Customized Convolutional Neural Network for Accurate Detection of Deep Fake Images in Video Collections

The motivation for this study is that the quality of deep fakes is constantly improving,which leads to the need to develop new methods for their detection.The proposed Customized Convolutional Neural Network method involves extracting structured data from video frames using facial landmark detection,which is then used as input to the CNN.The customized Convolutional Neural Network method is the date augmented-based CNN model to generate‘fake data’or‘fake images’.This study was carried out using Python and its libraries.We used 242 films from the dataset gathered by the Deep Fake Detection Challenge,of which 199 were made up and the remaining 53 were real.Ten seconds were allotted for each video.There were 318 videos used in all,199 of which were fake and 119 of which were real.Our proposedmethod achieved a testing accuracy of 91.47%,loss of 0.342,and AUC score of 0.92,outperforming two alternative approaches,CNN and MLP-CNN.Furthermore,our method succeeded in greater accuracy than contemporary models such as XceptionNet,Meso-4,EfficientNet-BO,MesoInception-4,VGG-16,and DST-Net.The novelty of this investigation is the development of a new Convolutional Neural Network(CNN)learning model that can accurately detect deep fake face photos.

Actor-Critic-Based UAV-Assisted Data Collection in the Wireless Sensor Network

Wireless Sensor Network(WSN)is widely utilized in large-scale distributed unmanned detection scenarios due to its low cost and flexible installation.However,WSN data collection encounters challenges in scenarios lacking communication infrastructure.Unmanned aerial vehicle(UAV)offers a novel solution for WSN data collection,leveraging their high mobility.In this paper,we present an efficient UAV-assisted data collection algorithm aimed at minimizing the overall power consumption of the WSN.Firstly,a two-layer UAV-assisted data collection model is introduced,including the ground and aerial layers.The ground layer senses the environmental data by the cluster members(CMs),and the CMs transmit the data to the cluster heads(CHs),which forward the collected data to the UAVs.The aerial network layer consists of multiple UAVs that collect,store,and forward data from the CHs to the data center for analysis.Secondly,an improved clustering algorithm based on K-Means++is proposed to optimize the number and locations of CHs.Moreover,an Actor-Critic based algorithm is introduced to optimize the UAV deployment and the association with CHs.Finally,simulation results verify the effectiveness of the proposed algorithms.

Nursing, Psychological Intervention and Self-Management after Deep Brain Stimulation in Parkinson’s Disease

Parkinson’s disease (PD) is a neurodegenerative disease mainly caused by motor disorders, mostly occurring in middle-aged and elderly people. The incidence of PD has been increasing year by year, and up to now, PD is still an incurable disease. However, more and more data show that early implementation of deep brain stimulation and early medical, psychological, social and other interventions can significantly improve the quality of life and prolong the survival time of patients with Parkinson’s disease (PD). Mental health guidance, cognitive behavioral intervention, psychogenic therapy and scientific nursing for PD patients may improve the functional recovery after Deep Brain Stimulation (DBS) for Parkinson’s disease. This paper discusses the nursing and psychological intervention methods of deep brain stimulation (DBS) implantation in patients with Parkinson’s disease (PD), aiming to scientifically discuss the clinical effect of nursing psychological intervention and improve the quality of life in patients with Parkinson’s disease. Basic nursing and psychological cognitive behavior intervention measures for PD patients can improve the daily activity ability of PD patients, improve the outcome of PD patients, and effectively improve the satisfaction of PD patients with nursing work, which has certain clinical promotion significance.

Impact of the Built Environment on the Mental Health of the Inhabitants in High-Rise Collective Housing in Algeria

For everybody, the house or the accommodation （and its environment） is a secure place, a haven space, which protects people from constraints of the everyday life. Unfortunately, and more or less everywhere in the third world, accommodation is for a great number of people a source of stress caused by daily obligations that people have to deal with. In Algeria, the majority of the high-rise collective housing estates through the country offers all the ingredients of a constrained urban environment for the inhabitants. For the most of the population, the accommodation appears as a vital need rather than a negotiable good. As a matter of fact, in the third world in general and particularly in Algeria, most of people live in communities where there is a shortage of accommodation and in which the social housing and its environment are often below the standards. Constructed on policies and a conception of housing which does not integrate at all the criteria of the sustainable development, millions of fiats have already been built. Millions are going to be built in the future, and will be of high-rise collective type. As underlined in this paper, it seems reasonable to think that such a degraded built environment will be unfavorable to the inhabitants and will have a negative impact on both their mental and physical health. The attempt is to demonstrate that there are evidences according to which the housing conditions, inside and outside the accommodation, contribute to create psychological distress and physiological diseases.

Intergroup Evaluation as an Indicator of Emotional Elaboration of Collective Traumas in National Historical Narratives

In a longitudinal content analytical study, the authors explored intergroup evaluation patterns in Hungarian history school-book narratives about the so-called ＂Trianon Peace Treaty＂ in 1920 which had approved the detachment of 2/3 of Hungary＇s territory by victorious countries of the First World War. The event has meant a major national trauma that has not been elaborated to date. The study aimed to find evaluation patterns in temporally changing narrative constructions which were diagnostic to the process of emotional elaboration of the trauma. School-books released between 1920 and 2000 were included in the study, by a 10-year sampling method. Analysis was performed by NARRCAT （Narrative Categorial Content Analytical Tool）, a computerized tool for narrative psychological content analysis, which is capable for identifying complex linguistic structures of psychological relevance in large databases of narratives. Four different evaluation patterns emerged in the narratives which roughly correspond to four different historical eras in Hungary. Results show that the aggressor-victim relation between the former Entente powers and Hungary has remained a part of the narrative representation of the treaty, reflecting the identity state of collective victimhood.

Psychological Emotion Analysis System for Special Needs Children Based on Neural Network

With the development of modern society and the improvement of living standards,care for special needs children has been increasingly highlighted,and numerous corresponding measures such as welfare homes,special education schools,and youth care centers have emerged.Due to the lack of systematic emotional companionship,the mental health of special needs children are bound to be affected.Nowadays,emotional education,analysis,and evaluation are mostly done by psychologists and emotional analysts,and these measures are unpopular.Therefore,many researchers at home and abroad have focused on the solution of psychological issues and the psychological assessment and emotional analysis of such children in their daily lives.In this paper,a special children’s psychological emotional analysis based on neural network is proposed,where the system sends the voice information to a cloud platform through intelligent wearable devices.To ensure that the data collected are valid,a series of pretreatments such as Chinese word segmentation,de-emphasis,and so on are put into the neural network model.The model is based on the further research of transfer learning and Bi-GRU model,which can meet the needs of Chinese text sentiment analysis.The completion rate of the final model test has reached 97%,which means that it is ready for use.Finally,a web page is designed,which can evaluate and detect abnormal psychological state,and at the same time,a personal emotion database can also be established.

双碳背景下铝电解绿色低碳技术体系构建

沈阳院铝电解绿色深度节能技术是针对国家双碳战略和行业阶梯电价背景而形成的一整套技术体系。其从铝电解工艺本质切入,着手于磁流体稳定性提升、良好热平衡维持和系统性节能降耗三个方向,包含了磁场原位升级技术、网络化自均衡母线技术、节能长寿命阴极及内衬技术、双侧精准集气技术以及相配套的生产操作及管理技术等。该技术体系已经在超过45条电解系列上应用,涉及了从180到600 kA的所有国内电解槽型,直流电耗最低达到了12200 kWh/t-Al以下。该技术的应用有助于缓解电解铝生产企业的节能降碳压力,同时提高电解生产的安全性和稳定性。

心理护理干预在儿童入托体检静脉采血中的应用

目的探讨心理护理干预在儿童入托体检静脉采血中的应用效果。方法选取2022年5月至2023年5月于寿光市人民医院进行入托体检静脉采血儿童100例作为研究对象,按照抽签法将其分为参照组(50例)与试验组(50例)。参照组采取常规静脉采血护理方案,试验组应用心理护理干预模式。比较两组疼痛程度、采血满意度。结果试验组疼痛程度评分低于参照组,采血满意度评分高于参照组,差异有统计学意义(P<0.05)。结论儿童入托体检静脉采血中,应用心理护理干预的效果明显,可减轻儿童的疼痛程度,提高采血满意度。

以风险评估为基础的集束化护理结合针对性心理干预在子宫全切术患者中的应用效果

目的探讨以风险评估为基础的集束化护理结合针对性心理干预在子宫全切术患者中的应用效果。方法选择2020年2月至2023年4月于滨州医学院附属医院行子宫全切术的80例患者,按照随机数字表法分为两组。观察组(40例)实施以风险评估为基础的集束化护理结合针对性心理干预,对照组(40例)实施常规护理,比较两组下肢深静脉血栓(DVT)发生率、心理状态。结果观察组下肢DVT发生率低于对照组(P<0.05);护理后观察组心理状态各评分较对照组更低(P<0.05)。结论以风险评估为基础的集束化护理结合针对性心理干预在子宫全切术患者中的应用效果较好,可改善心理状态,降低术后下肢DVT发生率。

Gloo+:利用在网计算技术加速分布式深度学习训练

在分布式深度学习训练中,聚合通信是主要的通信方式。在聚合通信优化的研究中,有软件层面的优化和硬件层面的优化。SHARP是Mellanox提出来的一种聚合通信网络卸载协议,是针对聚合通信在硬件上的优化,其将聚合操作卸载到网络中的交换机,进而缩短了聚合通信时间。在Gloo的基础上集成了SHARP技术,设计并实现了一个能够利用在网计算技术来加速分布式深度学习训练的聚合通信库——Gloo+。评估并比较了Gloo+、Gloo以及MPI中聚合操作的性能,并将Gloo+应用于分布式深度学习训练中,以此来检验其实战能力。对Gloo+的实验评估结果显示,在基准测试时,在消息大小较小的情况下,Gloo+相对于Gloo的加速比最高能达到100以上;相比于以太网模式下的MPI,其加速比最高也能达到50以上;相比于IB网模式下的MPI,其加速比在10以内。在分布式深度学习训练的实际应用中,Gloo+相比于Gloo加速比最高能达到1.1,相比于以太网模式下的MPI加速比最高有1.3,相比于IB网模式下的MPI加速比最高有0.5。

深海重要矿产资源开发研究进展

深海蕴藏了丰富的能源与矿产资源,包括多金属结核、多金属硫化物、富钴结壳、深海稀土、天然气水合物等,是人类社会未来可持续发展的重要储备。当前,社会迅猛发展对能源和关键矿产资源的需求持续增长,引发了又一轮深海矿产资源开发热潮。全球范围内,已发现的深海矿产资源主要集中于太平洋、大西洋和印度洋的国际海底区域,以及沿海国家专属经济区内的海底。与陆地矿床相比,深海矿床具有品位高、种类多、开采地点远离居民生活区、且易于运输转移的优势,但采矿又势必会影响海洋生物生存环境,破坏深海生态系统稳定性。如不加强对环境影响的评估和开采技术的研究,并制定相关的法律法规,未来的深海采矿对海洋环境的破坏将无法估量。鉴于此,本文从深海典型矿产资源开采现状、开采技术研究进展以及对深海生态系统影响等方面进行了系统梳理,并对相关制约因素进行了讨论并提出合理性建议,以期为未来深海采矿技术方案的制定提供有益借鉴。

不同水力集矿模型的矿粒采集及环境扰动特性数值研究

实现高效、低扰动海底集矿是深海多金属结核矿产资源开发面临的关键技术挑战.海底矿粒采集过程的力学行为十分复杂,是一个涉及三维紊动流动、离散粗颗粒运动与细颗粒土体破坏的流固土多物理场耦合过程.针对吸扬式、附壁射流式、射流冲采式3种主流的深海水力集矿方法,基于K-Epsilon两层模型和离散元模型分别模拟集矿液相湍流和固相矿粒,进行矿粒采集性能和环境扰动的数值模拟研究,分析集矿流量q m和集矿头拖曳速度v对矿粒采集率η、集矿流场湍动能k和海水-沉积物混合物体积分数φ的影响,探究集矿流场中流速、压力和矿粒的分布特征.结果表明:在相同q m和v下,射流冲采式模型获得的η最大,吸扬式模型获得的η最小;射流冲采式模型对近底流场扰动程度最大,深海沉积物扩散现象也最为明显,而吸扬式和附壁射流式模型造成的环境扰动程度较小,更有利于环保要求;附壁射流式模型的η对q m和v最不敏感,且该集矿模型可以较好地兼顾高效矿粒采集和低环境扰动.研究结果可为揭示集矿机理和设计研发高效低扰动集矿装置提供科学依据.

基于PycModel高效深度学习模型的心理咨询问题分类探究

大众心理健康问题日益受到广泛关注。为提高心理医疗资源的使用效率,采用互联网手段采集与心理咨询相关的数据,使用一种融合脚本筛选、数据整理以及利用Easydata进行主动学习式数据标注的新方法构建数据集,结合卷积神经网络和BERT预训练模型等技术,在textvec-base-chinese模型的基础上,提出了PycModel模型,以实现更加高效的心理咨询问题分类。实验结果显示,PycModel在心理咨询问题分类的准确率明显优于其他参照模型,该模型能够有效提高心理咨询的效率和效果,可以为心理健康服务提供有力的支持。

基于协同卷积的轻量级行为检测方法

时空行为检测是计算机视觉领域重要的研究方向,为了减小模型体量,提高检测速度,提出一种基于协同卷积(collective convolution, CoConv)的轻量级行为检测方法。将视频的时序信息转换为时空图像(spatio-temporal image, STI),利用协同卷积获取相同位置不同时间的时空特征信息。在YOLOv5的基础上将骨干网络和检测头部替换为协同卷积模块构建时空行为检测网络结构,通过后处理对时空图像的检测结果进行连接,快速形成视频结果,提高网络的行为检测性能。实验结果表明,提出的方法可以在保证准确率和不增加参数量的情况下,减少网络计算量,提高网络检测速度,且优于现有的行为检测方法。

患者的自主与同一性分析

个体同一性一直是医学伦理中的核心问题之一。深入探讨这一问题会发现对于患者自主的看法在一定程度上依赖于一个特定观点,即患者将以足够的程度保持与未来自我的心理联系。这意味着诸如信仰、欲望和价值观等心理因素会随着时间的推移而保持稳定,或者以某种方式与先前的心理因素联系在一起。意识到心理联系对于实现这一目标的重要性,并不需要在时间维度上对个体的本质做出形而上学的断言,即无需支持基于心理联系的个体同一性的形而上学理论。然而,如果假设患者在某种程度上会与未来自我保持心理上的联系,那么尊重患者自我决定的价值和道德权威才能得到理解。

自监督非等距三维模型簇对应关系计算方法

针对现有非等距三维模型簇对应关系计算方法准确率较低且泛化能力较差的问题,文中提出采用深度内外特征对齐算法的自监督非等距三维模型簇对应关系计算方法.首先,利用DiffusionNet直接学习原始的三维模型特征,获取具有鉴别力的特征描述符.然后,使用深度内外特征对齐算法计算非等距模型对之间的对应关系,采用局部流形调和基作为模型本征信息,并结合笛卡尔坐标等外部信息,实现内外部信息对齐一致,以无监督的方式自动生成对应结果.最后,构建非等距模型簇的加权无向图,根据相似几何模型存在内在相关性的原则,设计自监督多模型匹配算法,不断增强模型图最短路径的循环一致性,获得最优的非等距三维模型簇对应关系.实验表明,文中方法对应关系测地误差较小,结果较准确,能处理模型自身对称性影响对应关系计算的问题,具有良好的泛化能力.

无人机协同空天地一体化战场信息采集策略研究

围绕空天地一体化战场信息采集场景,建立优化目标为最大化系统效率和性能的信道分配和设备匹配模型。针对信道分配和设备匹配决策变量具有相关和耦合性的特点,提出一种基于遗传算法和深度强化学习的两阶段资源分配(GD2A)算法。通过仿真验证,分析超参数对GD2A算法的性能和收敛速度的影响。对比实验表明,GD2A算法能在6G空天地异构网络中保障军事通信的可靠性和安全性,最大化系统效用。

基于人工智能的新时代大学生心理障碍预测研究

心理障碍是影响新时代大学生的首要健康问题,目前尚无可用于筛查大学生心理障碍的早期预测工具。人工智能技术的迅速发展为心理健康领域的研究提供新的思路和方法。本研究基于浙江省不同层次高校在校生的样本数据资料,通过构建CNN-LSTM深度学习模型来识别大学生心理障碍的风险因素,并基于目标预测变量将CNN-LSTM与SVM、BP和CNN构建的模型进行比较,通过对比AUC等多个预测效果评价指标,最终验证构建的CNN-LSTM模型在大学生心理障碍预测上表现出最好的性能,且具有实际用于筛查大学生心理障碍的应用潜力。

Using 3D TCAD Simulation to Study Charge Collection of a p-n Junction in a 0.18μm Bulk Process

Single event transient of a real p-n junction in a 0.18μm bulk process is studied by 3D TCAD simulation. The impact of voltage, temperature, substrate concentration, and LET on SET is studied. Our simulation results demonstrate that biases in the range 1.62 to 1.98V influence DSET current shape greatly and total collected charge weakly. Peak current and charge collection within 2ns decreases as temperature increases,and temperature has a stronger influence on SET currents than on total charge. Typical variation of substrate concentration in modern VDSM processes has a negligible effect on SEEs. Both peak current and total collection charge increases as LET increases.

RISE-D3QN驱动的多无人机数据采集路径规划

无人机辅助物联网数据采集是高效且具有前景的方法。针对路径规划的优化资源分配问题,细化了电量消耗模型,并考虑了三个指标:数据量、时间效率和能源效率。该问题被建模为分布式局部可观测马尔可夫决策过程,并提出一种深度强化学习算法。具体地,将归一化的模型分为四个具体地的无人机电量消耗模型;基于离散动作离线深度强化学习架构,提出一种新的RISE(Rényi state entropy)-D3QN(dueling double deep Q network)算法,结合了内在奖励、优先经验回放和soft-max探索策略,可在无人机电池容量、物联网设备位置、物联网设备数据量、物联网设备数量发生变化的同时规划无人机群的路径。仿真结果表明,相比于传统的D3QN算法以及传统的DQN算法,在确保无人机安全飞行的同时,提高了无人机从物联网设备采集的数据量,并在以此为主要目标的情况下减少了无人机的飞行时间以及能量消耗。