Micro-expressions(ME)recognition is a complex task that requires advanced techniques to extract informative features fromfacial expressions.Numerous deep neural networks(DNNs)with convolutional structures have been pr...Micro-expressions(ME)recognition is a complex task that requires advanced techniques to extract informative features fromfacial expressions.Numerous deep neural networks(DNNs)with convolutional structures have been proposed.However,unlike DNNs,shallow convolutional neural networks often outperform deeper models in mitigating overfitting,particularly with small datasets.Still,many of these methods rely on a single feature for recognition,resulting in an insufficient ability to extract highly effective features.To address this limitation,in this paper,an Improved Dual-stream Shallow Convolutional Neural Network based on an Extreme Gradient Boosting Algorithm(IDSSCNN-XgBoost)is introduced for ME Recognition.The proposed method utilizes a dual-stream architecture where motion vectors(temporal features)are extracted using Optical Flow TV-L1 and amplify subtle changes(spatial features)via EulerianVideoMagnification(EVM).These features are processed by IDSSCNN,with an attention mechanism applied to refine the extracted effective features.The outputs are then fused,concatenated,and classified using the XgBoost algorithm.This comprehensive approach significantly improves recognition accuracy by leveraging the strengths of both temporal and spatial information,supported by the robust classification power of XgBoost.The proposed method is evaluated on three publicly available ME databases named Chinese Academy of Sciences Micro-expression Database(CASMEII),Spontaneous Micro-Expression Database(SMICHS),and Spontaneous Actions and Micro-Movements(SAMM).Experimental results indicate that the proposed model can achieve outstanding results compared to recent models.The accuracy results are 79.01%,69.22%,and 68.99%on CASMEII,SMIC-HS,and SAMM,and the F1-score are 75.47%,68.91%,and 63.84%,respectively.The proposed method has the advantage of operational efficiency and less computational time.展开更多
Compared with RGB videos and images,human bone data is less vulnerable to external factors and has stronger robustness.Therefore,behavior recognition methods based on skeletons are widely studied.Because graph convolu...Compared with RGB videos and images,human bone data is less vulnerable to external factors and has stronger robustness.Therefore,behavior recognition methods based on skeletons are widely studied.Because graph convolution network(GCN)can deal with the irregular topology data of hu-man skeletons very well,more and more researchers apply GCN to human behavior recognition.Tra-ditional graph convolution methods only consider the joints with physical connectivity or the same type when building the behavior recognition model based on human skeletons structure,which cannot capture higher-order information better.To solve this problem,Motif-GCN is used in this paper to ex-tract spatial features.The relationship between the joints with natural connection in the human body is encoded by the first Motif-GCN,and the possible relationship between the unconnected joints in the human skeleton is encoded by the second Motif-GCN.In this way,the relationship between non-physical joints can be strengthened.Then a two stream framework combining joint and bone informa-tion is used to capture more action information.Finally,experiments are conducted on two subdata-sets X-Sub and X-View of NTU-RGB+D,and the accuracy shown in Top-1 classification results is 89.5%and 95.4%respectively.The experimental results are 1.0%and 0.3%higher than those of the 2S-AGCN model respectively.The superiority of this method is also proved by the experimental results.展开更多
In peer-assisted streaming network, service provider needs to pre-pushing resources to some users for caching. This scheme can increase the resource availability. In traditional strategy, in order to guarantee the use...In peer-assisted streaming network, service provider needs to pre-pushing resources to some users for caching. This scheme can increase the resource availability. In traditional strategy, in order to guarantee the user＇s quality of service （QoS）, the pre-pushing action can only start at the time point that the certain user has not used the computer for more than 20 minutes. If the user comes back, the pre-pushing action will be stopped. We claim that this is not an efficient scheme. In this paper, we propose a novel pre-pushing scheme to improve the pre-pushing efficiency, while meeting the user＇s QoS requirements. The basic idea is using the user＇s available bandwidth as much as possible in the condition of meeting the user＇s QoS requirements. Then we design an available bandwidth calculating strategy. The numerical simulation demonstrates that our proposed scheme outperforms the traditional one.展开更多
Hydrologic data derived from digital elevation models (DEM) has been regarded as an effective method in the spatial analysis of geographical information systems (GIS). However, both DEM resolution and terrain complexi...Hydrologic data derived from digital elevation models (DEM) has been regarded as an effective method in the spatial analysis of geographical information systems (GIS). However, both DEM resolution and terrain complexity has impacts on the accuracy of hydrologic derivatives. In this study, a multi-resolution and multi-relief comparative approach was used as a major methodology to investigate the accuracy of hydrologic data derived from DEMs. The experiment reveals that DEM terrain representation error affects the accuracy of DEM hydrological derivatives (drainage networks and watershed etc.). Coarser DEM resolutions can usually cause worse results. However, uncertain result commonly exists in this calculation. The derivative errors can be found closely related with DEM vertical resolution and terrain roughness. DEM vertical resolution can be found closely related with the accuracy of DEM hydrological derivatives, especially in the smooth plain area. If the mean slope is less than 4 degrees, the derived hydrologic data are usually unreliable. This result may be helpful in estimating the accuracy of the hydrologic derivatives and determining the DEM resolution that is appropriate to the accuracy requirement of a particular user. By applying a threshold value to subset the cells of a higher accumulation flow, a stream network of a specific network density can be extracted. Some very important geomorphologic characteristics, e.g., shallow and deep gullies, can be separately extracted by means of adjusting the threshold value. However, such a flow accumulation- based processing method can not correctly derive those streams that pass through the working area because it is hard to accumulate enough flow direction values to express the stream channels at the stream's entrance area. Consequently, errors will definitely occur at the stream抯 entrance area. In addition, erroneous derivatives can also be found in deriving some particular rivers, e.g., perched (hanging up) rivers, anastomosing rivers and braided rivers. Therefore, more work should be done to develop and perfect the algorithms.展开更多
Geomorphological and Quaternarygeological field- and laboratory data (Fig.1) are introduced and interpreted with regard to the maximum Ice Age (LGM) glaciation of the Central-and South Karakoram in the Braldu-, Basna-...Geomorphological and Quaternarygeological field- and laboratory data (Fig.1) are introduced and interpreted with regard to the maximum Ice Age (LGM) glaciation of the Central-and South Karakoram in the Braldu-, Basna-, Shigar-and Indus valley system as well as on the Deosai plateau between the Skardu Basin and the Astor valley (Fig.2). These data result from two research expeditions in the years 1997 and 2000. They show that between c. 60 and 20 Ka the Central Karakorum and its south slope were covered by a continuous c. 125,000 km2 sized ice stream network. This ice stream network flowed together to a joint parent glacier, the Indus glacier. The tongue end of the Indus glacier reached down to 850 ～ 800 m a.s.l. In its centre the surface of this Indus ice stream network reached a height of a good 6000 m. Its most important ice thicknesses amounted to c. 2400 ～ 2900 m.展开更多
[Objective] The aim was to simulate the surface water flow of Dalinghe watershed based on SRTM DEM.[Method] By using ArcGIS ModelBuilder,and SRTM DEM data of Dalinghe watershed as input data,the model to simulate the ...[Objective] The aim was to simulate the surface water flow of Dalinghe watershed based on SRTM DEM.[Method] By using ArcGIS ModelBuilder,and SRTM DEM data of Dalinghe watershed as input data,the model to simulate the water flow of Dalinghe watershed was set up.[Result] The model realized automatic division of Dalinghe watershed area and extraction of stream network.In the meantime,it also made the choice of threshold during filling DEM and extracting stream network much easier.The division of the Dalinghe watershed was precise and the extraction result of Dalinghe and its corresponding level Ⅰ tributary river was close to actual stream network.However,the extraction of the smaller stream was less accurate.[Conclusion] The study provided scientific reference for the simulation of surface water network in future.展开更多
Super High Definition (SHD) movies were successfully transmitted as streaming contents at about 300 Mbps for the first time over a long distance IP network (more than 3,000 km), in conjunction with the experimental ve...Super High Definition (SHD) movies were successfully transmitted as streaming contents at about 300 Mbps for the first time over a long distance IP network (more than 3,000 km), in conjunction with the experimental verification of traffic control and scalable multicast technologies.展开更多
Real-time video streaming using ultra-wideband(UWB) technology is experimentally demonstrated along long-reach passive optical networks(LR-PONs) with different wired and wireless reaches. Experimental tests using exte...Real-time video streaming using ultra-wideband(UWB) technology is experimentally demonstrated along long-reach passive optical networks(LR-PONs) with different wired and wireless reaches. Experimental tests using external and direct modulation with UWB wireless radiation in the 10- and 60-GHz bands are performed. An ultra-bendable fiber is also considered for a last-mile distribution. The video quality at the output of the optical fiber infrastructure of the LR-PON is assessed using the error vector magnitude(EVM), and the link quality indicator(LQI) is used as a figure of merit after wireless radiation. An EVM below –17 dB is achieved for both externally and directly modulated LR-PONs comprising up to 125 km of optical fiber. EVM improvement is observed for longer LR-PONs when directly modulated lasers(DMLs) are used because of the amplitude gain provided by the combined effect of dispersion and DML's chirp. Compared with optical back-to-back operation, the LQI level degrades to the maximum around 20% for LR-PONs ranging between 75 and 125 km of fiber reach and with a wireless coverage of 2 m in the 10-GHz UWB band. The same level of LQI degradation is observed using the 60-GHz UWB band with a LR-PON integrating 101 km of access network, a last-mile distribution using ultra-bendable fiber, and a 5.2-m wireless link.展开更多
The fluvio-aeolian sediments in the alluvial plains of NW Haryana and adjoining northern Rajasthan are known to conceal a number of palaeochannels of the erstwhile Saraswati River and its tributaries,which were also t...The fluvio-aeolian sediments in the alluvial plains of NW Haryana and adjoining northern Rajasthan are known to conceal a number of palaeochannels of the erstwhile Saraswati River and its tributaries,which were also the sites for many pre-historic settlements dating back to the pre-Harappan era.Using digital remote sensing techniques like image enhancement,principal component analysis and mineral composite index mapping from Landsat ETM+data,and hydrological routing and stream network generation from DEM.展开更多
Active anomaly detection queries labels of sampled instances and uses them to incrementally update the detection model,and has been widely adopted in detecting network attacks.However,existing methods cannot achieve d...Active anomaly detection queries labels of sampled instances and uses them to incrementally update the detection model,and has been widely adopted in detecting network attacks.However,existing methods cannot achieve desirable performance on dynamic network traffic streams because(1)their query strategies cannot sample informative instances to make the detection model adapt to the evolving stream and(2)their model updating relies on limited query instances only and fails to leverage the enormous unlabeled instances on streams.To address these issues,we propose an active tree based model,adaptive and augmented active prior-knowledge forest(A3PF),for anomaly detection on network trafic streams.A prior-knowledge forest is constructed using prior knowledge of network attacks to find feature subspaces that better distinguish network anomalies from normal traffic.On one hand,to make the model adapt to the evolving stream,a novel adaptive query strategy is designed to sample informative instances from two aspects:the changes in dynamic data distribution and the uncertainty of anomalies.On the other hand,based on the similarity of instances in the neighborhood,we devise an augmented update method to generate pseudo labels for the unlabeled neighbors of query instances,which enables usage of the enormous unlabeled instances during model updating.Extensive experiments on two benchmarks,CIC-IDS2017 and UNSW-NB15,demonstrate that A3PF achieves significant improvements over previous active methods in terms of the area under the receiver operating characteristic curve(AUC-ROC)(20.9%and 21.5%)and the area under the precision-recall curve(AUC-PR)(44.6%and 64.1%).展开更多
基金supported by the Key Research and Development Program of Jiangsu Province under Grant BE2022059-3,CTBC Bank through the Industry-Academia Cooperation Project,as well as by the Ministry of Science and Technology of Taiwan through Grants MOST-108-2218-E-002-055,MOST-109-2223-E-009-002-MY3,MOST-109-2218-E-009-025,and MOST431109-2218-E-002-015.
文摘Micro-expressions(ME)recognition is a complex task that requires advanced techniques to extract informative features fromfacial expressions.Numerous deep neural networks(DNNs)with convolutional structures have been proposed.However,unlike DNNs,shallow convolutional neural networks often outperform deeper models in mitigating overfitting,particularly with small datasets.Still,many of these methods rely on a single feature for recognition,resulting in an insufficient ability to extract highly effective features.To address this limitation,in this paper,an Improved Dual-stream Shallow Convolutional Neural Network based on an Extreme Gradient Boosting Algorithm(IDSSCNN-XgBoost)is introduced for ME Recognition.The proposed method utilizes a dual-stream architecture where motion vectors(temporal features)are extracted using Optical Flow TV-L1 and amplify subtle changes(spatial features)via EulerianVideoMagnification(EVM).These features are processed by IDSSCNN,with an attention mechanism applied to refine the extracted effective features.The outputs are then fused,concatenated,and classified using the XgBoost algorithm.This comprehensive approach significantly improves recognition accuracy by leveraging the strengths of both temporal and spatial information,supported by the robust classification power of XgBoost.The proposed method is evaluated on three publicly available ME databases named Chinese Academy of Sciences Micro-expression Database(CASMEII),Spontaneous Micro-Expression Database(SMICHS),and Spontaneous Actions and Micro-Movements(SAMM).Experimental results indicate that the proposed model can achieve outstanding results compared to recent models.The accuracy results are 79.01%,69.22%,and 68.99%on CASMEII,SMIC-HS,and SAMM,and the F1-score are 75.47%,68.91%,and 63.84%,respectively.The proposed method has the advantage of operational efficiency and less computational time.
基金This work was supported by the Natural Science Foundation of Jiangsu Province(Grant No.BK20181310)the National Natural Science Foundation of China(Grant No.52079039).
基金the National Natural Science Foundation of China(No.61834005,61772417,61802304)the Shaanxi Province Key Research and Development Project(2021GY280).
文摘Compared with RGB videos and images,human bone data is less vulnerable to external factors and has stronger robustness.Therefore,behavior recognition methods based on skeletons are widely studied.Because graph convolution network(GCN)can deal with the irregular topology data of hu-man skeletons very well,more and more researchers apply GCN to human behavior recognition.Tra-ditional graph convolution methods only consider the joints with physical connectivity or the same type when building the behavior recognition model based on human skeletons structure,which cannot capture higher-order information better.To solve this problem,Motif-GCN is used in this paper to ex-tract spatial features.The relationship between the joints with natural connection in the human body is encoded by the first Motif-GCN,and the possible relationship between the unconnected joints in the human skeleton is encoded by the second Motif-GCN.In this way,the relationship between non-physical joints can be strengthened.Then a two stream framework combining joint and bone informa-tion is used to capture more action information.Finally,experiments are conducted on two subdata-sets X-Sub and X-View of NTU-RGB+D,and the accuracy shown in Top-1 classification results is 89.5%and 95.4%respectively.The experimental results are 1.0%and 0.3%higher than those of the 2S-AGCN model respectively.The superiority of this method is also proved by the experimental results.
基金the National Natural Science Foundation of China under Grant No.61202079,61308018，the China Postdoctoral Science Foundation under Grant No.2014T70031，the Fundamental Research Funds for the Central Universities under Grant No.FRF-TP-14-046A2
文摘In peer-assisted streaming network, service provider needs to pre-pushing resources to some users for caching. This scheme can increase the resource availability. In traditional strategy, in order to guarantee the user＇s quality of service （QoS）, the pre-pushing action can only start at the time point that the certain user has not used the computer for more than 20 minutes. If the user comes back, the pre-pushing action will be stopped. We claim that this is not an efficient scheme. In this paper, we propose a novel pre-pushing scheme to improve the pre-pushing efficiency, while meeting the user＇s QoS requirements. The basic idea is using the user＇s available bandwidth as much as possible in the condition of meeting the user＇s QoS requirements. Then we design an available bandwidth calculating strategy. The numerical simulation demonstrates that our proposed scheme outperforms the traditional one.
基金National Natural Science Foundation of China No. 49971065 Open Research Fund Program of LIESMARS+3 种基金 No. WKL 99-0302 High-visiting Scholar Fund of the Key Laboratory of Continental Dynamics Ministry of Education China
文摘Hydrologic data derived from digital elevation models (DEM) has been regarded as an effective method in the spatial analysis of geographical information systems (GIS). However, both DEM resolution and terrain complexity has impacts on the accuracy of hydrologic derivatives. In this study, a multi-resolution and multi-relief comparative approach was used as a major methodology to investigate the accuracy of hydrologic data derived from DEMs. The experiment reveals that DEM terrain representation error affects the accuracy of DEM hydrological derivatives (drainage networks and watershed etc.). Coarser DEM resolutions can usually cause worse results. However, uncertain result commonly exists in this calculation. The derivative errors can be found closely related with DEM vertical resolution and terrain roughness. DEM vertical resolution can be found closely related with the accuracy of DEM hydrological derivatives, especially in the smooth plain area. If the mean slope is less than 4 degrees, the derived hydrologic data are usually unreliable. This result may be helpful in estimating the accuracy of the hydrologic derivatives and determining the DEM resolution that is appropriate to the accuracy requirement of a particular user. By applying a threshold value to subset the cells of a higher accumulation flow, a stream network of a specific network density can be extracted. Some very important geomorphologic characteristics, e.g., shallow and deep gullies, can be separately extracted by means of adjusting the threshold value. However, such a flow accumulation- based processing method can not correctly derive those streams that pass through the working area because it is hard to accumulate enough flow direction values to express the stream channels at the stream's entrance area. Consequently, errors will definitely occur at the stream抯 entrance area. In addition, erroneous derivatives can also be found in deriving some particular rivers, e.g., perched (hanging up) rivers, anastomosing rivers and braided rivers. Therefore, more work should be done to develop and perfect the algorithms.
文摘Geomorphological and Quaternarygeological field- and laboratory data (Fig.1) are introduced and interpreted with regard to the maximum Ice Age (LGM) glaciation of the Central-and South Karakoram in the Braldu-, Basna-, Shigar-and Indus valley system as well as on the Deosai plateau between the Skardu Basin and the Astor valley (Fig.2). These data result from two research expeditions in the years 1997 and 2000. They show that between c. 60 and 20 Ka the Central Karakorum and its south slope were covered by a continuous c. 125,000 km2 sized ice stream network. This ice stream network flowed together to a joint parent glacier, the Indus glacier. The tongue end of the Indus glacier reached down to 850 ～ 800 m a.s.l. In its centre the surface of this Indus ice stream network reached a height of a good 6000 m. Its most important ice thicknesses amounted to c. 2400 ～ 2900 m.
文摘[Objective] The aim was to simulate the surface water flow of Dalinghe watershed based on SRTM DEM.[Method] By using ArcGIS ModelBuilder,and SRTM DEM data of Dalinghe watershed as input data,the model to simulate the water flow of Dalinghe watershed was set up.[Result] The model realized automatic division of Dalinghe watershed area and extraction of stream network.In the meantime,it also made the choice of threshold during filling DEM and extracting stream network much easier.The division of the Dalinghe watershed was precise and the extraction result of Dalinghe and its corresponding level Ⅰ tributary river was close to actual stream network.However,the extraction of the smaller stream was less accurate.[Conclusion] The study provided scientific reference for the simulation of surface water network in future.
文摘Super High Definition (SHD) movies were successfully transmitted as streaming contents at about 300 Mbps for the first time over a long distance IP network (more than 3,000 km), in conjunction with the experimental verification of traffic control and scalable multicast technologies.
基金supported by the Fundao para a Ciência e a Tecnologia from Portugal under projects PEst-OE/EEI/LA0008/2013 and TURBO-PTDC/EEATEL/104358/2008by the European FIVER-FP7-ICT-2009-4-249142 project
文摘Real-time video streaming using ultra-wideband(UWB) technology is experimentally demonstrated along long-reach passive optical networks(LR-PONs) with different wired and wireless reaches. Experimental tests using external and direct modulation with UWB wireless radiation in the 10- and 60-GHz bands are performed. An ultra-bendable fiber is also considered for a last-mile distribution. The video quality at the output of the optical fiber infrastructure of the LR-PON is assessed using the error vector magnitude(EVM), and the link quality indicator(LQI) is used as a figure of merit after wireless radiation. An EVM below –17 dB is achieved for both externally and directly modulated LR-PONs comprising up to 125 km of optical fiber. EVM improvement is observed for longer LR-PONs when directly modulated lasers(DMLs) are used because of the amplitude gain provided by the combined effect of dispersion and DML's chirp. Compared with optical back-to-back operation, the LQI level degrades to the maximum around 20% for LR-PONs ranging between 75 and 125 km of fiber reach and with a wireless coverage of 2 m in the 10-GHz UWB band. The same level of LQI degradation is observed using the 60-GHz UWB band with a LR-PON integrating 101 km of access network, a last-mile distribution using ultra-bendable fiber, and a 5.2-m wireless link.
文摘The fluvio-aeolian sediments in the alluvial plains of NW Haryana and adjoining northern Rajasthan are known to conceal a number of palaeochannels of the erstwhile Saraswati River and its tributaries,which were also the sites for many pre-historic settlements dating back to the pre-Harappan era.Using digital remote sensing techniques like image enhancement,principal component analysis and mineral composite index mapping from Landsat ETM+data,and hydrological routing and stream network generation from DEM.
基金Project supported by the National Science and Technology Major Project(No.2022ZD0115302)the National Natural Science Foundation of China(No.61379052)+1 种基金the Science Foundation of Ministry of Education of China(No.2018A02002)the Natural Science Foundation for Distinguished Young Scholars of Hunan Province,China(No.14JJ1026)。
文摘Active anomaly detection queries labels of sampled instances and uses them to incrementally update the detection model,and has been widely adopted in detecting network attacks.However,existing methods cannot achieve desirable performance on dynamic network traffic streams because(1)their query strategies cannot sample informative instances to make the detection model adapt to the evolving stream and(2)their model updating relies on limited query instances only and fails to leverage the enormous unlabeled instances on streams.To address these issues,we propose an active tree based model,adaptive and augmented active prior-knowledge forest(A3PF),for anomaly detection on network trafic streams.A prior-knowledge forest is constructed using prior knowledge of network attacks to find feature subspaces that better distinguish network anomalies from normal traffic.On one hand,to make the model adapt to the evolving stream,a novel adaptive query strategy is designed to sample informative instances from two aspects:the changes in dynamic data distribution and the uncertainty of anomalies.On the other hand,based on the similarity of instances in the neighborhood,we devise an augmented update method to generate pseudo labels for the unlabeled neighbors of query instances,which enables usage of the enormous unlabeled instances during model updating.Extensive experiments on two benchmarks,CIC-IDS2017 and UNSW-NB15,demonstrate that A3PF achieves significant improvements over previous active methods in terms of the area under the receiver operating characteristic curve(AUC-ROC)(20.9%and 21.5%)and the area under the precision-recall curve(AUC-PR)(44.6%and 64.1%).
