Learning compact binary code based on multiple heterogeneous features

A novel hashing method based on multiple heterogeneous features is proposed to improve the accuracy of the image retrieval system. First, it leverages the imbalanced distribution of the similar and dissimilar samples in the feature space to boost the performance of each weak classifier in the asymmetric boosting framework. Then, the weak classifier based on a novel linear discriminate analysis （LDA） algorithm which is learned from the subspace of heterogeneous features is integrated into the framework. Finally, the proposed method deals with each bit of the code sequentially, which utilizes the samples misclassified in each round in order to learn compact and balanced code. The heterogeneous information from different modalities can be effectively complementary to each other, which leads to much higher performance. The experimental results based on the two public benchmarks demonstrate that this method is superior to many of the state- of-the-art methods. In conclusion, the performance of the retrieval system can be improved with the help of multiple heterogeneous features and the compact hash codes which can be learned by the imbalanced learning method.

Hierarchical Optimization Method for Federated Learning with Feature Alignment and Decision Fusion

In the realm of data privacy protection,federated learning aims to collaboratively train a global model.However,heterogeneous data between clients presents challenges,often resulting in slow convergence and inadequate accuracy of the global model.Utilizing shared feature representations alongside customized classifiers for individual clients emerges as a promising personalized solution.Nonetheless,previous research has frequently neglected the integration of global knowledge into local representation learning and the synergy between global and local classifiers,thereby limiting model performance.To tackle these issues,this study proposes a hierarchical optimization method for federated learning with feature alignment and the fusion of classification decisions(FedFCD).FedFCD regularizes the relationship between global and local feature representations to achieve alignment and incorporates decision information from the global classifier,facilitating the late fusion of decision outputs from both global and local classifiers.Additionally,FedFCD employs a hierarchical optimization strategy to flexibly optimize model parameters.Through experiments on the Fashion-MNIST,CIFAR-10 and CIFAR-100 datasets,we demonstrate the effectiveness and superiority of FedFCD.For instance,on the CIFAR-100 dataset,FedFCD exhibited a significant improvement in average test accuracy by 6.83%compared to four outstanding personalized federated learning approaches.Furthermore,extended experiments confirm the robustness of FedFCD across various hyperparameter values.

Spatial Heterogeneity Modeling Using Machine Learning Based on a Hybrid of Random Forest and Convolutional Neural Network (CNN)

Spatial heterogeneity refers to the variation or differences in characteristics or features across different locations or areas in space. Spatial data refers to information that explicitly or indirectly belongs to a particular geographic region or location, also known as geo-spatial data or geographic information. Focusing on spatial heterogeneity, we present a hybrid machine learning model combining two competitive algorithms: the Random Forest Regressor and CNN. The model is fine-tuned using cross validation for hyper-parameter adjustment and performance evaluation, ensuring robustness and generalization. Our approach integrates Global Moran’s I for examining global autocorrelation, and local Moran’s I for assessing local spatial autocorrelation in the residuals. To validate our approach, we implemented the hybrid model on a real-world dataset and compared its performance with that of the traditional machine learning models. Results indicate superior performance with an R-squared of 0.90, outperforming RF 0.84 and CNN 0.74. This study contributed to a detailed understanding of spatial variations in data considering the geographical information (Longitude & Latitude) present in the dataset. Our results, also assessed using the Root Mean Squared Error (RMSE), indicated that the hybrid yielded lower errors, showing a deviation of 53.65% from the RF model and 63.24% from the CNN model. Additionally, the global Moran’s I index was observed to be 0.10. This study underscores that the hybrid was able to predict correctly the house prices both in clusters and in dispersed areas.

Heterogeneous data-driven aerodynamic modeling based on physical feature embedding

Aerodynamic surrogate modeling mostly relies only on integrated loads data obtained from simulation or experiment,while neglecting and wasting the valuable distributed physical information on the surface.To make full use of both integrated and distributed loads,a modeling paradigm,called the heterogeneous data-driven aerodynamic modeling,is presented.The essential concept is to incorporate the physical information of distributed loads as additional constraints within the end-to-end aerodynamic modeling.Towards heterogenous data,a novel and easily applicable physical feature embedding modeling framework is designed.This framework extracts lowdimensional physical features from pressure distribution and then effectively enhances the modeling of the integrated loads via feature embedding.The proposed framework can be coupled with multiple feature extraction methods,and the well-performed generalization capabilities over different airfoils are verified through a transonic case.Compared with traditional direct modeling,the proposed framework can reduce testing errors by almost 50%.Given the same prediction accuracy,it can save more than half of the training samples.Furthermore,the visualization analysis has revealed a significant correlation between the discovered low-dimensional physical features and the heterogeneous aerodynamic loads,which shows the interpretability and credibility of the superior performance offered by the proposed deep learning framework.

Event-based Two-stage Non-intrusive Load Monitoring Method Involving Multi-dimensional Features

This paper proposes an event-based two-stage Nonintrusive load monitoring(NILM)method involving multidimensional features,which is an essential technology for energy savings and management.First,capture appliance events using a goodness of fit test and then pair the on-off events.Then the multi-dimensional features are extracted to establish a feature library.In the first stage identification,several groups of events for the appliance have been divided,according to three features,including phase,steady active power and power peak.In the second stage identification,a“one against the rest”support vector machine(SVM)model for each group is established to precisely identify the appliances.The proposed method is verified by using a public available dataset;the results show that the proposed method contains high generalization ability,less computation,and less training samples.

Research on Privacy Disclosure Detection Method in Social Networks Based on Multi-Dimensional Deep Learning

In order to effectively detect the privacy that may be leaked through social networks and avoid unnecessary harm to users,this paper takes microblog as the research object to study the detection of privacy disclosure in social networks.First,we perform fast privacy leak detection on the currently published text based on the fastText model.In the case that the text to be published contains certain private information,we fully consider the aggregation effect of the private information leaked by different channels,and establish a convolution neural network model based on multi-dimensional features(MF-CNN)to detect privacy disclosure comprehensively and accurately.The experimental results show that the proposed method has a higher accuracy of privacy disclosure detection and can meet the real-time requirements of detection.

Multi-dimensional and Multi-threshold Airframe Damage Region Division Method Based on Correlation Optimization

In order to obtain the image of airframe damage region and provide the input data for aircraft intelligent maintenance,a multi-dimensional and multi-threshold airframe damage region division method based on correlation optimization is proposed.On the basis of airframe damage feature analysis,the multi-dimensional feature entropy is defined to realize the full fusion of multiple feature information of the image,and the division method is extended to multi-threshold to refine the damage division and reduce the impact of the damage adjacent region’s morphological changes on the division.Through the correlation parameter optimization algorithm,the problem of low efficiency of multi-dimensional multi-threshold division method is solved.Finally,the proposed method is compared and verified by instances of airframe damage image.The results show that compared with the traditional threshold division method,the damage region divided by the proposed method is complete and accurate,and the boundary is clear and coherent,which can effectively reduce the interference of many factors such as uneven luminance,chromaticity deviation,dirt attachment,image compression,and so on.The correlation optimization algorithm has high efficiency and stable convergence,and can meet the requirements of aircraft intelligent maintenance.

Robust signal recognition algorithm based on machine learning in heterogeneous networks

There are various heterogeneous networks for terminals to deliver a better quality of service. Signal system recognition and classification contribute a lot to the process. However, in low signal to noise ratio（SNR） circumstances or under time-varying multipath channels, the majority of the existing algorithms for signal recognition are already facing limitations. In this series, we present a robust signal recognition method based upon the original and latest updated version of the extreme learning machine（ELM） to help users to switch between networks. The ELM utilizes signal characteristics to distinguish systems. The superiority of this algorithm lies in the random choices of hidden nodes and in the fact that it determines the output weights analytically, which result in lower complexity. Theoretically, the algorithm tends to offer a good generalization performance at an extremely fast speed of learning. Moreover, we implement the GSM/WCDMA/LTE models in the Matlab environment by using the Simulink tools. The simulations reveal that the signals can be recognized successfully to achieve a 95% accuracy in a low SNR（0 dB） environment in the time-varying multipath Rayleigh fading channel.

基于空间特征交叉融合的轻量级图像超分辨率重建

近几年,深度学习技术显著提高了单幅图像超分辨率重建(Single Image Super-resolution Reconstruction,SISR)的性能,但基于深度学习技术的SISR算法存在模型参数量大、网络结构复杂、资源消耗多等问题。为解决这些问题,提出一种基于空间特征交叉融合的轻量级图像超分辨率重建算法,该算法使用多个局部特征融合模块和特征交叉增强模块组成非线性映射单元,通过残差学习逐步聚合图像特征,提取更加精准的残差信息。同时采用对称结构将特征映射到两个分支,通过执行特征交叉,对应元素相乘提取高频成分,细化特征,增加网络非线性。在每个特征交叉增强模块中使用异构卷积代替标准卷积拆分和融合两条分支,有效地降低网络的参数量,使网络在参数量和性能之间达到相对平衡。通过一个多级集成模块增强不同阶段特征的相关性。在基准数据集上的实验结果表明,新的重建算法在降低模型参数量的同时,峰值信噪比和结构相似度均取得了较好的结果,而且重建图像的边缘结构完整,整体轮廓清晰,细节更加丰富。

使用异质集成学习和心电信号异构特征融合的睡眠呼吸暂停分类方法

睡眠呼吸暂停(SA)会影响睡眠质量,增加心脑血管疾病风险,其准确分类有助于在SA早期阶段及时开展针对性治疗。本文提出一种使用异质集成学习和异构特征融合的SA分类新方法。首先从原始心电信号中提取小波时频谱,使用SE-ResNet作为初级分类器;然后提取RR间期序列和R峰值序列,使用1D CNN-LSTM作为初级分类器;再提取心率变异性特征,使用SVM作为初级分类器。最后采用堆叠法作为异质集成学习的融合策略,再使用另一个SVM作为次级分类器实现SA分类。在Apnea-ECG数据集上进行实验,所提出的SA分类方法的准确率为89.12%。实验结果表明,所提方法有效利用了各初级分类器的多样性和异构特征的互补性,其性能优于传统的SA分类方法。

基于特征匹配度与异类子模型融合的安全性评估方法

机器学习模型的好坏影响预测精度、输入与输出结果的拟合情况。在复杂系统中,使用单一模型评估系统安全性问题时容易受数据量、数据格式、模型结构以及环境干扰等因素影响,使得这个模型在解决某个问题的能力上比较出色,而在解决其他问题时,结果却不尽如人意。针对上述问题,提出一种基于特征匹配度和异类子模型融合的安全性评估方法。首先,按照采样数据的输出值划分不同规模的数据集并构建子模型;其次,通过计算每个新数据对于这些子模型的匹配度,进而得到每个子模型的权重;最后,根据权重大小融合所有子模型的子输出得到最终的多模型融合结果。所提方法对山东省济宁市霄云煤矿采掘数据集进行研究,实验结果表明,该方法与多样本单模型、少样本单模型和传统多模型方法相比,在以330/70的比例来构建子模型的情况下均方根误差(RMSE)分别降低了15.13%、51.67%和12.46%,该方法充分集成各子模型所能提供的有效信息,减少和分散单一模型的预测误差,以提高模型的预测精度和泛化能力。

多源异构数据和注意力门控机制的小麦产量预测

针对传统的单模态数据预测小麦产量存在精度不高的问题,提出一种结合多源异构数据和注意力门控机制的小麦产量预测方法。首先引入了特征级的门控策略,来捕获每个模态内部特征的信息变化;然后利用神经网络评估每个模态内的置信度分数,并构建模态间的有效信息获取模块;最后设计了基于Transformer的空间和通道注意力门控机制模块,将不同模态之间的有效信息进行充分的融合,从而获得最佳的预测特征表示。实验结果表明,所提方法与传统方法相比具有更高的预测精准度,RMSE和MAE分别仅为809kg/hm^(2)和522kg/hm^(2),R^(2)则达到了0.806,通过对河南省近10年的小麦产量进行预测,得到的三项评价指标均相对稳定,且展现出了较强的鲁棒性。消融实验也验证了该方法中的不同组件均能有效提高小麦产量的预测精度,可为相关部门制定保障粮食安全管理决策提供有力的数据支持。

基于“十字”标志物的红外图像与三维点云融合方法

红外热成像技术广泛应用于多个领域,建立含有空间和温度信息的三维温度场模型具有十分重要的意义,可以将该技术扩展到更多应用领域.为此,本文提出一种异源空间数据融合方法,融合红外图像和三维点云,得到三维温度场模型.针对红外相机与可见光相机成像原理存在差异,难以通过常用标定板进行内参标定的问题,基于红外相机成像特性设计并制作镂空圆孔标定板用于内参标定,所得内参平均重投影误差为0.03像素.针对红外相机与结构光相机的成像原理不同,现有标志物制作复杂、外参精度低的问题,基于不同材料的辐射度差异,设计制作“十字”标志物并将其用于联合标定.为解决同名特征点难以识别的问题,针对红外图像和三维点云分别设计了同名特征点提取方法,配合“十字”标志物进行同名特征点提取.红外图像和三维点云特征点提取方法的检测重复率分别为75%和92%,与传统方法相比两者的检测重复率均有所提升.利用该方法建立纸杯、工件和人脸的三维温度场模型.实验结果表明,使用镂空圆孔标定板能实现红外相机的内参标定,对“十字”标志物采用同名特征点提取方法能完成红外相机与结构光相机的联合标定.最终所得三维温度场模型的平均重投影误差为1.70像素,与现有方法相比模型精度有所提升.

基于多元语义特征和图卷积神经网络的短文本分类模型

在互联网和社交媒体迅猛发展的背景下,网络中出现了大量的短文本数据,这些数据在舆情监控、情感分析和新闻分类等自然语言处理领域展现出了非常高的经济和学术价值.但短文本数据固有的特征给短文本分类带来了不小的挑战,如文本稀疏和缺乏丰富的上下文语义等.针对这些问题,提出了一种结合多元语义特征和图卷积神经网络(GCN)的短文本分类模型,该模型通过哈尔滨工业大学的语言技术平台获取短文本的多种语义特征,并将这些语义特征同短文本一起构建一个多元异构图,然后将其作为GCN的输入,利用GCN学习短文本更深层特征,最后通过Softmax函数获取每个类别的概率分布,进而实现短文本分类.试验结果表明,本模型在短文本分类的F1评分上比传统单一模型提高了4%.

基于激光扫描的电子档案异构时态数据检测方法

研究一种基于激光扫描的电子档案异构时态数据检测方法,该方法利用激光扫描获取数据并生成电子档案。针对电子档案中的数据实施缺失数据填补和离群数据处理。提取电子档案数据的递归率、仙农熵以及歪度三个特征。结合随机森林算法,输入递归率、仙农熵以及歪度三个特征,得出电子档案异构时态数据类型。结果表明:1处和3处边坡工程的位移处于轻微状态;2处边坡工程的位移处于严重状态;4处边坡工程的位移处于安全状态;5处边坡工程的位移处于较严重状态。所研究方法的Kappa系数达到相对最大值,说明所研究方法的检测能力更高。

多源异构数据渐进式融合的虚假新闻检测

社交媒体平台上充斥着大量未经验证的信息,这些信息大多为不同来源的异构数据,其传播范围之广、速度之快,对个人和社会造成了严重危害。因此,有效检测和防范虚假新闻至关重要。针对当前虚假新闻检测模型局限于从单一数据来源获取新闻文本及视觉信息,导致新闻报道主观性较强、数据覆盖不全面的问题,提出了一种多源异构数据渐进式融合的虚假新闻检测模型。首先,进行多源异构数据的收集、筛选和清洗,由此构建了一个多源多模态数据集,其中包含关于每个事件的多个不同角度的报道;接着,通过将文本特征提取器和视觉特征提取器获取的特征输入多源融合模块,实现了不同来源特征之间的渐进式融合;同时,引入文本的情感特征和图像的频域特征,以实现多层次的特征提取;最后,采用软注意力机制进行特征集成。实验结果和分析表明,与已有的流行方法相比,所提模型有较好的检测效果,为大数据时代的虚假新闻检测提供了一种有效的解决方案。

改进Petri网下异构跨域用户身份验证方法

在异构跨域用户身份验证时,由于用户身份存在属性特征难以捕捉、状态描述误差大等问题,导致验证精准度较低,为有效解决这一问题,文中提出一种基于改进Petri网的用户身份验证算法。利用改进Petri网反映不同局域网中事物间依赖关系,将异构跨域用户活动模型与Petri网模型转换,采用同步用户的状态变化提取实时状态特征值。通过建立云间身份认证中心,计算用户不同身份代表性词语在一定周期内出现的次数和频率值。提取频率最高的词语,并计算该词语在待验证用户全部代表词语库中的权重值,对比权重值与验证熵值,将得到的验证参数输入到认证中心,输出验证结果。实验结果表明,所提方法在先验节点比例为8%时,验证准确率达到0.9,且召回率和综合指标均在0.6以上,具备一定的抗攻击性,整体实用性能强、利用价值高。

基于异构图神经网络的半监督网站主题分类

互联网网站数量快速增长使现有方法难以准确分类特定网站主题,如基于URL的方法无法处理未反映在URL中的主题信息,基于网页内容的方法受到数据稀疏性和语义关系捕捉的限制。为此,提出一种基于异构图神经网络的半监督网站主题分类方法HGNN-SWT。该方法不仅利用网站文本特征来弥补仅使用URL特征的不足,还利用异构图对网站文本和词语的稀疏关系进行建模,通过处理图中的节点和边关系来提高分类性能。同时引入基于随机游走的邻居节点采样方法,考虑节点的局部特征和全局图结构,并提出特征融合策略,捕捉网站文本数据的上下文关系和特征交互。通过在自制的Chinaz Website数据集上的实验,证明了HGNN-SWT方法在网站主题分类任务中相较于现有方法具有更高的准确率。

基于异质特征解构的多模态识别方法

为了解决多模态识别模型因异构模态数据分布之间存在交叉重叠,造成在提取异质特征过程中容易出现特征冗余的问题,提出基于异质特征解构(heterogeneous feature deconstruction,HFD)的多模态识别方法,即构建异质特征解构模型,通过梯度下降的方式训练特性特征提取器,并以梯度反转的方式训练共性特征提取器,提取具有不同模态特质的模态特性特征,以及具有模态不变属性的模态共性特征,进一步利用共性特征增强损失,提高共性特征间的相似度,解决异质特征之间冗余度高的问题。在CMU-MOSEI数据集上的对比实验和消融实验结果验证了基于异质特征解构的多模态识别方法能够有效提升识别性能。

基于空间异质运算的结构信息提取辅助遥感影像分类研究

以Landsat-8和高分一号数据为例,采用仅有光谱特征、3种纹理特征(概率统计、灰度共生矩阵、空间异质运算)辅助光谱特征的方法提取影像空-谱信息,并通过支持向量机分类器进行基于像元的地物分类。结果表明:1)纹理特征辅助光谱特征的地物分类精度明显优于仅使用光谱特征的分类,可提高8.62%~24.36%;2)相较于概率统计、灰度共生矩阵方法结果,空间异质运算结果分类精度在GF-1影像中分别提高了13.31%和2.03%,在Landsat-8影像中分别提高了11.62%和7.79%;3)对于线状地物,相较于概率统计、灰度共生矩阵方法结果,空间异质运算结果分类精度在GF-1数据中分别提高了29.31%和0.80%,在Landsat-8数据中分别提高了11.90%和6.64%,有效减小了分类误差。因此,空间异质运算提取的空间结构信息辅助光谱特征的分类方法能显著改善遥感图像的分类精度,为空间结构信息辅助遥感影像地物分类及线状地物的提取提供一种新的思路和方法。