after mixing the MWCNTs into the EDM dielectric fluid. The results indicate that the combination of parameters A5, B5, C5, and D5 might have produced maximum MRR, whereas A1, B1, C1, and D3 have produced minimum surface roughness (SR).     

Scattering of elastic P- and SV-waves by a circular coated nanofiber based on Gurtin–Murdoch model of surface elasticity: Scattering cross section results

In this article, an analytical study of elastic P- and SV-wave scattering by a circular nanofiber is presented. The nanofiber is assumed to be surrounded by an inhomogeneous interphase layer, and Gurtin–Murdoch's model of surface elasticity is utilized to study the surface/interface effects in the regions between the fiber and interphase and also interphase and matrix. The simultaneous effects of surface elasticity and interphase inhomogeneity are considered here; by taking the inhomogeneous interphase to be composed of several sublayers, a transfer matrix approach is used to find the unknown field variables and, consequently, the scattering cross sections. The results indicate that considering the effects of surface elasticity and interphase inhomogeneity has a considerable impact on the calculated scattering cross sections.     

Analytical and numerical investigation of the stiffness matrix for edge‐cracked circular shafts

This paper examines two engineering methods of evaluating the stress intensity factors for cracked beams and bars subjected to a combined loading and proposes innovative formulations, as far as the circular cross section is concerned. Based on the definition of the stress intensity factors, the compliance matrix is determined as the inverse of the stiffness matrix, modelling the cracked section of a beam through a line‐spring approximation with interactive forces computed within fracture mechanics. A comparative evaluation of numerical predictions based on the proposed methods is also performed with methods available from the literature. Results for free vibration analyses of beams with transverse non‐propagating open cracks are presented and compared in order to estimate the accuracy and efficiency of the proposed methods, where a good agreement is generally found. More specifically, two different coupling effects are herein analysed for circular beams subjected to a combined bending, axial and shear loading, first, and a combined bending, shear and torsion loading, subsequently.     

Effects of microstructural heterogeneity on very high cycle fatigue properties of 7050-T7451 aluminum alloy friction stir butt welds

In this study, the very high cycle fatigue (VHCF) properties of 7050-T7451 aluminum alloy and its friction stir welding (FSW) butt welds have been investigated. The results show that the failure of FSW joints still occurs at 7.0 × 10⁸ cycles. The fatigue properties of the FSW joints are superior to those of the base material, especially in the super long life regime. Most fatigue cracks initiate at the thermo-mechanically affected zone and heat affected zone on the advancing side of the FSW joints, and the susceptibility of these zones to fatigue is attributed to the metallurgical heterogeneity.     

Drug self-assembly: A phenomenon at the nanometer scale with major impact in the structure–biological properties relationship and the treatment of disease

Under water-rich conditions, small amphiphilic and hydrophobic drug molecules self-assemble into supramolecular nanostructures. Thus, substantial modifications in their interaction with cellular structures and the ability to reach intracellular targets could happen. Additionally, drug aggregates could be more toxic than the non-aggregated counterparts, or vice versa. Moreover, since self-aggregation reduces the number of effective “monomeric” molecules that interact with the target, the drug potency could be underestimated. In other cases, the activity could be ascribed to the non-aggregated molecule while it stems from its aggregates. Thus, drug self-assembly could mislead from drug throughput screening assays to advanced preclinical and clinical trials. Finally, aggregates could serve as crystallization nuclei. The impact that this phenomenon has on the biological performance of active compounds, the inconsistent and often controversial nature of the published data and the need for recommendations/guidelines as preamble of more harmonized research protocols to characterize drug self-aggregation were main motivations for this review. First, the key molecular and environmental parameters governing drug self-aggregation, the main drug families for which this phenomenon and the methods used for its characterization are described. Then, promising nanotechnology platforms investigated to prevent/control it towards a more efficient drug development process are briefly discussed.     

Improved robust exponential stability for Takagi–Sugeno fuzzy uncertain systems with time-varying delays

This paper concerns managing the robust exponential stability problem of uncertain Takagi–Sugeno fuzzy systems with time-varying delay by employing a further improved integral inequality matrix approach. Based on the linear matrix inequality (LMI) approach, delay-dependent robust exponential stability criteria have been developed. By taking the relationship among the time-varying delay, its upper bound and their difference into account, some less conservative LMI-based delay-dependent robust exponential stability criteria are obtained without ignoring any useful terms in the derivative of Lyapunov–Krasovskii functionals. Maximum allowable upper bound for time-varying delays is determined. Numerical examples are provided to show that the obtained results significantly improve the allowed upper bounds of delay size over some methods existing in the literature.     

The use of mindmapping software for patent search and management

Modern information technologies and in particular the internet have revolutionized the patent information professionals' work in terms of speed of access and information comprehensiveness from both company internal and external digital sources. Here, I describe how the digital mindmapping technique can be used to complement existing intellectual property management software solutions to meet the challenge of optimizing and managing patent search workflows as well as to rapidly organise and access highly dynamic, heterogeneous and scattered patent information sources. Both eye catching and highly memorable and at the same time self-explanatory mindmapping examples are presented. These were designed to include basic and advanced level digital mindmapping features tailored to significantly speed up and maintain a high work quality level of patent search professionals. A special emphasis is put on the great benefit of organising and accessing the plethora of internet-based worldwide online patent registers through mindmapping, both in terms of managing the constantly changing deep links to the actual search options for e.g., legal status information, and keeping track of the offered level of content.