Depressive semantic awareness from vlog facial and vocal streams via spatio-temporal transformer

With the rapid growth of information transmission via the Internet,efforts have been made to reduce network load to promote efficiency.One such application is semantic computing,which can extract and process semantic communication.Social media has enabled users to share their current emotions,opinions,and life events through their mobile devices.Notably,people suffering from mental health problems are more willing to share their feelings on social networks.Therefore,it is necessary to extract semantic information from social media(vlog data)to identify abnormal emotional states to facilitate early identification and intervention.Most studies do not consider spatio-temporal information when fusing multimodal information to identify abnormal emotional states such as depression.To solve this problem,this paper proposes a spatio-temporal squeeze transformer method for the extraction of semantic features of depression.First,a module with spatio-temporal data is embedded into the transformer encoder,which is utilized to obtain a representation of spatio-temporal features.Second,a classifier with a voting mechanism is designed to encourage the model to classify depression and non-depression effec-tively.Experiments are conducted on the D-Vlog dataset.The results show that the method is effective,and the accuracy rate can reach 70.70%.This work provides scaffolding for future work in the detection of affect recognition in semantic communication based on social media vlog data.

Providing Global Awareness in Distributed Dynamic Systems

The paper investigates applicability of the developed high-level model and technology for solution of diverse problems in large distributed dynamic systems which can provide sufficient awareness of their structures,organization,and functionalities.After the review of meanings of awareness and existing approaches for its expression and support,the paper shows application of the Spatial Grasp Model and Technology(SGT)and its basic Spatial Grasp Language(SGL)for very practical awareness solutions in large distributed dynamic systems,with obtaining any knowledge from any point inside or outside the system.The self-evolving,self-replicating,and self-recovering scenario code in SGL can effectively supervise distributed systems under any circumstances including rapidly changing number of their elements.Examples are provided in SGL for distributed networked systems showing how in any node any information about other nodes and links,including the whole system,can be obtained by using network requesting patterns based on recursive scenarios combining forward and backward network matching and coverage.The returned results may be automatically organized in networked patterns too.The presented exemplary solutions are parallel and fully distributed,without the need of using vulnerable centralized resources,also very compact.This can be explained by fundamentally different philosophy and ideology of SGT which is not based on traditional partitioned systems representation and multiple agent communications.On the contrary,SGT and its basic language supervise and control distributed systems by holistic self-spreading recursive code in wavelike,virus-like,and even“soul-like”mode.

Implementation of Capacity Development and Public Awareness for CTBT Verification Regime in Myanmar

Myanmar signed the Comprehensive Nuclear-Test-Ban Treaty(CTBT)in 1996 and ratified it in 2016.As a CTBT ratifying state,Myanmar needs to implement the capability development and public awareness concerning with CTBT verification activities.When implementing the CTBT verification regime,Myanmar considered the linkage between CTBT and sustainable development goals(SDGs).A National Data Centre(NDC)is needed in the country for collecting the data from the stations,differentiating the data quality,prioritizing and storage of the data to conduct with International Data Centre(IDC)of Comprehensive Nuclear-Test-Ban Treaty Organization(CTBTO).The seismic,hydroacoustic,infrasound and radionuclide technologies are important for identifying the signals generated by a nuclear explosion.Various capacity development and public awareness are important for initial step before the ratification process of the state party.Civil society,change school curriculum,participation in process,organizing workshops and seminars are carried out by policy makers’decision.International collaboration and citizen awareness are needed to understand the CTBT verification system for monitoring the nuclear explosions in a global context.The Atomic Energy Division(AED)will collaborate with external funding agents like CTBTO to improve skill of human resource for NDC implementation in Myanmar.Young scientists,science administrators and representative will be achieved from this national development initiative.This is focused on research and development in areas potentially relevant to the Treaty’s verification regime.

Context Awareness by Noise-Pattern Analysis of a Smart Factory

Recently,to build a smart factory,research has been conducted to perform fault diagnosis and defect detection based on vibration and noise signals generated when a mechanical system is driven using deep-learning technology,a field of artificial intelligence.Most of the related studies apply various audio-feature extraction techniques to one-dimensional raw data to extract sound-specific features and then classify the sound by using the derived spectral image as a training dataset.However,compared to numerical raw data,learning based on image data has the disadvantage that creating a training dataset is very time-consuming.Therefore,we devised a two-step data preprocessing method that efficiently detects machine anomalies in numerical raw data.In the first preprocessing process,sound signal information is analyzed to extract features,and in the second preprocessing process,data filtering is performed by applying the proposed algorithm.An efficient dataset was built formodel learning through a total of two steps of data preprocessing.In addition,both showed excellent performance in the training accuracy of the model that entered each dataset,but it can be seen that the time required to build the dataset was 203 s compared to 39 s,which is about 5.2 times than when building the image dataset.

Predictors of emotional awareness during childhood

Emotional awareness, the ability to identify and describe one’s own emotions, and those of other people, has been widely studied in adult samples, while only few studies have been conducted during childhood. Aim of the present study was to investigate the predictability of the main variables usually associated with the study of emotional awareness. The LEAS-C was administered to a group of 781 children aged 8 to 13 years old. The results showed that both gender and verbal skills are important predictors of children’s emotional awareness. Moreover, cognitive ability and emotion recognition also predicted LEAS-C scores. Finally, age emerged as a significant predictor of the complexity of children’s emotional awareness. The assessment of children’s emotional awareness may be significant for different health-related purposes, e.g. prevention and psychological support.

Multilayer Neural Network Based Speech Emotion Recognition for Smart Assistance

Day by day,biometric-based systems play a vital role in our daily lives.This paper proposed an intelligent assistant intended to identify emotions via voice message.A biometric system has been developed to detect human emotions based on voice recognition and control a few electronic peripherals for alert actions.This proposed smart assistant aims to provide a support to the people through buzzer and light emitting diodes(LED)alert signals and it also keep track of the places like households,hospitals and remote areas,etc.The proposed approach is able to detect seven emotions:worry,surprise,neutral,sadness,happiness,hate and love.The key elements for the implementation of speech emotion recognition are voice processing,and once the emotion is recognized,the machine interface automatically detects the actions by buzzer and LED.The proposed system is trained and tested on various benchmark datasets,i.e.,Ryerson Audio-Visual Database of Emotional Speech and Song(RAVDESS)database,Acoustic-Phonetic Continuous Speech Corpus(TIMIT)database,Emotional Speech database(Emo-DB)database and evaluated based on various parameters,i.e.,accuracy,error rate,and time.While comparing with existing technologies,the proposed algorithm gave a better error rate and less time.Error rate and time is decreased by 19.79%,5.13 s.for the RAVDEES dataset,15.77%,0.01 s for the Emo-DB dataset and 14.88%,3.62 for the TIMIT database.The proposed model shows better accuracy of 81.02%for the RAVDEES dataset,84.23%for the TIMIT dataset and 85.12%for the Emo-DB dataset compared to Gaussian Mixture Modeling(GMM)and Support Vector Machine(SVM)Model.

基于ERes-ECAM的动物声纹识别

声纹识别技术不仅在人类身份验证领域广泛应用,在动物种类识别方面也取得一定进展。现有模型存在特征表达能力不足的问题,同时,在保证性能的前提下,模型的时间复杂度和推理速度有待优化。提出用于发声动物嵌入学习的改进的残差块连接改进的上下文感知掩蔽(Enhanced Res2block connected Enhanced Context Aware Masking, ERes-ECAM)新型架构,采用了稠密连接的时延神经网络(Densely-connected Time Delay Neural Network, D-TDNN)作为骨干,为了解决模糊不相关噪声问题的同时能够提取更多有效的关键信息,在D-TDNN层中采用多粒度池化方法的改进的上下文感知掩蔽(Enhanced Context Aware Masking, ECAM)模块,前端连接残差模块,通过局部特征融合(Local Feature Fusion, LFF)的方式,将残差块内提取的特征进行融合来提取局部信息,提升了声纹验证系统的准确性和鲁棒性。在Anim-Celeb和Pig-Celeb两个测试集中分别实验,实验结果表明,所提架构的等错误率(Equal Error Rate, EER)分别达到6.88%和7.24%,同时,对动物种类和猪只种类识别准确率达到了93.12%和92.76%。

基于司法视角对诚实信用原则在专利领域的适用研究

近期,专利法领域引入诚实信用原则,规定申请专利和行使专利权应当遵循诚实信用原则,不得滥用专利权损害公共利益或者他人合法权益。本文首先分析专利审批过程中适用诚实信用条款可能面临的现实问题,在此基础上针对性地研究知识产权领域中诚实信用原则相关的司法解释和近期典型判例,并立足于国内专利申请现状和专利工作特点提出建议。首先,从行为认定角度提升不诚实信用行为的识别能力;其次,从规则完善角度平衡诚信相关证明责任;再次,从法条适用角度合理适用诚信相关条款;最后,从事后惩戒角度完善不诚信行为的惩戒机制。本文为建立诚实信用原则在专利领域中的适用方式提供参考,从而使诚实信用原则在知识产权强国建设的新征程中更好地发挥其鼓励科技创新、助力经济高质量发展的应有功能。

基于HGTC−YOLOv8n模型的煤矸识别算法研究

现有基于深度学习的煤矸识别方法在煤矿井下低照度、高噪声及运动模糊等复杂工况下存在煤矸识别精度低、小目标煤矸容易漏检、模型参数量和运算量大,难以部署到计算资源有限的设备中等问题,提出了一种基于HGTC−YOLOv8n模型的煤矸识别算法。采用HGNetv2网络替换YOLOv8n的主干网络,通过多尺度特征的有效提取,提高煤矸识别效果并减少模型的存储需求和计算资源消耗;在主干网络中嵌入三重注意力机制模块Triplet Attention,捕获不同维度间的交互信息,增强煤矸图像目标特征的提取,减少无关信息的干扰;选用内容感知特征重组模块(CARAFE)来改进YOLOv8n颈部特征融合网络上采样算子,利用上下文信息提高感受视野,提高小目标煤矸识别准确率。实验结果表明:①HGTC−YOLOv8n模型的平均精度均值为93.5%,模型的参数量为2.645×10^(6),浮点运算量为8.0×10^(9),帧速率为79.36帧/s。②平均精度均值较YOLOv8n模型提升了2.5%,参数量和浮点运算量较YOLOv8n模型分别下降了16.22%和10.11%。③与YOLO系列模型相比,HGTC−YOLOv8n模型的平均精度均值最高,且参数量和浮点运算量最少,检测速度较快,综合检测性能最佳。④基于HGTC−YOLOv8n模型的煤矸识别算法在煤矿井下复杂工况下,改善了煤矸识别精度低、小目标煤矸容易漏检等问题,满足煤矸图像实时检测要求。

激光技术在天基成像中的应用进展

针对高轨目标的精细特征识别、亚厘米级空间碎片的探测、全天时全天域成像的主动成像等需求,对目前3类采用激光技术的天基成像手段包括激光雷达三维成像技术、逆合成孔径激光雷达(ISAL)成像技术、激光主动照明成像技术的工作原理、成像性能、优势、应用发展现状进行了对比和总结归纳。对于几类天基激光成像手段,其关键技术均包括了发展小型化高功率激光器,以增强成像的信噪比和成像距离,此外,对于三维激光成像和ISAL成像技术,需要发展大面阵探测器的技术,来提高其成像的视场。总结目前的各类成像手段,ISAL是目前对天基远距离目标逆光成像的唯一手段,具有极大的发展潜力,但在应用时需要考虑逆光条件下的目标跟踪捕获,未来天基应用可以考虑采用逆合成孔径微波雷达技术(ISAR)进行视场引导跟踪。

基于关系感知和标签消歧的细粒度面部表情识别算法

细粒度表情识别任务因其包含更丰富真实的人类情感而备受关注.现有面部表情识别算法通过提取局部关键区域等方式学习更优的图像表征.然而,这些方法忽略了图像数据集内在的结构关系,且没有充分利用标签间的语义关联度以及图像和标签间的相关性,导致所学特征带来的性能提升有限.其次,现有细粒度表情识别方法并未有效利用和挖掘粗细粒度的层级关系,因而限制了模型的识别性能.此外,现有细粒度表情识别算法忽略了由于标注主观性和情感复杂性导致的标签歧义性问题,极大影响了模型的识别性能.针对上述问题,本文提出一种基于关系感知和标签消歧的细粒度面部表情识别算法(fine-grained facial expression recognition algorithm based on Relationship-Awareness and Label Disambiguation,RALD).该算法通过构建层级感知的图像特征增强网络,充分挖掘图像之间、层级标签之间以及图像和标签之间的依赖关系,以获得更具辨别性的图像特征.针对标签歧义性问题,算法设计了基于近邻样本的标签分布学习模块,通过整合邻域信息进行标签消歧,进一步提升模型识别性能.在细粒度表情识别数据集FG-Emotions上算法的准确度达到97.34%,在粗粒度表情识别数据集RAF-DB上比现有主流表情分类方法提高了0.80%~4.55%.

基于舍入误差的神经网络量化方法

深度神经网络需要付出高昂的计算成本,降低神经网络推理的功耗和延迟,是将神经网络集成到对功耗和计算严格要求的边缘设备上的关键所在。针对这一点,提出一种采用舍入误差的端到端神经网络训练后量化方法,缓解神经网络量化到低比特宽时带来的精度下降问题。该方法只需采用小批量且无标注的数据进行训练,且在不同的神经网络结构上都有十分不错的表现,RegNetX-3.2GF在权重和激活数的比特宽均为4的情况下分类准确率下降不到2%。

人形机器人的法律治理基本架构

人形机器人具有跨场景的通用性,可以在多种场景中与人类展开合作,可能通过全面的运用形成人机共生社会。在人机共生社会中,法律治理应当超出特定的场景设计,形成更具一般性的架构。跨场景应用的人形机器人在态势感知和情感识别方面的特征,决定了人形机器人的法律治理不同于一般意义上的机器人。以此为出发点对人形机器人的法律治理基本架构进行设计,平衡技术发展与社会风险,形成合理的激励机制。

基于上下文感知与图注意力网络的人机协作装配人员作业意图识别方法

针对人机协作装配现有研究尚未充分综合利用复杂装配环境中各要素的三维空间特征与视觉特征,并考虑装配环境中丰富的上下文信息,使得复杂装配环境下人员作业意图识别精度较低,提出结合装配环境各要素的三维空间特征和视觉特征,基于图注意力网络实现人员作业意图的高精度识别方法。首先利用FasterR-CNN神经网络对装配场景中的各要素,如人员、机器人、零件等进行目标检测,得到各要素的空间位置信息,同时从网络中提取各要素的视觉特征信息;然后结合图注意力网络推理装配过程中人员对不同作业对象的作业意图,如搬运、组装、触碰等;最后通过人机协作场景下的齿轮装配实验对所提方法进行验证。实验结果表明,相比深度卷积神经网络,所提方法在识别准确性、场景泛化性等方面具有优越性。

基于眼动追踪技术的引航员情景意识识别研究

情景意识(Situation Awareness,SA)缺失是引航作业人为失误的主要根源,有效识别引航员SA水平是保障引航安全的关键。针对现有测量方法的侵入性、主观性、间断性问题,设计典型场景引航模拟试验,利用眼动特征开展引航员SA水平识别研究。应用眼动追踪技术实时采集引航员的眼动试验数据;采用独立样本t检验和Mann-Whitney检验方法研究SA与眼动指标的关联效应;基于敏感眼动指标结合支持向量机(Support Vector Machine,SVM)构建引航员SA识别模型,实现引航操作任务下SA的有效识别。结果显示:高SA组的视觉搜索效率大于低SA组,高SA组主要关注窗外视景而低SA组更多关注电子海图,高SA组比低SA组获取信息的能力更强;平均扫视次数(Average Saccade Count,ASC)、注视总时间百分比(Fixation Duration Percentage,FDP)、平均注视持续时间(Average Fixation Duration,AFD)3个眼动指标与引航员SA水平具有显著相关性;采用“ASC+FDP+AFD”组合指标时且在5 s时间窗口长度下,SA识别准确率达到92.23%。研究结果表明了该识别模型的有效性。

基于标签语义信息感知的少样本命名实体识别方法

在少样本命名实体识别方法中,目前广泛应用的方法是基于原型网络的两阶段模型。但是,该方法未充分利用实体标签中的语义信息,且在距离计算中过度依赖实体类型原型向量,导致模型泛化能力差。针对这些问题,提出一种基于标签语义信息感知的少样本命名实体识别方法。该方法是一种先进行实体跨度检测,再判断实体类型的两阶段方法。在构建实体类型原型向量时,将对应实体类型包含的语义信息考虑在内,通过维度转换层将其与原型向量相融合。在对新样本进行实体识别时,将实体类型的正负样本与实体类型原型向量组成实体类型三元组,依据样本到三元组的距离对其进行分类。在多个数据集上的实验结果证明,该模型的性能比以往的模型有较大的提升。

基于理论-实验-项目三位一体的制造工程认知实践课程改革

为了使学生在学习制造工程认识实践课程中,将理论知识的学习、实验技能的训练及实践能力的培养有机融合,提出构建理论-实验-项目三位一体的课题体系。理论部分主要起到扩大学生的知识面,了解制造业的前沿的目的;实验部分主要通过设计不同的实验环节,让学生掌握基本的实验技能;实践项目是基于学习的理论知识及掌握的实验技能,通过完整的项目选题、项目实施及项目展示培养学生的创新能力及实践动手能力。通过理论-实验-项目并行推进、融汇贯通,从而达到学生知识、技能、能力的全方位提升。同时建立与前期课程CAD和工程制图的衔接,打通由“工程表达”到“工程实现”的渠道,通过在课程中引入“精度”“配合”“成本”“效率”等概念,为学生后续系统学习机械设计、机械制造基础、先进制造实践等专业课程起到引导作用,从而实现课程的衔接与贯通。

专项任务与课程学分互换研究——以信息安全技术应用专业为例

为彰显信息安全技术应用专业“实战型教学”特色,鼓励和促进学生个性发展和创新发展,解决信息安全专业学生因参与信息安全专业专项任务而导致课程学习时间不足、课程学分损失等问题,提出了信息安全专业专项任务和课程学分互换研究机制。结合学校信息安全技术应用专业发展现状和教育政策,对信息安全专项任务的种类、专项任务的成果和学分认定标准、学分互换标准和算法等进行深入研究,形成实施方案。将研究内容和实施方案运用于信息安全技术应用专业专项任务的开展中,成立专项任务学分评定委员会,推动信息安全专项任务学分认定与互换制度有效实施,践行“以学生发展为本”的个性化育人理念,创新“技能与学分融通”的高职网络安全人才培养模式,有效地推动了学校学分制教学发展。

居民对艾滋病防控服务认知情况与防控行为的关系研究

目的了解居民对艾滋病防控服务的接受情况、知晓度和认可度,并探讨其与防控行为的关系。方法采用方便抽样的方法,对成都市936名居民进行问卷调查。结果有78.85%的被调查者仅接受过1种甚至没有接受过艾滋病防控服务。在防控服务方面,居民满意度最高的是健康教育的内容。调查对象对艾滋病知识的知晓度较高,但对艾滋病及其患者的接受度较低。Logistic回归分析结果显示:年龄、学历、防控服务满意度、艾滋病知识知晓度和对艾滋病及其患者的接受度是居民艾滋病防控行为的影响因素。结论居民的艾滋病预防控制行为与其态度和意识关系密切,应当强化艾滋病防控服务的覆盖范围和内容,进一步加大艾滋病防控专业人员和专项资金的配备,引导大众消除对艾滋病的歧视并选择正确的防控行为,促进形成良好的艾滋病防控氛围。

城市儿童友好空间建设思考

本文结合《城市儿童友好空间建设导则》和《实施手册》的编制,探讨我国儿童成长的需求特征和儿童友好空间建设存在的问题,通过国内外儿童友好空间建设原则和实施策略的比对,提出现阶段我国儿童友好空间建设应注重系统推进,开展儿童需求调研,注重既有空间评估,积极推动建立包容性的儿童参与机制,通过多部门合作,完善建设、服务、运营全生命周期的策划。