Efficient User Identity Linkage Based on Aligned Multimodal Features and Temporal Correlation

User identity linkage(UIL)refers to identifying user accounts belonging to the same identity across different social media platforms.Most of the current research is based on text analysis,which fails to fully explore the rich image resources generated by users,and the existing attempts touch on the multimodal domain,but still face the challenge of semantic differences between text and images.Given this,we investigate the UIL task across different social media platforms based on multimodal user-generated contents(UGCs).We innovatively introduce the efficient user identity linkage via aligned multi-modal features and temporal correlation(EUIL)approach.The method first generates captions for user-posted images with the BLIP model,alleviating the problem of missing textual information.Subsequently,we extract aligned text and image features with the CLIP model,which closely aligns the two modalities and significantly reduces the semantic gap.Accordingly,we construct a set of adapter modules to integrate the multimodal features.Furthermore,we design a temporal weight assignment mechanism to incorporate the temporal dimension of user behavior.We evaluate the proposed scheme on the real-world social dataset TWIN,and the results show that our method reaches 86.39%accuracy,which demonstrates the excellence in handling multimodal data,and provides strong algorithmic support for UIL.

Enhanced Temporal Correlation for Universal Lesion Detection

Universal lesion detection(ULD)methods for computed tomography(CT)images play a vital role in the modern clinical medicine and intelligent automation.It is well known that single 2D CT slices lack spatial-temporal characteristics and contextual information compared to 3D CT blocks.However,3D CT blocks necessitate significantly higher hardware resources during the learning phase.Therefore,efficiently exploiting temporal correlation and spatial-temporal features of 2D CT slices is crucial for ULD tasks.In this paper,we propose a ULD network with the enhanced temporal correlation for this purpose,named TCE-Net.The designed TCE module is applied to enrich the discriminate feature representation of multiple sequential CT slices.Besides,we employ multi-scale feature maps to facilitate the localization and detection of lesions in various sizes.Extensive experiments are conducted on the DeepLesion benchmark demonstrate that thismethod achieves 66.84%and 78.18%for FS@0.5 and FS@1.0,respectively,outperforming compared state-of-the-art methods.

Layer chicken microbiota:a comprehensive analysis of spatial and temporal dynamics across all major gut sections

Background The gut microbiota influences chicken health,welfare,and productivity.A diverse and balanced microbiota has been associated with improved growth,efficient feed utilisation,a well-developed immune system,disease resistance,and stress tolerance in chickens.Previous studies on chicken gut microbiota have predominantly focused on broiler chickens and have usually been limited to one or two sections of the digestive system,under con-trolled research environments,and often sampled at a single time point.To extend these studies,this investigation examined the microbiota of commercially raised layer chickens across all major gut sections of the digestive system and with regular sampling from rearing to the end of production at 80 weeks.The aim was to build a detailed picture of microbiota development across the entire digestive system of layer chickens and study spatial and temporal dynamics.Results The taxonomic composition of gut microbiota differed significantly between birds in the rearing and pro-duction stages,indicating a shift after laying onset.Similar microbiota compositions were observed between proven-triculus and gizzard,as well as between jejunum and ileum,likely due to their anatomical proximity.Lactobacil-lus dominated the upper gut in pullets and the lower gut in older birds.The oesophagus had a high proportion of Proteobacteria,including opportunistic pathogens such as Gallibacterium.Relative abundance of Gallibacterium increased after peak production in multiple gut sections.Aeriscardovia was enriched in the late-lay phase compared to younger birds in multiple gut sections.Age influenced microbial richness and diversity in different organs.The upper gut showed decreased diversity over time,possibly influenced by dietary changes,while the lower gut,specifi-cally cecum and colon,displayed increased richness as birds matured.However,age-related changes were inconsist-ent across all organs,suggesting the influence of organ-specific factors in microbiota maturation.Conclusion Addressing a gap in previous research,this study explored the microbiota across all major gut sections and tracked their dynamics from rearing to the end of the production cycle in commercially raised layer chickens.This study provides a comprehensive understanding of microbiota structure and development which help to develop targeted strategies to optimise gut health and overall productivity in poultry production.

Spatial compartmentalization and temporal stability of associated microbiota in Pacific oyster Crassostrea gigas

The Pacific oyster Crassostrea gigas,one of the most exploited molluscs in the world,has suffered from massive mortality in recent decades,and the occurrence mechanisms have not been well characterized.In this study,to reveal the relationship of associated microbiota to the fitness of oysters,temporal dynamics of microbiota in the gill,hemolymph,and hepatopancreas of C.gigas during April 2018-January 2019 were investigated by 16 S rRNA gene sequencing.The microbiota in C.gigas exhibited tissue heterogeneity,of which Spirochaetaceae was dominant in the gill and hemolymph while Mycoplasmataceae enriched in the hepatopancreas.Co-occurrence network demonstrated that the gill microbiota exhibited higher inter-taxon connectivity while the hemolymph microbiota had more modules.The richness(Chao 1 index)and diversity(Shannon index)of microbial community in each tissue showed no significant seasonal variations,except for the hepatopancreas having a higher richness in the autumn.Similarly,beta diversity analysis indicated a relatively stable microbiota in each tissue during the sampling period,showing relative abundance of the dominant taxa exhibiting temporal dynamics.Results indicate that the microbial community in C.gigas showed a tissue-specific stability with temporal dynamics in the composition,which might be essential for the tissue functioning and environmental adaption in oysters.This work provides a baseline microbiota in C.gigas and is helpful for the understanding of host-microbiota interaction in oysters.

Spatial and temporal variation of water clarity in typical reservoirs in the Beijing-Tianjin-Hebei region observed by GF 1-WFV satellite data

Rapidly monitoring regional water quality and the changing trend is of great practical and scientific significance,especially for the Beijing-Tianjin-Hebei(BTH)region of China where water resources are relatively scarce and inland water bodies are generally small.The remote sensing data of the GF 1 satellite launched in 2013 have characteristics of high spatial and temporal resolution,which can be used for the dynamic monitoring of the water environment in small lakes and reservoirs.However,the water quality remote sensing monitoring model based on the GF 1 satellite data for lakes and reservoirs in BTH is still lacking because of the considerable differences in the optical characteristics of the lakes and reservoirs.In this paper,the typical reservoirs in BTH-Guanting Reservoir,Yuqiao Reservoir,Panjiakou Reservoir,and Daheiting Reservoir are taken as the study areas.In the atmospheric correction of GF 1-WFV,the relative radiation normalized atmospheric correction was adopted after comparing it with other methods,such as 6 S and FLAASH.In the water clarity retrieval,a water color hue angle based model was proposed and outperformed other available published models,with the R 2 of 0.74 and MRE of 31.7%.The clarity products of the four typical reservoirs in the BTH region in 2013-2019 were produced using the GF 1-WFV data.Based on the products,temporal and spatial changes in clarity were analyzed,and the main influencing factors for each water body were discussed.It was found that the clarity of Guanting,Daheiting,and Panjiakou reservoirs showed an upward trend during this period,while that of Yuqiao Reservoir showed a downward trend.In the influencing factors,the water level of the water bodies can be an important factor related to the water clarity changes in this region.

Spatial correlations in time and frequency domains between chlorophyll-a concentration and environmental factors in the Bohai Sea

Based on the reconstructed MODIS data and ECMWF reanalysis data from 2003 to 2021,spatial correlations between chlorophyll a(Chl a)and sea surface temperature(SST),photosynthetically available radiation(PAR),aerosol optical thickness(AOT),and wind speed(WS)in the Bohai Sea were analyzed from the perspective of time domain and frequency domain.Results indicate that the frequency domain analysis was more conducive to revealing the correlations between Chl a and environmental factors.The spatial pattern of time-domain correlations was similar to the isobaths of the Bohai Sea,which was positive in shallow waters and negative in deep waters for SST,PAR,and AOT,and was reversed for WS.Frequency-domain correlations were obtained by performing Fourier Transform and were higher than correlations in time domain.The spatial distributions indicated that the effects of SST and PAR on Chl a were greater than AOT and WS in the Bohai Sea.Additionally,cross-spectrum analysis was applied to explore the response relationships.A depth-dependent pattern was shown in correlations and time lags,indicating that the influential mechanism of environmental factors on Chl-a concentration is related to seawater depth.

Spatial correlations in ground motion intensity measuring from the 2023 Türkiye earthquake

When evaluating an area's seismic risk or resilience,it is necessary to use the spatial correlation to analyze the ground motion parameters of multiple sites together in an earthquake.These two large earthquakes in Türkiye provided the possibility for spatial correlation analysis of ground motion intensity measurements in this area.Based on the strong motion records provided by The Disaster and Emergency Management Authority of Türkiye(AFAD),this study uses the local ground motion prediction equation in Türkiye to give spatial correlation analysis of Intensity Measurements.This study gives an exponential model based on a semivariogram and compares it with the correlation model obtained from previous studies.

Temporal and spatial differences in human activities performed in Urban Green Spaces of Vilnius(Lithuania)

Urban green spaces(UGS)are relevant to city well-being,as recognized by the United Nations’Sustainable Development Goals(SDGs).However,few studies have studied the temporal use of UGS.This work assessed the seasonal,weekly,and daily use of three urban green spaces(Vingis Park,Bernardino Garden,and Jomantas Park)in Vilnius(Lithuania).The study is based on an on-site observation-based survey,which recorded users’characteristics,activities,and weather conditions during summer and winter.The results showed that UGS’s seasonal,weekly,and daily use differed according to park and users’characteristics.Parks with a higher diversity of facilities had a high seasonal difference in the number of observed activities.User numbers were higher in the summer for activities with children,social activities,sports,and water activities than in the winter.Jomantas Park had the lowest variability in user characteristics.Weather variables were linked to changes in users’activities.Higher precipitation and lower temperature were associated with reducing the number of users and the diversity of registered activities.Most of the stationary activities were observed during summer.The diversity of the observed activities was associated with the available facilities rather than the park size.The distribution of stationary activities was spatially correlated with facility/equipment(benches,playgrounds,sports,and fitness equipment)and proximity to water features.The results of this study are relevant for UGS design,planning,and management.

Temporal and Spatial Dynamics of Vaccine-Derived Poliovirus (VDPV) in Democratic Republic of Congo from 2018 to 2023

Background: The Democratic Republic of Congo (DRC) has been facing outbreaks of VDPV since 2017. These wild poliovirus variants are responsible for poliomyelitis, which is in the process of eradication.. In the following lines, we try to show the evolution of VDPV cases across the country in order to understand their chronological dynamics and seasonal influence. Methods: We conducted a cross-sectional study of of VDPV notified in the DRC from 2018 to 2023. Maps of the spatial dynamics of VDPV cases were produced from attack rates with QGIS® (3.22.8). As for temporal dynamics, time series were decomposed and presented in the form of graphs showing the chronological evolution of VDPV cases and their seasonal trend, using R.4.0 software package. Results: A total of 1196 Cases of VDPV types 1, 2 and 3 were recorded in the biological confirmation databases of the INRB and the Expanded Program of Immunization during the study period across25 provinces. The eastern part of the country reporting the most cases. The general trend is upwards, with a peak in 2022 of 527 cases, whereas in 2021 there was a notable drop of 31 cases. Analysis of the temporal breakdown suggests a seasonal pattern, with peaks between the months of September and December, considered being rainy periods in some provinces. Conclusion: During the 6 years of our study (2018 - 2023) almost all the Health Zones were hit by VDPV epidemics. The eastern part was the most impacted. The seasonal component is well marked suggesting a rise in detection in the rainy season and during pivotal periods of climate change.

Spatial and Temporal Distribution Characteristics of Snowstorm in Ulanqab City in Recent 30 Years

Based on the data of daily snowfall and weather phenomena of 11 national meteorological stations in Ulanqab City from 1991 to 2020,the spatial and temporal distribution characteristics of snowstorm were analyzed.The results show that the snowstorm in Ulanqab had obvious seasonal distribution characteristics,mainly happening in spring(March-May)and autumn(September-November).It also had obvious regional distribution in space,and the snowstorm center appeared in Chahar Right Wing Middle Banner and Jining District,namely the east side of the Yinshan Mountain.In the past 30 years,the amount of snowstorm in the whole year in Ulanqab showed a certain fluctuation trend,and the number of snowstorm days had shown an obvious upward trend since 2011.

Spatiotemporal Prediction of Urban Traffics Based on Deep GNN

Traffic prediction already plays a significant role in applications like traffic planning and urban management,but it is still difficult to capture the highly non-linear and complicated spatiotemporal correlations of traffic data.As well as to fulfil both long-termand short-termprediction objectives,a better representation of the temporal dependency and global spatial correlation of traffic data is needed.In order to do this,the Spatiotemporal Graph Neural Network(S-GNN)is proposed in this research as amethod for traffic prediction.The S-GNN simultaneously accepts various traffic data as inputs and investigates the non-linear correlations between the variables.In terms of modelling,the road network is initially represented as a spatiotemporal directed graph,with the features of the samples at the time step being captured by a convolution module.In order to assign varying attention weights to various adjacent area nodes of the target node,the adjacent areas information of nodes in the road network is then aggregated using a graph network.The data is output using a fully connected layer at the end.The findings show that S-GNN can improve short-and long-term traffic prediction accuracy to a greater extent;in comparison to the control model,the RMSE of S-GNN is reduced by about 0.571 to 9.288 and the MAE(Mean Absolute Error)by about 0.314 to 7.678.The experimental results on two real datasets,Pe MSD7(M)and PEMS-BAY,also support this claim.

Spatiotemporal Distributions of Bacterioplankton Communities in the Qiantang River(Hangzhou Section),China

The spatial and temporal distribution of bacterioplankton communities plays a vital role in understanding the ecological dynamics and health of aquatic ecosystems.In this study,we conducted a comprehensive investigation of the bacterioplankton communities in the Qiantang River(Hangzhou section).Water samples were collected quarterly from seven sites over a one-year period.Physical and chemical parameters,including dissolved oxygen(DO),water temperature(WT),chemical oxygen demand(COD),nitrite(NO_(2)^(-)),active phosphate(PO_(4)^(3-))and other indices were determined.In this study,theαdiversity,βdiversity and abundance differences of bacterial communities were investigated using 16S rRNA high-throughput sequencing analysis.The spatial and temporal distribution characteristics and main driving factors of the bacterioplankton community structure and diversity were discussed.The results showed that a total of 57 phyla were detected in the bacterioplankton community,among which Proteobacteria and Actinomycetes were the two dominant groups with the highest relative abundance.The results of PCoA based on Bray-Curtis distance showed that the sampling season had a slightly greater effect on the changes in bacterioplankton community structure in the Qiantang River.Mantel and partial Mantel test showed that environmental variables(Mantel r=0.6739,P<0.0001;partial Mantel r=0.507,P=0.0001)were more important than geographical distance(Mantel r=0.5322,P<0.001;partial Mantel r=0.1563,P=0.001).The distance attenuation model showed that there was significant distance attenuation in all four seasons,and the maximum limit of bacterial community diffusion was found in spring.RDA analysis showed that nine environmental factors,including COD,WT and DO,significantly affected community distribution(P<0.05).This study provides valuable insights into the spatial and temporal distribution characteristics of bacterioplankton communities,shedding light on their ecological roles in the Qiantang River.The information obtained can guide future research on the interactions between bacterioplankton and their environment,enabling the development of effective conservation strategies and sustainable management practices for aquatic ecosystems.

Employing temporal channel correlation in user selection for MIMO broadcast systems

Taking the time varying nature of wireless channels into account, two user selection schemes with lower complexity are developed for multiple-input multiple-output broadcast （MIMO BC）systems. According to the relationship between coherence time and Doppler frequency, an information frame is divided into several segments. At the beginning of each segment, the user selection is carded out with the greedy selection algorithm. In the simplified user selection algorithms employing the temporal correlation （SUSTC）, the selection results are applied for all the remaining slots in each segment. But in the improved simplified user selection algorithms employing the temporal correlation（ISUSTC）, at the remaining slots, users are kept with favorable channel conditions selected at the previous slot, and other users are updated from the candidate pool to communicate simultaneously. Simulations show that compared with the greedy user selection method, the proposed algorithms can reduce the selection complexity with a little sum capacity loss.

The Spatiotemporal Distribution Characteristics of Cloud Types and Phases in the Arctic Based on CloudSat and CALIPSO Cloud Classification Products

The cloud type product 2B-CLDCLASS-LIDAR based on CloudSat and CALIPSO from June 2006 to May 2017 is used to examine the temporal and spatial distribution characteristics and interannual variability of eight cloud types(high cloud, altostratus, altocumulus, stratus, stratocumulus, cumulus, nimbostratus, and deep convection) and three phases(ice,mixed, and water) in the Arctic. Possible reasons for the observed interannual variability are also discussed. The main conclusions are as follows:(1) More water clouds occur on the Atlantic side, and more ice clouds occur over continents.(2)The average spatial and seasonal distributions of cloud types show three patterns: high clouds and most cumuliform clouds are concentrated in low-latitude locations and peak in summer;altostratus and nimbostratus are concentrated over and around continents and are less abundant in summer;stratocumulus and stratus are concentrated near the inner Arctic and peak during spring and autumn.(3) Regional averaged interannual frequencies of ice clouds and altostratus clouds significantly decrease, while those of water clouds, altocumulus, and cumulus clouds increase significantly.(4) Significant features of the linear trends of cloud frequencies are mainly located over ocean areas.(5) The monthly water cloud frequency anomalies are positively correlated with air temperature in most of the troposphere, while those for ice clouds are negatively correlated.(6) The decrease in altostratus clouds is associated with the weakening of the Arctic front due to Arctic warming, while increased water vapor transport into the Arctic and higher atmospheric instability lead to more cumulus and altocumulus clouds.

Analysis of Spatiotemporal Changes in Integrated Land Productivity in Chongqing from 1997 to 2023

[Objectives]Integrated land productivity can reflect the comprehensive utilization of land and the overall output level,which is the most basic and commonly used indicator in assessing land use efficiency.This thesis aims to analyze the spatial and temporal changes of integrated land productivity in Chongqing from 1997 to 2023 in order to assess its land use efficiency.[Methods]This study measured the integrated land productivity of Chongqing Municipality,the only municipality directly under the central government in the western part of China,over the past 26 years(1997-2023)through relevant surveys and statistical data,and analyzed in depth the integrated land productivity of the 38 districts and counties under the jurisdiction of Chongqing,as well as the functional sub-districts of the"one district and two clusters"and the"one district and two clusters"in Chongqing.It also analyzes the characteristics of spatial and temporal differences in land productivity in 38 districts and counties under the jurisdiction of Chongqing and"one district and two clusters".[Results]The results of the study show that over the past 26 years,the integrated land productivity of Chongqing has shown an annual growth trend,and the integrated land productivity of the 38 districts and counties and the functional subregions of"one district and two clusters"has also increased significantly,but the average annual growth rate of the integrated land productivity varies among different regions.From the perspective of spatial differences,there are significant differences in land productivity among the 38 districts and counties of Chongqing and the functional subregions of"one district and two clusters",which are mainly due to the different natural conditions,economic development levels and functional positioning of each region.[Conclusions]Based on the results of the study and the actual situation of Chongqing,this paper puts forward the leading measures to improve the integrated land productivity,with a view to providing a reference basis for Chongqing to improve the efficiency of land use and promote the sustainable use of land resources.

Spatiotemporal evolution and influencing factors of urban resilience in the Yellow River Basin,China

The Yellow River Basin of China is a key region that contains myriad interactions between human activities and natural environment.Industrialization and urbanization promote social-economic development,but they also have generated a series of environmental and ecological issues in this basin.Previous researches have evaluated urban resilience at the national,regional,urban agglomeration,city,and prefecture levels,but not at the watershed level.To address this research gap and elevate the Yellow River Basin’s urban resilience level,we constructed an urban resilience evaluation index system from five dimensions:industrial resilience,social resilience,environmental resilience,technological resilience,and organizational resilience.The entropy weight method was used to comprehensively evaluate urban resilience in the Yellow River Basin.The exploratory spatial data analysis method was employed to study the spatiotemporal differences in urban resilience in the Yellow River Basin in 2010,2015,and 2020.Furthermore,the grey correlation analysis method was utilized to explore the influencing factors of these differences.The results of this study are as follows:(1)the overall level of urban resilience in the Yellow River Basin was relatively low but showed an increasing trend during 2010–2015,and significant spatial distribution differences were observed,with a higher resilience level in the eastern region and a low-medium resilience level in the western region;(2)the differences in urban resilience were noticeable,with industrial resilience and social resilience being relatively highly developed,whereas organizational resilience and environmental resilience were relatively weak;and(3)the correlation ranking of resilience influencing factors was as follows:science and technology level>administrative power>openness>market forces.This research can provide a basis for improving the resilience level of cities in the Yellow River Basin and contribute to the high-quality development of the region.

Flight parameter calculation method of multi-projectiles using temporal and spatial information constraint

The dynamic parameters of multiple projectiles that are fired using multi-barrel weapons in highfrequency continuous firing modes are important indicators to measure the performance of these weapons.The characteristics of multiple projectiles are high randomness and large numbers launched in a short period of time,making it very difficult to obtain the real dispersion parameters of the projectiles due to the occlusion or coincidence of multiple projectiles.Using six intersecting-screen testing system,in this paper,we propose an association recognition and matching algorithm of multiple projectiles using a temporal and spatial information constraint mechanism.We extract the output signal from each detection screen and then use the wavelet transform to process the output signal.We present a method to identify and extract the time values on which the projectiles pass through the detection screens using the wavelet transform modulus maximum theory.We then use the correlation of the output signals of three parallel detection screens to establish a correlation coefficient recognition constraint function for the multiple projectiles.Based on the premise of linear projectile motion,we establish a temporal and spatial constraint matching model using the projectile’s position coordinates in each detection screen and the projectile’s time constraints within the multiple intersecting-screen geometry.We then determine the time values of the multiple projectiles in each detection screen using an iterative search cycle registration,and finally obtain the flight parameters for the multiple projectiles in the presence of uncertainty.The proposed method and algorithm were verified experimentally and can solve the problem of uncertainty in projectiles flight parameter under different multiple projectile firing states.

Spatial Correlation Module for Classification of Multi-Label Ocular Diseases Using Color Fundus Images

To prevent irreversible damage to one’s eyesight,ocular diseases(ODs)need to be recognized and treated immediately.Color fundus imaging(CFI)is a screening technology that is both effective and economical.According to CFIs,the early stages of the disease are characterized by a paucity of observable symptoms,which necessitates the prompt creation of automated and robust diagnostic algorithms.The traditional research focuses on image-level diagnostics that attend to the left and right eyes in isolation without making use of pertinent correlation data between the two sets of eyes.In addition,they usually only target one or a few different kinds of eye diseases at the same time.In this study,we design a patient-level multi-label OD(PLML_ODs)classification model that is based on a spatial correlation network(SCNet).This model takes into consideration the relevance of patient-level diagnosis combining bilateral eyes and multi-label ODs classification.PLML_ODs is made up of three parts:a backbone convolutional neural network(CNN)for feature extraction i.e.,DenseNet-169,a SCNet for feature correlation,and a classifier for the development of classification scores.The DenseNet-169 is responsible for retrieving two separate sets of attributes,one from each of the left and right CFI.After then,the SCNet will record the correlations between the two feature sets on a pixel-by-pixel basis.After the attributes have been analyzed,they are integrated to provide a representation at the patient level.Throughout the whole process of ODs categorization,the patient-level representation will be used.The efficacy of the PLML_ODs is examined using a soft margin loss on a dataset that is readily accessible to the public,and the results reveal that the classification performance is significantly improved when compared to several baseline approaches.

Using BlazePose on Spatial Temporal Graph Convolutional Networks for Action Recognition

The ever-growing available visual data(i.e.,uploaded videos and pictures by internet users)has attracted the research community’s attention in the computer vision field.Therefore,finding efficient solutions to extract knowledge from these sources is imperative.Recently,the BlazePose system has been released for skeleton extraction from images oriented to mobile devices.With this skeleton graph representation in place,a Spatial-Temporal Graph Convolutional Network can be implemented to predict the action.We hypothesize that just by changing the skeleton input data for a different set of joints that offers more information about the action of interest,it is possible to increase the performance of the Spatial-Temporal Graph Convolutional Network for HAR tasks.Hence,in this study,we present the first implementation of the BlazePose skeleton topology upon this architecture for action recognition.Moreover,we propose the Enhanced-BlazePose topology that can achieve better results than its predecessor.Additionally,we propose different skeleton detection thresholds that can improve the accuracy performance even further.We reached a top-1 accuracy performance of 40.1%on the Kinetics dataset.For the NTU-RGB+D dataset,we achieved 87.59%and 92.1%accuracy for Cross-Subject and Cross-View evaluation criteria,respectively.

Deepfake Video Detection Based on Improved CapsNet and Temporal–Spatial Features

Rapid development of deepfake technology led to the spread of forged audios and videos across network platforms,presenting risks for numerous countries,societies,and individuals,and posing a serious threat to cyberspace security.To address the problem of insufficient extraction of spatial features and the fact that temporal features are not considered in the deepfake video detection,we propose a detection method based on improved CapsNet and temporal–spatial features(iCapsNet–TSF).First,the dynamic routing algorithm of CapsNet is improved using weight initialization and updating.Then,the optical flow algorithm is used to extract interframe temporal features of the videos to form a dataset of temporal–spatial features.Finally,the iCapsNet model is employed to fully learn the temporal–spatial features of facial videos,and the results are fused.Experimental results show that the detection accuracy of iCapsNet–TSF reaches 94.07%,98.83%,and 98.50%on the Celeb-DF,FaceSwap,and Deepfakes datasets,respectively,displaying a better performance than most existing mainstream algorithms.The iCapsNet–TSF method combines the capsule network and the optical flow algorithm,providing a novel strategy for the deepfake detection,which is of great significance to the prevention of deepfake attacks and the preservation of cyberspace security.