Human Stress Recognition by Correlating Vision and EEG Data

Because stress has such a powerful impact on human health,we must be able to identify it automatically in our everyday lives.The human activity recognition(HAR)system use data from several kinds of sensors to try to recognize and evaluate human actions automatically recognize and evaluate human actions.Using the multimodal dataset DEAP(Database for Emotion Analysis using Physiological Signals),this paper presents deep learning(DL)technique for effectively detecting human stress.The combination of vision-based and sensor-based approaches for recognizing human stress will help us achieve the increased efficiency of current stress recognition systems and predict probable actions in advance of when fatal.Based on visual and EEG(Electroencephalogram)data,this research aims to enhance the performance and extract the dominating characteristics of stress detection.For the stress identification test,we utilized the DEAP dataset,which included video and EEG data.We also demonstrate that combining video and EEG characteristics may increase overall performance,with the suggested stochastic features providing the most accurate results.In the first step,CNN(Convolutional Neural Network)extracts feature vectors from video frames and EEG data.Feature Level(FL)fusion that combines the features extracted from video and EEG data.We use XGBoost as our classifier model to predict stress,and we put it into action.The stress recognition accuracy of the proposed method is compared to existing methods of Decision Tree(DT),Random Forest(RF),AdaBoost,Linear Discriminant Analysis(LDA),and KNearest Neighborhood(KNN).When we compared our technique to existing state-of-the-art approaches,we found that the suggested DL methodology combining multimodal and heterogeneous inputs may improve stress identification.

Sleep spindles detection from human sleep EEG signals using autoregressive (AR) model: a surrogate data approach

A new algorithm for the detection of sleep spindles from human sleep EEG with surrogate data approach is presented. Surrogate data ap-proach is the state of the art technique for nonlinear spectral analysis. In this paper, by developing autoregressive (AR) models on short segment of the EEG is described as a superposition of harmonic oscillating with damping and frequency in time. Sleep spindle events are detected, whenever the damping of one or more frequencies falls below a prede-fined threshold. Based on a surrogate data, a method was proposed to test the hypothesis that the original data were generated by a linear Gaussian process. This method was tested on human sleep EEG signal. The algorithm work well for the detection of sleep spindles and in addition the analysis reveals the alpha and beta band activities in EEG. The rigorous statistical framework proves essential in establishing these results.

Mechanical Fault Diagnosis Based on Band-phase-randomized Surrogate Data and Multifractal

The vibration signals of machinery with various faults often show clear nonlinear characteristics.Currently,fractal dimension analysis as the common useful method for nonlinear signal analysis,is a kind of single fractal form,which only reflects the overall irregularity of signals,but cannot describe its local scaling properties.For comprehensive revealing of internal properties,a combinatorial method based on band-phase-randomized（BPR） surrogate data and multifractal is introduced.BPR surrogate data method is effective to eliminate nonlinearity in specified frequency band for a fault signal,which can be utilized to detect nonlinear degree in whole fault signal by nonlinear titration method,and the overall nonlinear distribution of fault signal is displayed in nonlinear characteristic curve that can be used to analyze the fault signal qualitatively.Then multifractal theory as a quantitative analysis method is used to describe geometrical characteristics and local scaling properties,and asymmetry coefficient of multifractal spectrum and multifractal entropy for fault signals are extracted as new criterions to diagnose machinery faults.Several typical faults include rotor misalignment,transversal crack,and static-dynamic rubbing fault are analyzed,and the results indicate that those faults can be distinguished by the proposed method effectively,which provides a qualitative and quantitative analysis way in the field of machinery fault diagnosis.

Test for random in electrical signals time series of CO_2 short circuit transition welding process by the method of surrogate data

This paper introduced the basic theory and algorithm of the surrogate data method, which proposed a rigorous way to detect the random and seemingly stochastic characteristics in a system. The Gaussian data and the Rossler data were used to show the availability and effectivity of this method. According to the analysis by this method based on the short-circuiting current signals under the conditions of the same voltage and different wire feed speeds, it is demonstrated that the electrical signals time series exhibit apparently randomness when the welding parameters do not match. However, the electrical signals time series are deterministic when a match is found. The stability of short-circuiting transfer process could be judged exactly by the method of surrogate data.

Individualization of Data-Segment-Related Parameters for Improvement of EEG Signal Classification in Brain-Computer Interface

In electroencephalogram (EEG) modeling techniques, data segment selection is the first and still an important step. The influence of a set of data-segment-related parameters on feature extraction and classification in an EEG-based brain-computer interface (BCI) was studied. An auto search algorithm was developed to study four datasegment-related parameters in each trial of 12 subjects’ EEG. The length of data segment (LDS), the start position of data (SPD) segment, AR order, and number of trials (NT) were used to build the model. The study showed that, compared with the classification ratio (CR) without parameter selection, the CR was increased by 20% to 30% with proper selection of these data-segment-related parameters, and the optimum parameter values were subject-dependent. This suggests that the data-segment-related parameters should be individualized when building models for BCI.

Detecting Nonlinear Dynamics Using BDS Test and Surrogate Data in Financial Time Series

Physicists experimentalists use many observations of a phenomenon, which are the unknown equations that describe it, in order to understand the dynamics and obtain information on their future behavior. In this article we study the possibility of reproducing the dynamics of the phenomenon using only a measurement scale. The Whitney immersion theorem ideas are presented and generalization of Sauer for fractal sets to rebuild the asymptotic behavior of the phenomena and to investigate evidence of nonlinear dynamics in the reproduced dynamics using the Brock, Dechert, Scheinkman test (BDS). The applications are made in the financial market which are only known stock prices.

基于Otsu的EEG通道选择情绪识别研究

脑电信号情绪识别是数据人机交互(HCI)技术的一种,实时情感识别对于模型性能要求较高,为实现以较低的运算成本获取较高的识别精度,采用时域滑动窗口的方法扩充样本量,基于Otsu算法筛选出含有最多情绪特征信息的通道,并利用快速傅里叶变换进行脑电信号频段提取,以功率谱密度作为特征,构建了基于支持向量机等分类模型,对高唤醒-低唤醒(HA-LA)和高效价-低效价(HV-LV)两种任务进行分类。实验表明,使用SVM分类器在HA-LA情绪识别任务中得到(82.2±0.4)%的识别准确率,在HV-LV情绪识别任务中得到(83.4±0.3)%的识别准确率。所提出的时域滑动窗口能有效提取含有情绪的脑电信号,在减少数据量的情况下仍获得了不错的情绪识别性能,为实时情感识别的脑机接口提供了一种高效的模型。

Ensemble Deep Learning with Chimp Optimization Based Medical Data Classification

Eye state classification acts as a vital part of the biomedical sector,for instance,smart home device control,drowsy driving recognition,and so on.The modifications in the cognitive levels can be reflected via transforming the electro-encephalogram(EEG)signals.The deep learning(DL)models automated extract the features and often showcased improved outcomes over the conventional clas-sification model in the recognition processes.This paper presents an Ensemble Deep Learning with Chimp Optimization Algorithm for EEG Eye State Classifi-cation(EDLCOA-ESC).The proposed EDLCOA-ESC technique involves min-max normalization approach as a pre-processing step.Besides,wavelet packet decomposition(WPD)technique is employed for the extraction of useful features from the EEG signals.In addition,an ensemble of deep sparse autoencoder(DSAE)and kernel ridge regression(KRR)models are employed for EEG Eye State classification.Finally,hyperparameters tuning of the DSAE model takes place using COA and thereby boost the classification results to a maximum extent.An extensive range of simulation analysis on the benchmark dataset is car-ried out and the results reported the promising performance of the EDLCOA-ESC technique over the recent approaches with maximum accuracy of 98.50%.

基于拓扑数据分析的驾驶疲劳EEG数据处理与优化分析研究

为提高驾驶疲劳脑电(EEG)数据处理与分析的准确性和鲁棒性,提出一种基于拓扑数据分析(TDA)的驾驶人疲劳脑电分析方法。首先利用汽车性能虚拟仿真平台开展驾驶实验,通过驾驶人状态反馈和面部特征视频,标记脑电数据,形成清醒和疲劳二分数据集。之后利用EEGLAB预处理数据,剔除噪声并保留0.3~30 Hz频带,直接从时域EEG数据中提取拓扑特征。此外还提取了经典频域特征α波能量和α/β用于对比分析。最后使用支持向量机进行分类。结果表明,基于持久同源(PH)的拓扑特征取得了高达88.7%的准确率和91.4%的召回率,与经典频域特征性能相当,且对脑电伪影的鲁棒性明显更好,在未剔除EEG伪影的情况下仍取得了87.4%的准确率和89.7%的召回率。综上所述,本文提出的用于驾驶疲劳脑电信号处理与分析的TDA方法抗干扰特性好、处理成本低、经济性高,有助于稳定、高效地处理驾驶人脑电数据并检测驾驶疲劳状态,具有较大的科学实际应用价值。

基于多模态数据的在线学习认知负荷评估

近年来,认知负荷过载成为影响在线学习效果的一个重要因素。为解决此问题,文章聚焦在线学习认知负荷评估,首先设计了基于多模态数据的在线学习认知负荷评估研究框架,包含多模态数据采集、多模态特征提取和评估模型构建三个部分。接着,文章通过实验,采集学习者在特定在线学习环境下的多模态数据,提取人脸表情特征、眼动追踪指标特征和脑电信号特征,构建在线学习认知负荷评估模型。之后,文章对评估模型的准确性进行验证,发现在五种机器学习算法模型中,KNN模型的综合性能最佳,故将此模型作为最终的在线学习认知评估模型;同时,文章对多模态数据进行了分析,发现多模态融合方法在认知负荷评估上具有优越性。文章构建的评估模型可赋能在线学习平台实现认知负荷的实时评估,进而实现基于认知负荷的个性化学习,增强学习动机,提升学习效果,促进在线教育质量提升。

基于双子空间PCA降维的脑力负荷分类

人类社会至今的飞速发展使得大量体力劳动被机械工程替代,工作者的任务重心也从体力劳动逐渐转变为脑力劳动,对操作者脑力负荷进行实时评估以增强工作效率在当下有着重大意义。目前人类对于脑力负荷评估共有3种方式,有研究表明,采用生物电信号进行脑力负荷分类效果较其余两种方法更客观。但脑电信号经过特征提取后维数极高,所需数据量和运算量巨大,需要对其进行降维。目前降维方面最广泛运用的两种算法为主成分分析(principal component analysis,PCA)和线性判别分析(linear discriminate analysis,LDA)。针对PCA的非监督性和LDA的特征冗余敏感性,提出一种二分类下基于双子空间主成分分析的降维算法,分别对不同类别的训练集数据进行主成分分析,并将所有训练集数据映射到生成的空间中,再次进行PCA-LDA降维,以此提高降维后数据的可分性。实验结果表明,双子空间PCA-LDA降维算法在二分类任务下测试集精度整体高于单子空间PCA-LDA算法,以此为脑力负荷分类领域和高维数据降维领域提供了新思路。

数据驱动的机身对接过程仿真技术研究

针对传统机身对接仿真中理论模型偏差较大、有限元模型消耗时间长等问题,开展数据驱动的机身对接过程仿真技术研究。首先通过虚实融合技术,构建六自由度平台运动学模型和机身–调姿平台位姿转换模型,通过数据交互实现仿真数据实时传输从而驱动虚拟模型,利用测量数据和虚拟模型进行了机身对接仿真,确定了对接过程位姿参数,为调姿平台参数修正提供数据基础。之后为提升数据求解效率,根据求解机身变形的有限元模型,计算出变形量并转化成机身位置变化量,将机身姿态角和位置变化量作为输入输出值构建代理模型,并验证方法有效性。最后开发了机身对接过程仿真系统,以机身试验件的对接过程为例,验证了仿真系统的可行性。

基于集成式张量域自适应的运动想象脑电分类

实际应用中脑电信号一直面临采集成本高、用户间差异大等问题,制约着基于脑电信号的运动想象领域的发展。针对跨受试者运动想象脑电信号识别任务,本研究提出了一种基于集成式张量域自适应的迁移学习方法。首先采用改进的欧氏空间对齐方法对多维脑电数据进行协方差对齐,消除原始数据的边缘分布偏移;其次提出基于张量子空间的改进联合概率分布方法,求得不同类别的映射子空间并实现未知标签的目标域识别分类。分别在7人200个样本和9人144个样本的BCI数据集上进行了实验,平均准确率达到82.18%和76.45%,证明了该方法在跨域分类识别上具有很好的性能。同时对于该方法各环节的效果也进行了可视化验证,展示了该集成式方法在跨域问题上的效果。

基于多模态数据的在线学习认知风格智能识别

在人工智能技术迅速发展的背景下,多模态数据分析在教育领域的重要性日益凸显。本研究通过提取学习者在线学习中产生的脑电数据、表情数据和眼动数据等,构建了基于多模态数据的认知风格智能识别框架,并运用六种机器学习模型验证其有效性。研究结果显示,四种机器学习模型的识别准确率超过80%,且最佳模型的准确率达89.17%,F1分数达0.9241。此外,与基于单模态数据的模型相比,基于多模态数据的认知风格模型识别表现更佳。这表明,多模态数据融合策略具有优越性,有助于加强在线学习平台的适应性和个性化。

基于STM32和USB虚拟串口的EEG信号采集仪设计

便携式脑电信号采集系统采用三级放大器将脑电信号逐级放大数千倍,采用仪表放大芯片INA129作为前置放大器,ADS8568作为模拟数字转换器,对32路EEG信号进行放大滤波和同步采集。以STM32微处理器作为核心,通过基于USB的虚拟串口将原始数据传给计算机,在计算机上使用MATLAB对数据进行数字滤波处理,最终将波形实时显示在GUI中。使用USB接口方便与主机相连,保证了传输带宽和稳定性,虚拟串口方便MATLAB软件的开发。与传统EEG信号采集仪相比,具有便携性强、成本低廉、便于重构等优点。

高功率腔体滤波器的电磁-热-力多物理场智能代理模型及快速预测

电磁多物理场计算在模拟高功率微波器件方面至关重要。为了快速且精确地获取高功率腔体滤波器中由电磁引发的多物理场响应,本文构建了一种基于数据驱动的电磁多物理场代理模型。首先利用有限元方法建立腔体滤波器的多物理场响应数据库,随后基于该数据库构建数据驱动的人工神经网络代理模型。结果对比表明:该代理模型能够精确且快速地预测高功率腔体滤波器在不同输入功率下的S参数曲线,预测精度超过98%,预测时间少于0.2 s;相较于传统的电磁多物理场数值计算方法,计算速度提升了3个数量级以上。因此,本文代理模型的快速精确预测能力对新型高功率微波器件的稳定性分析、可靠性评估和优化设计具有重要的理论指导意义和实际工程应用价值。

脑电信号情绪识别关键技术研究进展

随着人机交互技术的不断进步和广泛应用,对用户情绪的准确识别变得日益重要。情绪识别技术已经在多个领域展现出巨大潜力,包括医疗诊断、交通安全和教育等方面。而基于脑电的情绪识别成为了情绪识别领域中的热门研究方向。首先,本文介绍情绪连续和情绪离散模型的基本概念,总结常用的脑电公开数据集,并对数据集的规模、情绪标签以及它们对脑电情绪识别任务的影响进行了比较分析。其次,因为不同频带和脑电通道对于情绪识别的影响各不相同,课题组汇总脑电情绪识别关键频带和通道的相关研究,并从文献中总结归纳出脑电情绪识别的关键频带范围以及具有丰富情绪信息的脑电通道位置。接着,介绍四类脑电情绪特征且给出对应的特征提取方法,也指出各种脑电特征的提取难度和目前在情绪识别中的应用效果。然后,对基于深度学习的脑电情绪识别中的数据增强技术和注意力机制进行了阐述,指出数据增强技术的主流趋势和生成的人工情绪特征的类型,对各种注意力机制的作用方式和侧重点进行了对比分析。数据增强技术用来解决脑电数据量不足的问题,注意力机制则对情绪识别准确率的提高起到了关键作用。最后,对未来脑电情绪识别模型的通用性和脑电采集设备的研究方向做了一定的展望。

一种基于深度学习的时序病变数据段分类方法

针对在大规模时序医疗数据的分析中现有检测方法检测精度低、检测速度慢等问题,文中提出了一种基于深度学习的时序病变数据段分类方法。该方法在TSTKS(Ternary Search Trees and modified Kolmogorov-Smirnov)算法和滑动窗口理论的基础上,利用深度学习技术实现了对病变数据段的快速准确分类。文中以利用该方法对病变数据段进行分类的结果作为依据,实现了滑动窗口大小的动态调整。通过对真实癫痫脑电信号(Electroencephalogram,EEG)进行分析,证明了所提病变数据段分类方法和基于该分类方法的滑动窗口动态调整机制具有检测速度快、精度较高等优点,可以为大规模时序数据的快速分析研究提供一种新选择。

基于参考向量关联估计的离线多目标优化算法

很多实际工程和科学问题都是计算费时的多目标优化问题,这类问题中每个候选解的评价往往都非常费时,因此仅允许使用少量真实评价。论文采用离线数据驱动的进化算法求解计算费时多目标优化问题,以期节省优化时间。论文通过训练代理模型来估计候选解的收敛性,采用最近邻样本估计候选解与参考向量的关联关系,减少了使用目标估值计算候选解与参考向量夹角大小所产生的误差累积。使用DTLZ测试集验证论文算法的有效性,论文算法与离线数据驱动的优化算法MS-RV以及三个经典在线数据驱动优化算法进行对比,实验结果表明论文提出的算法在保证性能的前提下,可以减少使用真实的评价次数。

面向抗冲击结构尺寸优化的自训练分类判断优化设计方法

抗冲击结构在军民安全领域有着广泛且重要的应用,但其优化设计因涉及极端载荷下的强非线性而面临结构响应求解耗时与灵敏度分析困难的挑战.针对该问题,提出了一种自训练分类判断优化设计方法.该方法采用基于支持向量机的自训练分类判断代理模型和遗传算法,求解抗冲击结构尺寸优化列式.与基于回归代理模型的传统方法不同,自训练分类判断代理模型可以减少样本集构建计算成本.同时,提出了面向分类判断代理模型的遗传算法约束处理策略及个体适应度计算方法.基于该方法开展的波纹夹芯结构抗爆性能和聚脲/陶瓷复合板抗侵彻性能优化设计,验证了其有效性和高效性.为抗冲击结构设计提供了一种新的高效优化方法,以期保障极端冲击载荷下装备结构服役性能和人员安全.