Classification-Based Fraud Detection for Payment Marketing and Promotion

Nowadays, many payment service providers use the discounts and other marketing strategies to promote their products. This also raises the issue of people who deliberately take advantage of such promotions to reap financial benefits. These people are known as ‘scalper parties’ or ‘econnoisseurs’ which can constitute an underground industry. In this paper, we show how to use machine learning to assist in identifying abnormal scalper transactions. Moreover, we introduce the basic methods of Decision Tree and Boosting Tree, and show how these classification methods can be applied in the detection of abnormal transactions. In addition, we introduce a graph computing method, which implicitly describes the characteristics of people and merchants through node correlation, in order to mine deep features. Because of the volume of large data, we carried out reasonable block calculation, and succeeded in reducing a large amount of data to a series of segments, thereby decreasing the computational resources and memory requirements. Compared with other work on abnormal transaction detection, we pay more attention to creating and using the portraits of merchants or individuals to assist in decision-making. After data analysis and model building, we find that focusing on only one transaction or one day does not yield a comprehensive number of characteristics, and many characteristics can be obtained by examining the transactions of a person or a merchant over a period of time. Furthermore, a large number of characteristics can be obtained from transactions in a period of time. After GBDT (Gradient Boosting Decision Tree) based classification prediction and analysis, we can conclude that there is a clear distinction between abnormal trading shops and conventional shops, facilitating the clustering of abnormal merchants. By filtering transaction data from multiple dimensions, multiple sub-graphs can be obtained. After hierarchical clustering, the abnormal trading group is mined and classified according to its features. Finally, we build a scoring model and apply it to the big data platform of one of China’s largest payment service providers to help enterprises identify abnormal trading groups and specific marketing strategies.     

针对大视差图像拼接的显性子平面配准

目的 针对图像拼接中大视差图像难以配准的问题,提出一种显性子平面自动配准算法。方法 假设大视差图像包含多个显性子平面且每个平面内所含特征点密集分布。对该假设进行了验证性实验。所提算法以特征点分布为依据,通过聚类算法实现子平面分割,进而对子平面进行局部配准。首先,使用层次聚类算法对已匹配的特征点聚类,通过一种本文设计的拼接误差确定分组数目,并以各组特征点的聚类中心为新的聚类中心对重叠区域再聚类,分割出目标图像的显性子平面。然后,求解每个显性子平面的投影参数,并采用就近原则分配非重叠区域的单应性矩阵。结果 采用公共数据集对本文算法进行测试,并与Auto-Stitching、微软Image Composite Editor两种软件及全局投影拼接方法（Baseline）、尽可能投影算法（APAP）进行对比,采用均方根误差作为配准精度的客观评判标准。实验结果表明,该算法在拼接大视差图像时,能有效地配准局部区域,解决软件和传统方法由误配准引起的鬼影、错位等问题。其均方根误差比Baseline方法平均减小55%左右。与APAP算法相比,均方根误差平均相差10%左右,但可视化配准效果相同且无需调节复杂参数,可实现自动配准。结论 提出的显性子平面自动配准算法,通过分割图像所含子平面进而实现局部配准。该方法具有较高的配准精度,在大视差图像配准方面,优于部分软件及算法,可应用于图像拼接中大视差图像的自动配准。     

Euryhaline Hydrogel with Constant Swelling and Salinity-Enhanced Mechanical Strength in a Wide Salinity Range

In nature, many biological organisms possess a unique osmoregulation feature that enables them to survive in environments of different salinity, which is called euryhaline characteristics (e.g., salmon that can survive in freshwater and seawater). Drawing inspiration from these salinity-tolerant organisms, here a strategy that integrates two polymer chain segments with different salinity tolerances is reported to produce a euryhaline hydrogel with stable water retention, constant swelling properties, superoleophobicity, and low-adhesion to oil in aqueous environments over a wide range of salinity. The formation of internal dynamic complementary crosslinks is a key structural factor of the euryhaline attributes. The euryhaline hydrogel-coated meshes can be successfully utilized in various oil/aqueous phase separation in a wide range of salinity. Furthermore, by creating a double network with dynamic bonds, superior euryhaline hydrogel with unique salinity-enhanced mechanical strength can be obtained. It is anticipated that the euryhaline hydrogel will have broad application prospects in complex and variable ionic environments.     

Bioinspired Tumor Calcification Enables Early Detection and Elimination of Lung Cancer

Precise diagnosis of cancer in an early stage and treatments with a reliable response, high selectivity, and negligible side effects is urgently needed. However, current cancer management involves low-resolution metrics and delayed visual confirmation of tumor foci in imaging findings, and the toxicity of chemo- and radiotherapy unavoidably damages normal tissue and disrupts the immune balance of cancer patients. Here, a polypeptide is synthesized that preferentially targets lung cancer cells rather than normal lung epithelial cells and induces calcium precipitation specifically on the plasma membrane of lung cancer cells without additional supplementary calcium. Polypeptide-induced cellular calcification can ideally facilitate medical imaging for identifying early-stage lung cancer and distinguish cancer from benign nodules. Physiological and spontaneous calcification of tumors is induced by polypeptides and sharply prolongs the survival of tumor-bearing mice without evidence of systemic side effects. This tumor cell-selective calcification process provides an attractive, safe, and unprecedented approach for accurately visualizing and treating cancer in patients with early-stage disease in the clinic. It has broad implications in developing simple physiological reactions for diagnosing and treating cancer and provides a new horizon for drug discovery.     

基于水足迹的黄河流域生态补偿标准研究

流域生态补偿机制的建立为黄河流域生态保护与高质量发展提供了重要保障。基于水足迹视角,探讨了黄河流域九省(区)2011—2020年水资源利用情况,确定了生态补偿的主、客体及生态补偿标准。结果表明：黄河流域水足迹整体呈现增长态势,构成比例从大到小依次为农业、工业、生活、出口虚拟水、进口虚拟水足迹和生态水足迹;黄河流域生态补偿标准存在波动变化且为补偿主体,2015年补偿标准最高,为3 415.42×10⁸元,2020年最低,为12.75×10⁸元,补偿金额年均值为1 772.51×10⁸元。宁夏回族自治区、山西省、河南省和山东省为生态补偿的主体呈上升趋势,补偿金额年均值排名为河南省＞山东省＞山西省＞宁夏回族自治区。四川省是处于较快上升趋势的生态补偿客体,受补偿金额年均值为3 060.09×10⁸元;青海省为较慢上升趋势的生态补偿客体,补偿金额年均值为1 340.14×10⁸元。甘肃省是由生态补偿主体逐渐转变为客体;内蒙古自治区仅在2013年为生态补偿客体,其他年份则为补偿主体;陕西省生态补偿的主、客体身份不断发生变化。黄河流域应发挥区域优势推动农业高质量发展,建立完善的生态补偿机制,通过创新节水技术和提高用水效率等方式缓解水资源需求压力。     

固相研磨法制备AgI/Ag3PO4复合光催化剂及其光催化性能

采用固相研磨法，以Ag₃PO₄和KI为原料在研钵中研磨，机械化学效应作用下发生固相反应，在Ag₃PO₄表面包覆生成了AgI纳米粒子。改变研磨时间和反应物中KI的摩尔分数能够调控AgI/Ag₃PO₄复合光催化剂的组成和形貌。AgI/Ag₃PO₄复合光催化剂显示较强的可见光吸收性质，其光催化活性显著高于Ag₃PO₄或AgI。反应物中KI摩尔分数为10%，研磨时间为10min时制备得到的样品显示出最高的光催化活性，可见光照射1h，罗丹明B降解率达到99%，并具有较好的稳定性和循环利用效果。AgI壳层可以避免磷酸银的光腐蚀，提高了磷酸银在水中的结构稳定性。AgI的导带和价带位置比Ag₃PO₄更负，因此AgI的光生电子容易迁移到Ag₃PO₄表面，同时Ag₃PO₄产生的光生空穴可以迁移到AgI的价带上，降低了光生电子和空穴的复合，提高了光催化效率。     

一种可改善失步振荡模式的风电虚拟惯性控制策略

高渗透率风电并网导致原有系统等值惯性下降,给电力系统失步振荡后的稳定控制过程带来新的挑战。但是目前尚未有文献将风电虚拟惯性控制应用于电力系统发生失步振荡后的稳定控制领域。提出一种可改善失步振荡模式的风电虚拟惯性控制策略,通过引入风机并网点频率偏差来修改风机有功输出,系统遭受大扰动后,风电机组可持续向系统提供大量功率支援。风电机组通过虚拟惯性控制向系统注入/吸收功率,能有效抑制系统群内功率振荡,使得群内同调,从而改善系统失步振荡模式。仿真结果验证了所提虚拟惯性控制策略的正确性和有效性。     

Zero thermal expansion in cubic MgZrF6

Isotropic zero thermal expansion (ZTE) is rare but intriguing physical property in materials. Here, we report an isotropic ZTE property in a double ReO₃‐type compound of MgZrF₆, which exhibits a negligible value of coefficient of thermal expansion (α_l = ?7.94 × 10^?7 K^?1 (XRD), α_l = ?4.22 × 10^?7 K^?1 (dilatometry), 300‐675 K). The ZTE mechanism of MgZrF₆ is understood by the joint studies of temperature dependence of crystal structure and lattice dynamics. Interestingly, different magnitudes of atomic displacement parameters (ADPs) for the fluorine atoms in MZrF₆ (M = Ca, Ni, Mg) are found. The strong temperature sensitivity of ADPs demonstrates intensive transverse thermal vibration of fluorine atoms, which contributes essentially to the negative thermal expansion of CaZrF₆. By contrast, for NiZrF₆ with positive thermal expansion, the temperature response of ADPs is weak. Moderate transverse thermal vibration takes place in MgZrF₆, and ZTE appears. Furthermore, lattice dynamics of MgZrF₆ is studied by temperature‐dependent Raman spectroscopy, which reveals the ZTE mechanism. In particular, the F_2g and A_g modes, corresponding to the bending and stretching vibrations of fluorine atoms, respectively, neither soften nor harden over the whole temperature range, which is correlated with the isotropic ZTE property of MgZrF₆.     

Structure and properties of microwave transparent crosslinked polystyrene prepared through 3D printing bulk polymerization

Polystyrene (PS) crosslinked with divinylbenzene is one of the most promising microwave (MW) transparent polymers for structural application. However, preparation of it has been always inhibited by severe thermal runaway. Under this background, novel technology referred to as “3D printing bulk polymerization” is innovated with great advantages on heat release control. Comparing to the multistage batch polymerization as previously reported, the products obtained in present work are also superior on both the thermal–mechanical properties and dielectric properties at MW band. The modulus is improved from 2.43 to 3.06 GPa, and is increased to over 120 °C after crosslinking, while the ultralow dielectric loss feature of PS is well maintained with the loss factor to the order of 10^?4. In‐depth investigation about the effects of crosslinking on main‐chain segments has found it extend to secondary relaxation originates from pendent phenyl groups and resulted in a dispersion shift toward higher band. Based on the Raman study, a possible mechanism is also proposed that the torsional relaxation of side groups is governed by a potential barrier arises from constraint by main‐chain of several adjacent atoms through interaction between them. © 2017 Wiley Periodicals, Inc. J. Appl. Polym. Sci. 2017 , 134 , 44865.     

永磁同步电机双矢量模型预测磁链控制

针对传统模型预测转矩控制定子磁链和转矩脉动大、计算负担重、权重系数调节繁杂的问题,提出了一种基于定子磁链矢量跟踪误差最小化的双矢量模型预测磁链控制方法。首先,将传统模型预测中定子磁链和转矩误差的价值函数转化为对定子磁链跟踪误差的价值函数,避免了权重系数的整定;接着基于无差拍思想预测出参考电压矢量,减轻了处理器计算负担;然后为进一步改善系统稳定性,引入了双矢量占空比计算方法;最后,基于磁链矢量跟踪误差最小化原理,确定所选电压矢量的持续时间。通过实验验证所提控制策略的正确性与有效性,结果表明相较于传统模型预测转矩控制,所提出的双矢量模型预测磁链控制提高了永磁同步电机控制系统的动、稳态运行性能。