Aiming at the problems of low detection accuracy and large model size of existing object detection algorithms applied to complex road scenes,an improved you only look once version 8(YOLOv8)object detection algorithm f...Aiming at the problems of low detection accuracy and large model size of existing object detection algorithms applied to complex road scenes,an improved you only look once version 8(YOLOv8)object detection algorithm for infrared images,F-YOLOv8,is proposed.First,a spatial-to-depth network replaces the traditional backbone network's strided convolution or pooling layer.At the same time,it combines with the channel attention mechanism so that the neural network focuses on the channels with large weight values to better extract low-resolution image feature information;then an improved feature pyramid network of lightweight bidirectional feature pyramid network(L-BiFPN)is proposed,which can efficiently fuse features of different scales.In addition,a loss function of insertion of union based on the minimum point distance(MPDIoU)is introduced for bounding box regression,which obtains faster convergence speed and more accurate regression results.Experimental results on the FLIR dataset show that the improved algorithm can accurately detect infrared road targets in real time with 3%and 2.2%enhancement in mean average precision at 50%IoU(mAP50)and mean average precision at 50%—95%IoU(mAP50-95),respectively,and 38.1%,37.3%and 16.9%reduction in the number of model parameters,the model weight,and floating-point operations per second(FLOPs),respectively.To further demonstrate the detection capability of the improved algorithm,it is tested on the public dataset PASCAL VOC,and the results show that F-YOLO has excellent generalized detection performance.展开更多
Firstly,neural network based on improved particle swarm optimization (PSO)algorithm is introduced in this paper. Based on the data collected from projects in typical areas,the concrete carbonation depth is assessed wi...Firstly,neural network based on improved particle swarm optimization (PSO)algorithm is introduced in this paper. Based on the data collected from projects in typical areas,the concrete carbonation depth is assessed with consideration of various factors such as unit cement consumption (C),unit water consumption (W),binder material content (B),water binder ratio (W/B ),concrete strength (MPa),rapid carbonization days (D),fly ash consumption of unit volume concrete(FA),fly ash percentage of total cementitious materials (FA%),expansion agent consumption of unit volume concrete(EA),expansion agent percentage of total cementitious materials (FA%).Gaining the data from project-experiment,a model is presented to calculate and forecast carbonation depth using neural network based on improved PSO algorithm. The calculation results indicate that this algorithm accord with the prediction carbonation depth of concrete accuracy requirements and has a better convergence and generalization,worth being popularized.展开更多
This paper outlines the application of the multi-layer perceptron artificial neural network (ANN), ordinary kriging (OK), and inverse distance weighting (IDW) models in the estimation of local scour depth around bridg...This paper outlines the application of the multi-layer perceptron artificial neural network (ANN), ordinary kriging (OK), and inverse distance weighting (IDW) models in the estimation of local scour depth around bridge piers. As part of this study, bridge piers were installed with bed sills at the bed of an experimental flume. Experimental tests were conducted under different flow conditions and varying distances between bridge pier and bed sill. The ANN, OK and IDW models were applied to the experimental data and it was shown that the artificial neural network model predicts local scour depth more accurately than the kriging and inverse distance weighting models. It was found that the ANN with two hidden layers was the optimum model to predict local scour depth. The results from the sixth test case showed that the ANN with one hidden layer and 17 hidden nodes was the best model to predict local scour depth. Whereas the results from the fifth test case found that the ANN with three hidden layers was the best model to predict local scour depth.展开更多
With the global climate change,the high-altitude detection is more and more important in the climate prediction,and the input-output characteristic curve of the air pressure sensor is offset due to the interference of...With the global climate change,the high-altitude detection is more and more important in the climate prediction,and the input-output characteristic curve of the air pressure sensor is offset due to the interference of the tested object and the environment under test,and the nonlinear error is generated.Aiming at the difficulty of nonlinear correction of pressure sensor and the low accuracy of correction results,depth neural network model was established based on wavelet function,and Levenberg-Marquardt algorithm is used to update network parameters to realize the nonlinear correction of pressure sensor.The experimental results show that compared with the traditional neural network model,the improved depth neural network not only accelerates the convergence rate,but also improves the correction accuracy,meets the error requirements of upper-air detection,and has a good generalization ability,which can be extended to the nonlinear correction of similar sensors.展开更多
The defense in depth methodology was popularized in the early 2000’s amid growing concerns for information security;this paper will address the shortcomings of early implementations. In the last two years, many suppo...The defense in depth methodology was popularized in the early 2000’s amid growing concerns for information security;this paper will address the shortcomings of early implementations. In the last two years, many supporters of the defense in depth security methodology have changed their allegiance to an offshoot method dubbed the defense in breadth methodology. A substantial portion of this paper’s body will be devoted to comparing real-world usage scenarios and discussing the flaws in each method. A major goal of this publication will be to assist readers in selecting a method that will best benefit their personal environment. Scenarios certainly exist where one method may be clearly favored;this article will help identify the factors that make one method a clear choice over another. This paper will strive not only to highlight key strengths and weaknesses for the two strategies listed, but also provide the evaluation techniques necessary for readers to apply to other popular methodologies in order to make the most appropriate personal determinations.展开更多
The purpose of this research was to determine whether the Linear Regression Analysis can be effectively applied to the prioritization of defense-in-depth security tools and procedures to reduce cyber threats during th...The purpose of this research was to determine whether the Linear Regression Analysis can be effectively applied to the prioritization of defense-in-depth security tools and procedures to reduce cyber threats during the Global Corona Virus Pandemic. The way this was determined or methods used in this study consisted of scanning 20 peer reviewed Cybersecurity Articles from prominent Cybersecurity Journals for a list of defense in depth measures (tools and procedures) and the threats that those measures were designed to reduce. The methods further involved using the Likert Scale Model to create an ordinal ranking of the measures and threats. The defense in depth tools and procedures were then compared to see whether the Likert scale and Linear Regression Analysis could be effectively applied to prioritize and combine the measures to reduce pandemic related cyber threats. The results of this research reject the H0 null hypothesis that Linear Regression Analysis does not affect the relationship between the prioritization and combining of defense in depth tools and procedures (independent variables) and pandemic related cyber threats (dependent variables).展开更多
An intelligent system including both a neural network(NN) and a self adjusting fuzzy controller(FC) for modeling and control of the penetration depth during gas tungsten arc welding(GTAW) process is presented in this...An intelligent system including both a neural network(NN) and a self adjusting fuzzy controller(FC) for modeling and control of the penetration depth during gas tungsten arc welding(GTAW) process is presented in this paper. The discussion is mainly focused on two parts. One is the modeling of the penetration depth with NN. A visual sensor CCD is used to obtain the image of the molten pool. A neural network model is established to estimate the penetration depth from the welding current, pool width and seam gap. It is demonstrated that the proposed neural network can produce highly complex nonlinear multi variable model of the GTAW process that offer the accurate prediction of welding penetration depth. Another is the control for the penetration depth with FC.A self adjusting fuzzy controller is proposed,which used for controlling the penetration depth.The control parameters are adjusted on line automatically according to the controlling errors of penetration and the errors can be decreased sharply. The effectiveness of the proposed intelligent methods is demonstrated by the real experiments and the improved performance results are obtained.展开更多
Studied in this article is whether the Bayesian Network Model (BNM) can be effectively applied to the prioritization of defense in-depth security tools and procedures and to the combining of those measures to reduce c...Studied in this article is whether the Bayesian Network Model (BNM) can be effectively applied to the prioritization of defense in-depth security tools and procedures and to the combining of those measures to reduce cyber threats. The methods used in this study consisted of scanning 24 peer reviewed Cybersecurity Articles from prominent Cybersecurity Journals using the Likert Scale Model for the article’s list of defense in depth measures (tools and procedures) and the threats that those measures were designed to reduce. The defense in depth tools and procedures are then compared to see whether the Likert scale and the Bayesian Network Model could be effectively applied to prioritize and combine the measures to reduce cyber threats attacks against organizational and private computing systems. The findings of the research reject the H0 null hypothesis that BNM does not affect the relationship between the prioritization and combining of 24 Cybersecurity Article’s defense in depth tools and procedures (independent variables) and cyber threats (dependent variables).展开更多
自监督单目深度估计受到了国内外研究人员的广泛关注。现有基于深度学习的自监督单目深度估计方法主要采用编码器-解码器结构。然而,这些方法在编码过程中对输入图像进行下采样操作,导致部分图像信息,尤其是图像的边界信息丢失,进而影...自监督单目深度估计受到了国内外研究人员的广泛关注。现有基于深度学习的自监督单目深度估计方法主要采用编码器-解码器结构。然而,这些方法在编码过程中对输入图像进行下采样操作,导致部分图像信息,尤其是图像的边界信息丢失,进而影响深度图的精度。针对上述问题,提出一种基于拉普拉斯金字塔的自监督单目深度估计方法(Self-supervised Monocular Depth Estimation Based on the Laplace Pyramid,LpDepth)。此方法的核心思想是:首先,使用拉普拉斯残差图丰富编码特征,以弥补在下采样过程中丢失的特征信息;其次,在下采样过程中使用最大池化层突显和放大特征信息,使编码器在特征提取过程中更容易地提取到训练模型所需要的特征信息;最后,使用残差模块解决过拟合问题,提高解码器对特征的利用效率。在KITTI和Make3D等数据集上对所提方法进行了测试,同时将其与现有经典方法进行了比较。实验结果证明了所提方法的有效性。展开更多
针对目前雷达干扰抑制决策智能化程度低的问题,提出了一种基于双深度优先经验回放和可变贪婪算法改进的双重竞争深度Q网络(double dueling deep Q network,D3QN)决策的雷达干扰抑制方法。首先对雷达目标回波和干扰混合信号进行特征提取...针对目前雷达干扰抑制决策智能化程度低的问题,提出了一种基于双深度优先经验回放和可变贪婪算法改进的双重竞争深度Q网络(double dueling deep Q network,D3QN)决策的雷达干扰抑制方法。首先对雷达目标回波和干扰混合信号进行特征提取;然后根据信号特征通过可变贪婪算法选择动作作用于干扰,并将动作前后的信号特征存储于双深度优先经验回放池后,经过学习决策出最优的干扰抑制策略;最后使用该策略抑制干扰后输出。实验结果表明,该方法有效改善了信号的脉压结果,显著提升了信号的信干噪比,相较于基于D3QN的传统干扰抑制方法,在策略准确率和收敛速度上分别提升了7.3%和8.7%。展开更多
Real-time hand gesture recognition technology significantly improves the user's experience for virtual reality/augmented reality(VR/AR) applications, which relies on the identification of the orientation of the ha...Real-time hand gesture recognition technology significantly improves the user's experience for virtual reality/augmented reality(VR/AR) applications, which relies on the identification of the orientation of the hand in captured images or videos. A new three-stage pipeline approach for fast and accurate hand segmentation for the hand from a single depth image is proposed. Firstly, a depth frame is segmented into several regions by histogrambased threshold selection algorithm and by tracing the exterior boundaries of objects after thresholding. Secondly, each segmentation proposal is evaluated by a three-layers shallow convolutional neural network(CNN) to determine whether or not the boundary is associated with the hand. Finally, all hand components are merged as the hand segmentation result. Compared with algorithms based on random decision forest(RDF), the experimental results demonstrate that the approach achieves better performance with high-accuracy(88.34% mean intersection over union, mIoU) and a shorter processing time(≤8 ms).展开更多
Pattern matching is a fundamental approach to detect malicious behaviors and information over Internet, which has been gradually used in high-speed network traffic analysis. However, there is a performance bottleneck ...Pattern matching is a fundamental approach to detect malicious behaviors and information over Internet, which has been gradually used in high-speed network traffic analysis. However, there is a performance bottleneck for multi-pattern matching on online compressed network traffic(CNT), this is because malicious and intrusion codes are often embedded into compressed network traffic. In this paper, we propose an online fast and multi-pattern matching algorithm on compressed network traffic(FMMCN). FMMCN employs two types of jumping, i.e. jumping during sliding window and a string jump scanning strategy to skip unnecessary compressed bytes. Moreover, FMMCN has the ability to efficiently process multiple large volume of networks such as HTTP traffic, vehicles traffic, and other Internet-based services. The experimental results show that FMMCN can ignore more than 89.5% of bytes, and its maximum speed reaches 176.470MB/s in a midrange switches device, which is faster than the current fastest algorithm ACCH by almost 73.15 MB/s.展开更多
基金supported by the National Natural Science Foundation of China(No.62103298)。
文摘Aiming at the problems of low detection accuracy and large model size of existing object detection algorithms applied to complex road scenes,an improved you only look once version 8(YOLOv8)object detection algorithm for infrared images,F-YOLOv8,is proposed.First,a spatial-to-depth network replaces the traditional backbone network's strided convolution or pooling layer.At the same time,it combines with the channel attention mechanism so that the neural network focuses on the channels with large weight values to better extract low-resolution image feature information;then an improved feature pyramid network of lightweight bidirectional feature pyramid network(L-BiFPN)is proposed,which can efficiently fuse features of different scales.In addition,a loss function of insertion of union based on the minimum point distance(MPDIoU)is introduced for bounding box regression,which obtains faster convergence speed and more accurate regression results.Experimental results on the FLIR dataset show that the improved algorithm can accurately detect infrared road targets in real time with 3%and 2.2%enhancement in mean average precision at 50%IoU(mAP50)and mean average precision at 50%—95%IoU(mAP50-95),respectively,and 38.1%,37.3%and 16.9%reduction in the number of model parameters,the model weight,and floating-point operations per second(FLOPs),respectively.To further demonstrate the detection capability of the improved algorithm,it is tested on the public dataset PASCAL VOC,and the results show that F-YOLO has excellent generalized detection performance.
文摘Firstly,neural network based on improved particle swarm optimization (PSO)algorithm is introduced in this paper. Based on the data collected from projects in typical areas,the concrete carbonation depth is assessed with consideration of various factors such as unit cement consumption (C),unit water consumption (W),binder material content (B),water binder ratio (W/B ),concrete strength (MPa),rapid carbonization days (D),fly ash consumption of unit volume concrete(FA),fly ash percentage of total cementitious materials (FA%),expansion agent consumption of unit volume concrete(EA),expansion agent percentage of total cementitious materials (FA%).Gaining the data from project-experiment,a model is presented to calculate and forecast carbonation depth using neural network based on improved PSO algorithm. The calculation results indicate that this algorithm accord with the prediction carbonation depth of concrete accuracy requirements and has a better convergence and generalization,worth being popularized.
文摘This paper outlines the application of the multi-layer perceptron artificial neural network (ANN), ordinary kriging (OK), and inverse distance weighting (IDW) models in the estimation of local scour depth around bridge piers. As part of this study, bridge piers were installed with bed sills at the bed of an experimental flume. Experimental tests were conducted under different flow conditions and varying distances between bridge pier and bed sill. The ANN, OK and IDW models were applied to the experimental data and it was shown that the artificial neural network model predicts local scour depth more accurately than the kriging and inverse distance weighting models. It was found that the ANN with two hidden layers was the optimum model to predict local scour depth. The results from the sixth test case showed that the ANN with one hidden layer and 17 hidden nodes was the best model to predict local scour depth. Whereas the results from the fifth test case found that the ANN with three hidden layers was the best model to predict local scour depth.
基金This paper is supported by the following funds:National Key R&D Program of China(2018YFF01010100)National natural science foundation of China(61672064),Beijing natural science foundation project(4172001)Advanced information network Beijing laboratory(PXM2019_014204_500029).
文摘With the global climate change,the high-altitude detection is more and more important in the climate prediction,and the input-output characteristic curve of the air pressure sensor is offset due to the interference of the tested object and the environment under test,and the nonlinear error is generated.Aiming at the difficulty of nonlinear correction of pressure sensor and the low accuracy of correction results,depth neural network model was established based on wavelet function,and Levenberg-Marquardt algorithm is used to update network parameters to realize the nonlinear correction of pressure sensor.The experimental results show that compared with the traditional neural network model,the improved depth neural network not only accelerates the convergence rate,but also improves the correction accuracy,meets the error requirements of upper-air detection,and has a good generalization ability,which can be extended to the nonlinear correction of similar sensors.
文摘The defense in depth methodology was popularized in the early 2000’s amid growing concerns for information security;this paper will address the shortcomings of early implementations. In the last two years, many supporters of the defense in depth security methodology have changed their allegiance to an offshoot method dubbed the defense in breadth methodology. A substantial portion of this paper’s body will be devoted to comparing real-world usage scenarios and discussing the flaws in each method. A major goal of this publication will be to assist readers in selecting a method that will best benefit their personal environment. Scenarios certainly exist where one method may be clearly favored;this article will help identify the factors that make one method a clear choice over another. This paper will strive not only to highlight key strengths and weaknesses for the two strategies listed, but also provide the evaluation techniques necessary for readers to apply to other popular methodologies in order to make the most appropriate personal determinations.
文摘The purpose of this research was to determine whether the Linear Regression Analysis can be effectively applied to the prioritization of defense-in-depth security tools and procedures to reduce cyber threats during the Global Corona Virus Pandemic. The way this was determined or methods used in this study consisted of scanning 20 peer reviewed Cybersecurity Articles from prominent Cybersecurity Journals for a list of defense in depth measures (tools and procedures) and the threats that those measures were designed to reduce. The methods further involved using the Likert Scale Model to create an ordinal ranking of the measures and threats. The defense in depth tools and procedures were then compared to see whether the Likert scale and Linear Regression Analysis could be effectively applied to prioritize and combine the measures to reduce pandemic related cyber threats. The results of this research reject the H0 null hypothesis that Linear Regression Analysis does not affect the relationship between the prioritization and combining of defense in depth tools and procedures (independent variables) and pandemic related cyber threats (dependent variables).
文摘An intelligent system including both a neural network(NN) and a self adjusting fuzzy controller(FC) for modeling and control of the penetration depth during gas tungsten arc welding(GTAW) process is presented in this paper. The discussion is mainly focused on two parts. One is the modeling of the penetration depth with NN. A visual sensor CCD is used to obtain the image of the molten pool. A neural network model is established to estimate the penetration depth from the welding current, pool width and seam gap. It is demonstrated that the proposed neural network can produce highly complex nonlinear multi variable model of the GTAW process that offer the accurate prediction of welding penetration depth. Another is the control for the penetration depth with FC.A self adjusting fuzzy controller is proposed,which used for controlling the penetration depth.The control parameters are adjusted on line automatically according to the controlling errors of penetration and the errors can be decreased sharply. The effectiveness of the proposed intelligent methods is demonstrated by the real experiments and the improved performance results are obtained.
文摘Studied in this article is whether the Bayesian Network Model (BNM) can be effectively applied to the prioritization of defense in-depth security tools and procedures and to the combining of those measures to reduce cyber threats. The methods used in this study consisted of scanning 24 peer reviewed Cybersecurity Articles from prominent Cybersecurity Journals using the Likert Scale Model for the article’s list of defense in depth measures (tools and procedures) and the threats that those measures were designed to reduce. The defense in depth tools and procedures are then compared to see whether the Likert scale and the Bayesian Network Model could be effectively applied to prioritize and combine the measures to reduce cyber threats attacks against organizational and private computing systems. The findings of the research reject the H0 null hypothesis that BNM does not affect the relationship between the prioritization and combining of 24 Cybersecurity Article’s defense in depth tools and procedures (independent variables) and cyber threats (dependent variables).
文摘自监督单目深度估计受到了国内外研究人员的广泛关注。现有基于深度学习的自监督单目深度估计方法主要采用编码器-解码器结构。然而,这些方法在编码过程中对输入图像进行下采样操作,导致部分图像信息,尤其是图像的边界信息丢失,进而影响深度图的精度。针对上述问题,提出一种基于拉普拉斯金字塔的自监督单目深度估计方法(Self-supervised Monocular Depth Estimation Based on the Laplace Pyramid,LpDepth)。此方法的核心思想是:首先,使用拉普拉斯残差图丰富编码特征,以弥补在下采样过程中丢失的特征信息;其次,在下采样过程中使用最大池化层突显和放大特征信息,使编码器在特征提取过程中更容易地提取到训练模型所需要的特征信息;最后,使用残差模块解决过拟合问题,提高解码器对特征的利用效率。在KITTI和Make3D等数据集上对所提方法进行了测试,同时将其与现有经典方法进行了比较。实验结果证明了所提方法的有效性。
文摘针对目前雷达干扰抑制决策智能化程度低的问题,提出了一种基于双深度优先经验回放和可变贪婪算法改进的双重竞争深度Q网络(double dueling deep Q network,D3QN)决策的雷达干扰抑制方法。首先对雷达目标回波和干扰混合信号进行特征提取;然后根据信号特征通过可变贪婪算法选择动作作用于干扰,并将动作前后的信号特征存储于双深度优先经验回放池后,经过学习决策出最优的干扰抑制策略;最后使用该策略抑制干扰后输出。实验结果表明,该方法有效改善了信号的脉压结果,显著提升了信号的信干噪比,相较于基于D3QN的传统干扰抑制方法,在策略准确率和收敛速度上分别提升了7.3%和8.7%。
文摘Real-time hand gesture recognition technology significantly improves the user's experience for virtual reality/augmented reality(VR/AR) applications, which relies on the identification of the orientation of the hand in captured images or videos. A new three-stage pipeline approach for fast and accurate hand segmentation for the hand from a single depth image is proposed. Firstly, a depth frame is segmented into several regions by histogrambased threshold selection algorithm and by tracing the exterior boundaries of objects after thresholding. Secondly, each segmentation proposal is evaluated by a three-layers shallow convolutional neural network(CNN) to determine whether or not the boundary is associated with the hand. Finally, all hand components are merged as the hand segmentation result. Compared with algorithms based on random decision forest(RDF), the experimental results demonstrate that the approach achieves better performance with high-accuracy(88.34% mean intersection over union, mIoU) and a shorter processing time(≤8 ms).
基金supported by China MOST project (No.2012BAH46B04)
文摘Pattern matching is a fundamental approach to detect malicious behaviors and information over Internet, which has been gradually used in high-speed network traffic analysis. However, there is a performance bottleneck for multi-pattern matching on online compressed network traffic(CNT), this is because malicious and intrusion codes are often embedded into compressed network traffic. In this paper, we propose an online fast and multi-pattern matching algorithm on compressed network traffic(FMMCN). FMMCN employs two types of jumping, i.e. jumping during sliding window and a string jump scanning strategy to skip unnecessary compressed bytes. Moreover, FMMCN has the ability to efficiently process multiple large volume of networks such as HTTP traffic, vehicles traffic, and other Internet-based services. The experimental results show that FMMCN can ignore more than 89.5% of bytes, and its maximum speed reaches 176.470MB/s in a midrange switches device, which is faster than the current fastest algorithm ACCH by almost 73.15 MB/s.
