基于快速神经网络架构搜索的鲁棒图像水印网络算法

为解决深度学习在图像水印算法中计算量大且模型冗余的问题，提高图像水印算法在抵抗噪声、旋转和剪裁等攻击时的鲁棒性，提出基于快速神经网络架构搜索（neural architecture search，NAS）的鲁棒图像水印网络算法。通过多项式分布学习快速神经网络架构搜索算法，在预设的搜索空间中搜索最优网络结构，进行图像水印的高效嵌入与鲁棒提取。首先，将子网络中线性连接的全卷积层设置为独立的神经单元结构，并参数化表示结构单元内节点的连接，预先设定结构单元内每个神经元操作的搜索空间；其次，在完成一个批次的数据集训练后，依据神经元操作中的被采样次数和平均损失函数值动态更新概率；最后，重新训练搜索完成的网络。水印网络模型的参数量较原始网络模型缩减了92%以上，大大缩短了模型训练时间。由于搜索得到的网络结构更为紧凑，本文算法具有较高的时间性能和较好的实验效果，在隐藏图像时，对空域信息的依赖比原始网络更少。对改进前后的2个网络进行了大量鲁棒性实验，对比发现，本文算法在CIFAR-10数据集上对抵抗椒盐噪声和旋转、移除像素行（列）等攻击优势显著；在ImageNet数据集上对抵抗椒盐高斯噪声、旋转、中值滤波、高斯滤波、JPEG压缩、裁剪等攻击优势显著，特别是对随机移除行（列）和椒盐噪声有较强的鲁棒性。     

An efficient edge based data structure for the compressible Reynolds-averaged Navier–Stokes equations on hybrid unstructured meshes

An efficient edge based data structure has been developed in order to implement an unstructured vertex based finite volume algorithm for the Reynolds-averaged Navier–Stokes equations on hybrid meshes. In the present approach, the data structure is tailored to meet the requirements of the vertex based algorithm by considering data access patterns and cache efficiency. The required data are packed and allocated in a way that they are close to each other in the physical memory. Therefore, the proposed data structure increases cache performance and improves computation time. As a result, the explicit flow solver indicates a significant speed up compared to other open-source solvers in terms of CPU time. A fully implicit version has also been implemented based on the PETSc library in order to improve the robustness of the algorithm. The resulting algebraic equations due to the compressible Navier–Stokes and the one equation Spalart–Allmaras turbulence equations are solved in a monolithic manner using the restricted additive Schwarz preconditioner combined with the FGMRES Krylov subspace algorithm. In order to further improve the computational accuracy, the multiscale metric based anisotropic mesh refinement library PyAMG is used for mesh adaptation. The numerical algorithm is validated for the classical benchmark problems such as the transonic turbulent flow around a supercritical RAE2822 airfoil and DLR-F6 wing-body-nacelle-pylon configuration. The efficiency of the data structure is demonstrated by achieving up to an order of magnitude speed up in CPU times.     

游客情感计算的文本大数据挖掘方法比较研究

旅游文本大数据以其方便、快捷和低门槛的特点为游客情感计算提供了极大便利，已经成为旅游大数据的主要来源之一。基于大数据理论和情感理论，以文本大数据为数据源，在全面梳理国内外情感计算相关成果的基础上，利用人工智能中的逻辑/算法编程方法、机器学习方法、深度学习方法对旅游文本大数据进行挖掘，探索最佳的基于文本大数据的游客情感计算方法。研究发现：（1）基于情感词典的游客情感计算模型，其核心是构建情感词典和设计情感计算规则，方法简单，容易实现，适用语料范围广。（2）机器学习，用统计学方法抽取文本中的特征项，具有非线性特征，可靠性较线性特征的情感词典方法高。（3）基于深度学习技术的游客情感计算，效果良好，准确率在85%以上。训练多领域的文本语料易于移植，实用性强，且泛化能力好，较适合大数据时代游客情感计算研究。     

Simultaneous determination of cinnamaldehyde and its metabolite in rat tissues by gas chromatography–mass spectrometry

Cinnamaldehyde (CA), an active ingredient isolated from the traditional Chinese medicine Cortex Cinnamomi, has a wide range of bioactivities. To clarify the distribution characteristics of CA, a selective and sensitive method utilizing gas chromatography–mass spetrometry was initially developed for simultaneously determining the concentration of CA and its metabolite cinnamyl alcohol in rat tissues. Selected ion masses of m/z 131, 105 and 92 were chosen, and separation of the analytes was performed on a DB‐5 ms (30 m × 0.25 mm, 0.25 µm, thickness) capillary column by gas chromatography–mass spectrometry. The calibration curves demonstrated good linearity and reproducibility over the range of 20–2000 and 20–4000 ng/mL for various tissue samples. Recoveries ranged from 86.8 to 107.5%, while intra‐ and interday relative standard deviations were all <11.3%. The analysis method was successfully applied in tissue distribution studies for CA and cinnamyl alcohol. As CA and cinnamyl alcohol may inter‐convert to one another, simultaneous determination of both analytes provides a comparative and accurate data for tissue study. The concentrations of CA and cinnamyl alcohol remaining in spleen were the highest among the main organs, including heart, liver, spleen, lung, kidney and brain. In addition, there was no long‐term accumulation of CA in rat tissues. Copyright © 2014 John Wiley & Sons, Ltd.     

Lithium Insertion Mechanism in Iron-Based Oxyfluorides with Anionic Vacancies Probed by PDF Analysis

The mechanism of lithium insertion that occurs in an iron oxyfluoride sample with a hexagonal–tungsten–bronze (HTB)-type structure was investigated by the pair distribution function. This study reveals that upon lithiation, the HTB framework collapses to yield disordered rutile and rock salt phases followed by a conversion reaction of the fluoride phase toward lithium fluoride and nanometer-sized metallic iron. The occurrence of anionic vacancies in the pristine framework was shown to strongly impact the electrochemical activity, that is, the reversible capacity scales with the content of anionic vacancies. Similar to FeOF-type electrodes, upon de-lithiation, a disordered rutile phase forms, showing that the anionic chemistry dictates the atomic arrangement of the re-oxidized phase. Finally, it was shown that the nanoscaling and structural rearrangement induced by the conversion reaction allow the in situ formation of new electrode materials with enhanced electrochemical properties.     

均匀磁场下碳离子笔形束的剂量变化分析及位置修正方法

分析外加均匀磁场对于碳离子笔形束剂量分布的影响,并考虑修正这种影响,为磁共振成像引导碳离子放射治疗的临床应用提供指导。本文利用蒙特卡罗方法模拟计算了不同能量碳离子笔形束在不同强度磁场下的剂量分布情况,发现垂直于碳离子束入射方向的均匀磁场对于碳离子笔形束射程缩短的影响很小,磁场对碳离子束的主要影响是引起束流横向偏转,特别是碳离子束布拉格峰位置的横向侧移。横向侧移程度与碳离子束的能量和磁场强度相关,根据模拟结果,得到了一个计算碳离子束布拉格峰在磁场中相对横向偏转的方程,并提出一种校正外加磁场引起的碳离子束布拉格峰横移的角度修正方法。这些结果可用于指导磁共振图像引导碳离子放射治疗计划系统的研发。     

Definition and detection of data-based uniqueness in evaluating bilinear (two-way) chemical measurements

Multivariate curve resolution methods, frequently used in analyzing bilinear data sets, result in ambiguous decomposition in general. Implementing the adequate constraints may lead to reduce the so-called rotational ambiguity drastically, and in the most favorable cases to the unique solution. However, in some special cases, non-negativity constraint as minimal information of the system is a sufficient condition to resolve profiles uniquely. Although, several studies on exploring the uniqueness of the bilinear non-negatively constrained multivariate curve resolution methods have been made in the literature, it has still remained a mysterious question. In 1995, Manne published his profile-based theorems giving the necessary and sufficient conditions of the unique resolution. In this study, a new term, i.e., data-based uniqueness is defined and investigated in details, and a general procedure is suggested for detection of uniquely recovered profile(s) on the basis of data set structure in the abstract space. Close inspection of Borgen plots of these data sets leads to realize the comprehensive information of local rank, and these argumentations furnish a basis for data-based uniqueness theorem. The reported phenomenon and its exploration is a new stage (it can be said fundament) in understanding and describing the bilinear (matrix-type) chemical data in general.     

Semilinear wave models with friction and viscoelastic damping

In this paper, we study the global (in time) existence of small data solutions to the Cauchy problem for the semilinear wave equation with friction, viscoelastic damping, and a power nonlinearity. We are interested in the connection between regularity assumptions for the data and the admissible range of exponents p in the power nonlinearity.     

Spectral estimates for saddle point matrices arising in weak constraint four‐dimensional variational data assimilation

We consider the large sparse symmetric linear systems of equations that arise in the solution of weak constraint four‐dimensional variational data assimilation, a method of high interest for numerical weather prediction. These systems can be written as saddle point systems with a 3 × 3 block structure but block eliminations can be performed to reduce them to saddle point systems with a 2 × 2 block structure, or further to symmetric positive definite systems. In this article, we analyse how sensitive the spectra of these matrices are to the number of observations of the underlying dynamical system. We also obtain bounds on the eigenvalues of the matrices. Numerical experiments are used to confirm the theoretical analysis and bounds.     

Enhancing signal-to-noise ratio and resolution in low-field NMR relaxation measurements using post-acquisition digital filters

The traditional way to enhance signal-to-noise ratio (SNR) of nuclear magnetic resonance (NMR) signals is to increase the number of scans. However, this procedure increases the measuring time that can be prohibitive for some applications. Therefore, we have tested the use of several post-acquisition digital filters to enhance SNR up to one order of magnitude in time domain NMR (TD-NMR) relaxation measurements. The procedures were studied using continuous wave free precession (CWFP-T₁) signals, acquired with very low flip angles that contain six times more noise than the Carr–Purcell–Meiboom–Gill (CPMG) signal of the same sample and experimental time. Linear (LI) and logarithmic (LO) data compression, low-pass infinity impulse response (LP), Savitzky–Golay (SG), and wavelet transform (WA) post-acquisition filters enhanced the SNR of the CWFP-T₁ signals by at least six times. The best filters were LO, SG, and WA that have high enhancement in SNR without significant distortions in the ILT relaxation distribution data. Therefore, it was demonstrated that these post-acquisition digital filters could be a useful way to denoise CWFP-T₁, as well as CPMG noisy signals, and consequently reducing the experimental time. It was also demonstrated that filtered CWFP-T₁ method has the potential to be a rapid and nondestructive method to measure fat content in beef and certainly in other meat samples.