基于近邻分类的增量学习分类算法研究

为解决分类器学习新样本知识的问题,提出一种基于近邻算法的增量学习算法。该算法以最近邻算法为基础,首先计算新样本与标准样本之间的匹配度,找到最佳匹配样本和次佳匹配样本,然后通过与匹配度阈值进行比较来决定是类内学习还是类别学习。算法采用UCI中的标准数据集进行实验并应用于车辆识别仿真,其结果验证了该算法的有效性。实验进一步研究了匹配度阈值的选择和初始化样本数量选取对分类正确率影响。     

基于图模型与加权损失策略的视频行人重识别研究

针对监控视频中行人外观、姿态相似等现象导致的视频行人重识别准确率低的问题进行了研究,提出了一种基于图模型的视频行人重识别方法,有效利用了视频中的时序信息,实现跨帧及帧内区域的信息交互。具体来说,利用跨帧分块区域间的关联信息建立区域节点间的固有关系,并进行特征传播迭代更新区域信息。另一方面,在度量学习过程中,提出了一种加权损失函数策略,这个方法将先前挖掘策略中的二进制分配法(即丢弃或保留该样本)优化为连续分数分配法,解决了可用样本未被有效利用的问题。将模型在MARS和DukeMTMC-VideoReID两个数据集上进行了评估,实验结果证实了提出方法的有效性。     

Master data management for manufacturing big data: a method of evaluation for data network

In the process of manufacturing, a large amount of manufacturing data is produced by different departments and different domain. In order to realise data sharing and linkage among supply chains, master data management method has been used. Through master data management, the key data can be shared and distributed uniformly. However, since these cross-domain data form a data network through the association of master data, how to evaluate the effectiveness and rationality of this network becomes the major issue in the proposed method. In this paper, a model of the master data network is built based on the theory of set pair analysis. In order to verify the master data, an evaluation method for the network is proposed. Finally, a case was presented to validate this network model and evaluation method.

     

A localization method for stagnant water in city road traffic image

Stagnant water on roads has always been a major cause of traffic jams and accidents. Traditional urban waterlogging monitoring and warning system is mainly based on a large amount of historical data and predictive network, which has low accuracy and weak generalization ability. Considering the deep neural network algorithms have demonstrated strong capabilities in computer vision tasks such as object detection, we aim to apply them to road stagnant water detection. In this paper, a novel automatic stagnant water localization method under weak supervision based on visual image is proposed. First, the template matching method is applied to extract road information from the traffic image. Then, due to the complexity of data annotation, we locate stagnant water in image based on Class Activation Maps (CAM) mechanism, which is a weakly supervised method. The detection model consists of the ResNet-18 and the Grad-CAM++ mechanism. Finally, based on the heat map and template, we set a suitable threshold to segment stagnant water area in image. In the experiments, the precision and recall for road stagnant water classification by the proposed model are 99.39% and 99.60%, while the Intersection over Union (IoU) for stagnant water area segmentation is up to 63%. These show that our method is effective for road stagnant water localization.

     

Microstructure and Properties of Hydroxyapatite-Containing Ceramic Coatings on Magnesium Alloys by One-Step Micro-Arc Oxidation

Protection of Metals and Physical Chemistry of Surfaces - Mg and its alloys are considered as biodegradable metallic implant materials, are attracting great attentions. However, the corrosion...     

Coherent Sub-Nanometer Interface between Crystalline and Amorphous Materials Boosts Electrochemical Synthesis of Hydrogen Peroxide

There are enormous yet largely underexplored exotic phenomena and properties emerging from interfaces constructed by diverse types of components that may differ in composition, shape, or crystal structure. It remains poorly understood the unique properties a coherent interface between crystalline and amorphous materials may evoke, and there lacks a general strategy to fabricate such interfaces. It is demonstrated that by topotactic partial oxidation heterostructures composed of coherently registered crystalline and amorphous materials can be constructed. As a proof-of-concept study, heterostructures consisting of crystalline P₃N₅ and amorphous P₃N₅O_x can be synthesized by creating amorphous P₃N₅O_x from crystalline P₃N₅ without interrupting the covalent bonding across the coherent interface. The heterostructure is dictated by nanometer-sized short-range-ordered P₃N₅ domains enclosed by amorphous P₃N₅O_x matrix, which entails simultaneously fast charge transfer across the interface and bicomponent synergistic effect in catalysis. Such a P₃N₅/P₃N₅O_x heterostructure attains an optimal adsorption energy for *OOH intermediates and exhibits superior electrocatalytic performance toward H₂O₂ production by adopting a selectivity of 96.68% at 0.4 V_RHE and a production rate of 321.5 mmol h⁻¹ g_catalyst⁻¹ at −0.3 V_RHE. The current study provides new insights into the synthetic strategy, chemical structure, and catalytic property of a sub-nanometer coherent interface formed between crystalline and amorphous materials.     

A person re-identification algorithm based on pyramid color topology feature

Multimedia Tools and Applications - Due to illumination variations, person re-identification algorithms based on color features are not robust in practical applications. Different persons may have...     

多比特数字信号频率测量方法

文中提出一种可直接测量多比特数字信号频率的方法，叙述了基本原理及实施方法。此方法的特点是实施简单，成本低廉、精度高。当信号频率较高且量化的比特数较多时，上述优越性别明显。     

数字式自动衰减器设计

文中介绍以数控衰减器ＡＤ７１１８为核心配以衰减控制码电路组成数字式自动衰减器的设计。自动衰减器量根据输入信号电平高低自动调节并具有衰减分贝值的数字显示及数字输出。