Versatile Video Coding (VVC) introduces the new quad-tree with a nested multi-type tree (QTMT) block division structure, which increases the flexibility of block division, the more complex block division structure increases the coding complexity of VVC by nearly 26 times compared with High-Efficiency Video Coding (HEVC). Therefore, it is urgent to reduce the coding complexity of VVC. In this paper, we propose a fast CU division method based on optimal stopping theory and block texture decision. Firstly, by analyzing the division depth of the Coding Tree Unit (CTU) at the same position as neighboring frames, we use the optimal stopping theory to determine the optimal division layer of the current CTU, to terminate the division process in advance. Then, by judging the texture direction of the current Coding Unit (CU), the calculation of several CU division methods is selected to be skipped, thus reducing the computational effort of coding. The experimental results show that the coding time of this scheme is reduced by 45.65% on average, while the BDBR only increases by 1.64%.
通用视频编码(VVC)引入了一种新的四叉树嵌套多类型树(QTMT)块划分结构，增加了块划分的灵活性，更复杂的块划分结构使VVC的编码复杂度比高效视频编码(HEVC)提高了近26倍。因此，降低VVC的编码复杂度是当务之急。本文提出了一种基于最优停止理论和块纹理判定的快速CU分割方法。首先，通过分析相邻帧在同一位置的编码树单元(CTU)的分割深度，利用最优停止理论确定当前CTU的最优分割层，提前终止分割过程;然后，通过判断当前编码单元(Coding Unit, CU)的纹理方向，选择几种CU划分方法的计算跳过，从而减少编码的计算量。实验结果表明，该方案的编码时间平均缩短了45.65%，而BDBR仅增加了1.64%。
{"title":"Optimal Stopping Theory-Enabled VVC Intra Prediction with Texture","authors":"Yucheng Li, Xiantao Jiang, Wei Li, Jiayuan Jin, Dezhi Han, Tian Song, Fei Yu","doi":"10.1109/CCISP55629.2022.9974416","DOIUrl":"https://doi.org/10.1109/CCISP55629.2022.9974416","url":null,"abstract":"Versatile Video Coding (VVC) introduces the new quad-tree with a nested multi-type tree (QTMT) block division structure, which increases the flexibility of block division, the more complex block division structure increases the coding complexity of VVC by nearly 26 times compared with High-Efficiency Video Coding (HEVC). Therefore, it is urgent to reduce the coding complexity of VVC. In this paper, we propose a fast CU division method based on optimal stopping theory and block texture decision. Firstly, by analyzing the division depth of the Coding Tree Unit (CTU) at the same position as neighboring frames, we use the optimal stopping theory to determine the optimal division layer of the current CTU, to terminate the division process in advance. Then, by judging the texture direction of the current Coding Unit (CU), the calculation of several CU division methods is selected to be skipped, thus reducing the computational effort of coding. The experimental results show that the coding time of this scheme is reduced by 45.65% on average, while the BDBR only increases by 1.64%.","PeriodicalId":431851,"journal":{"name":"2022 7th International Conference on Communication, Image and Signal Processing (CCISP)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124406639","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The swarm information network requires outstanding features such as flexible networking, fast networking, security, and controllability. According to the information interaction requirements of the swarm information network, a Software Defined Network (SDN) modularized swarm network information architecture is proposed, the business requirements and characteristics of the swarm network are analyzed, and the swarm network information architecture mode is designed to improve the openness and reliability of the swarm network information system, which provides support for swarms to carry out tasks autonomously in complex scenes.
群信息网络需要组网灵活、组网快速、安全、可控性等突出特点。根据蜂群信息网络的信息交互需求，提出了一种软件定义网络(Software Defined network, SDN)模块化的蜂群网络信息架构，分析了蜂群网络的业务需求和特点，设计了蜂群网络信息架构模式，提高了蜂群网络信息系统的开放性和可靠性，为蜂群在复杂场景下自主执行任务提供了支持。
{"title":"Construction Technology of Software Defined Modular Swarm Network Information Architecture","authors":"Yuyong Cui, Dawei Liu, Shengzhe Wang, Xinvi Gao, Mutian Guo","doi":"10.1109/CCISP55629.2022.9974179","DOIUrl":"https://doi.org/10.1109/CCISP55629.2022.9974179","url":null,"abstract":"The swarm information network requires outstanding features such as flexible networking, fast networking, security, and controllability. According to the information interaction requirements of the swarm information network, a Software Defined Network (SDN) modularized swarm network information architecture is proposed, the business requirements and characteristics of the swarm network are analyzed, and the swarm network information architecture mode is designed to improve the openness and reliability of the swarm network information system, which provides support for swarms to carry out tasks autonomously in complex scenes.","PeriodicalId":431851,"journal":{"name":"2022 7th International Conference on Communication, Image and Signal Processing (CCISP)","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116939563","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-11-01DOI: 10.1109/CCISP55629.2022.9974507
Wentao Cai, Hao Zhao, Heng Wang, Xue Deng
Aiming at the problem of the low recognition accuracy caused by the arbitrary characters. In this paper, we propose an arbitrary direction character recognition network. Firstly, a lightweight spatial transformation network (STNet) is designed based on the MobileNetV2, which is used to extract the spatial features of the arbitrary characters and perform spatial transformation. Simultaneously, we introduced the SE attention block into the feature extraction backbone network, which makes the network focuses on the key regions of characters. Then, we build a text recognizer based on recurrent neural network and introduce the Connectionist Temporal Classification (CTC) loss to achieve the flexible alignment between the visual features and the prediction outputs. Extensive experiments are carried out on the IIIT5K and a self-made inkjet characters dataset. The recognition accuracy of our proposed method reaches 95.7% and 86.3% respectively. Compared with the benchmarks, the maximum accuracy of the proposed method is improved by 17.5%. Experimental results show the effectiveness of our proposed method.
{"title":"Arbitrary Direction Inkjet Character Recognition Based on Spatial Transformation","authors":"Wentao Cai, Hao Zhao, Heng Wang, Xue Deng","doi":"10.1109/CCISP55629.2022.9974507","DOIUrl":"https://doi.org/10.1109/CCISP55629.2022.9974507","url":null,"abstract":"Aiming at the problem of the low recognition accuracy caused by the arbitrary characters. In this paper, we propose an arbitrary direction character recognition network. Firstly, a lightweight spatial transformation network (STNet) is designed based on the MobileNetV2, which is used to extract the spatial features of the arbitrary characters and perform spatial transformation. Simultaneously, we introduced the SE attention block into the feature extraction backbone network, which makes the network focuses on the key regions of characters. Then, we build a text recognizer based on recurrent neural network and introduce the Connectionist Temporal Classification (CTC) loss to achieve the flexible alignment between the visual features and the prediction outputs. Extensive experiments are carried out on the IIIT5K and a self-made inkjet characters dataset. The recognition accuracy of our proposed method reaches 95.7% and 86.3% respectively. Compared with the benchmarks, the maximum accuracy of the proposed method is improved by 17.5%. Experimental results show the effectiveness of our proposed method.","PeriodicalId":431851,"journal":{"name":"2022 7th International Conference on Communication, Image and Signal Processing (CCISP)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122906302","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-11-01DOI: 10.1109/CCISP55629.2022.9974588
Yiyuan Wang, Jun Chang, Zhongkui Lu, Fuhui Yu, Jiaqi Wei, Yan Xu
In view of the requirement of high speed and low delay in 5G system, traditional channel estimation algorithms are difficult to meet the requirements. This paper regards the channel estimation problem in communication systems as an image processing problem in deep learning, and proposes a channel estimation network based on ConvLSTM network. Convolutional neural network is used in channel estimation, and LSTM structure is introduced to capture the correlation of the channel. The parameters are set to generate the channel data information set of the physical downlink shared channel (PDSCH) based on the 5G new radio (NR) standard, which is used to evaluate the performance of the proposed and existing algorithms. Experimental simulations show that the proposed algorithm has obvious performance improvement and strong robustness compared with least squares algorithm, practical channel estimation and T-CNN network based on image processing.
{"title":"Channel estimation of 5G OFDM system based on ConvLSTM network","authors":"Yiyuan Wang, Jun Chang, Zhongkui Lu, Fuhui Yu, Jiaqi Wei, Yan Xu","doi":"10.1109/CCISP55629.2022.9974588","DOIUrl":"https://doi.org/10.1109/CCISP55629.2022.9974588","url":null,"abstract":"In view of the requirement of high speed and low delay in 5G system, traditional channel estimation algorithms are difficult to meet the requirements. This paper regards the channel estimation problem in communication systems as an image processing problem in deep learning, and proposes a channel estimation network based on ConvLSTM network. Convolutional neural network is used in channel estimation, and LSTM structure is introduced to capture the correlation of the channel. The parameters are set to generate the channel data information set of the physical downlink shared channel (PDSCH) based on the 5G new radio (NR) standard, which is used to evaluate the performance of the proposed and existing algorithms. Experimental simulations show that the proposed algorithm has obvious performance improvement and strong robustness compared with least squares algorithm, practical channel estimation and T-CNN network based on image processing.","PeriodicalId":431851,"journal":{"name":"2022 7th International Conference on Communication, Image and Signal Processing (CCISP)","volume":"39 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123033187","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-11-01DOI: 10.1109/CCISP55629.2022.9974344
Fuquan Zhu, Liping Yang
The special statistical property produced by radiative correction has a serious impact on prediction accuracy. Back pixel search (BPS) algorithm is currently the most effective way to solve this problem. However, the effectiveness of BPS algorithm depends on optimal threshold and the prediction accuracy of the first prediction. In this paper, an effective lossless compression method for hyperspectral image based on conventional recursive least squares (CRLS) algorithm and BPS algorithm with adaptive threshold is proposed. Firstly, the CRLS predictor is adopted in the first prediction to improve the accuracy of predicted reference values. Afterwards, a recursive error mean estimation with scaling factor is used to estimate the optimal search threshold in the BPS predictor. Finally, the arithmetic encoder is used to entropy-encode the residuals generated by prediction. The experimental results on Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) images set show that this method significantly improves the compression effect and reduces the computational complexity compared with the typical methods already reported.
{"title":"Lossless Compression for Hyperspectral Images Using Back Pixel Search with Adaptive Threshold","authors":"Fuquan Zhu, Liping Yang","doi":"10.1109/CCISP55629.2022.9974344","DOIUrl":"https://doi.org/10.1109/CCISP55629.2022.9974344","url":null,"abstract":"The special statistical property produced by radiative correction has a serious impact on prediction accuracy. Back pixel search (BPS) algorithm is currently the most effective way to solve this problem. However, the effectiveness of BPS algorithm depends on optimal threshold and the prediction accuracy of the first prediction. In this paper, an effective lossless compression method for hyperspectral image based on conventional recursive least squares (CRLS) algorithm and BPS algorithm with adaptive threshold is proposed. Firstly, the CRLS predictor is adopted in the first prediction to improve the accuracy of predicted reference values. Afterwards, a recursive error mean estimation with scaling factor is used to estimate the optimal search threshold in the BPS predictor. Finally, the arithmetic encoder is used to entropy-encode the residuals generated by prediction. The experimental results on Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) images set show that this method significantly improves the compression effect and reduces the computational complexity compared with the typical methods already reported.","PeriodicalId":431851,"journal":{"name":"2022 7th International Conference on Communication, Image and Signal Processing (CCISP)","volume":"274 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128988857","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-11-01DOI: 10.1109/CCISP55629.2022.9974159
Cai Xiang, Kang Yihua, Ma Hongbao, Hong Ning, Qiu Gongzhe
Machine vision is more and more widely used in industrial inspection. In order to improve the industrial production of steel pipes, machine vision is used to replace manual inspection of steel pipe surface defects. Because the surface of the steel pipe is cylindrical, there are different degrees of deformation and information loss in the edge parts on both sides of the image when capturing the steel pipe surface image. Aiming at the deformation and distortion problems of the edge of the steel pipe image, a mathematical model is established and calculated. According to the principle of cylindrical projection, this paper deduces the expression of cylindrical orthographic projection, and puts forward a restoration algorithm of cylindrical back projection, which realizes the cylindrical expansion of steel pipe image, and solves the problems of defect shape distortion caused by the deformation of the edges on both sides of the image and the difficulty of defect detection. The results show that this method is effective and lays a foundation for steel pipe detection.
{"title":"Research on preprocessing method based on cylindrical image restoration of steel pipe","authors":"Cai Xiang, Kang Yihua, Ma Hongbao, Hong Ning, Qiu Gongzhe","doi":"10.1109/CCISP55629.2022.9974159","DOIUrl":"https://doi.org/10.1109/CCISP55629.2022.9974159","url":null,"abstract":"Machine vision is more and more widely used in industrial inspection. In order to improve the industrial production of steel pipes, machine vision is used to replace manual inspection of steel pipe surface defects. Because the surface of the steel pipe is cylindrical, there are different degrees of deformation and information loss in the edge parts on both sides of the image when capturing the steel pipe surface image. Aiming at the deformation and distortion problems of the edge of the steel pipe image, a mathematical model is established and calculated. According to the principle of cylindrical projection, this paper deduces the expression of cylindrical orthographic projection, and puts forward a restoration algorithm of cylindrical back projection, which realizes the cylindrical expansion of steel pipe image, and solves the problems of defect shape distortion caused by the deformation of the edges on both sides of the image and the difficulty of defect detection. The results show that this method is effective and lays a foundation for steel pipe detection.","PeriodicalId":431851,"journal":{"name":"2022 7th International Conference on Communication, Image and Signal Processing (CCISP)","volume":"111 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129277801","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In recent years, global warming and climate extremes have occurred frequently, seriously affecting human life, production and sustainable development.To study the correlation between carbon emissions and climate change, complex networks and big data statistical analysis methods were used to construct a multi-factor climate network of Yunnan carbon emissions and multi-factor climate network by determining the connectivity of edges through Pearson correlation coefficients, and sliding series correlation was used to analyse the effect of carbon emissions on meteorological factors.The results show that carbon emissions show a positive correlation with temperature, wind speed and sunshine, and a negative correlation with air pressure, precipitation and humidity.On long time scales carbon emissions have an impact on changes in meteorological factors, with an immediate effect on wind speed and a trend from lagging to immediate on precipitation.The study provides some theoretical reference for the control of CO2 emissions in Yunnan Province.
{"title":"Analysis of the effect of carbon emissions on meteorological factors in Yunnan province","authors":"Guilan Luo, Xin Ma, Xuan Liu, Anshun Hu, Caikui Wang, Lianbiao Fang","doi":"10.1109/CCISP55629.2022.9974575","DOIUrl":"https://doi.org/10.1109/CCISP55629.2022.9974575","url":null,"abstract":"In recent years, global warming and climate extremes have occurred frequently, seriously affecting human life, production and sustainable development.To study the correlation between carbon emissions and climate change, complex networks and big data statistical analysis methods were used to construct a multi-factor climate network of Yunnan carbon emissions and multi-factor climate network by determining the connectivity of edges through Pearson correlation coefficients, and sliding series correlation was used to analyse the effect of carbon emissions on meteorological factors.The results show that carbon emissions show a positive correlation with temperature, wind speed and sunshine, and a negative correlation with air pressure, precipitation and humidity.On long time scales carbon emissions have an impact on changes in meteorological factors, with an immediate effect on wind speed and a trend from lagging to immediate on precipitation.The study provides some theoretical reference for the control of CO2 emissions in Yunnan Province.","PeriodicalId":431851,"journal":{"name":"2022 7th International Conference on Communication, Image and Signal Processing (CCISP)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114614867","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Target detection task plays the most fundamental and important role in computer vision. The appearance of deep learning method has produced a positive effect on target detection, but multi-scale target detection is poor. The reasons could be attributed to two aspects; the first one is that the small target tends to contain less semantic information, which leads algorithm be hard to detect it; the other is that the sample distribution in the practical application scenarios is random, and the different-scaled target features will interfere with each other, which poses negative effect on multi-scale target detection. Based on existing technical issues, we propose an anchor-free frame for the multi-scale target detection (AFMTD) algorithm as solution. First, from the direction of feature fusion, we propose a spatial attention fusion module (SAFM), which designs same scale transformation (SST) based on Bi-FPN, strengthens the valuable information between adjacent feature layers, and suppresses interference features, improving the detection accuracy and resolution ability of targets of different scales. Then, from the direction of anchor-free frame detection, the heatmap-based multi-scale detection module (HMDM) is proposed; by introducing a scale distribution mechanism (SDM) and Heatmap-IOU (HIOU) loss function, the module allocates different targets to different corresponding feature maps, which makes the model converge faster and more accurately. Through experiments on the MS COCO dataset, our approach achieved 40.5% average precision (AP), and the AP of large, medium, and small-scale targets is 24.5%, 44.1%, and 53.9%, respectively.
目标检测任务是计算机视觉中最基础、最重要的任务。深度学习方法的出现对目标检测产生了积极的影响，但多尺度目标检测效果不佳。原因可以归结为两个方面;一是小目标往往包含较少的语义信息，导致算法难以检测到小目标;二是实际应用场景中的样本分布是随机的，不同尺度的目标特征会相互干扰，对多尺度目标检测产生不利影响。针对目前存在的技术问题，提出了一种无锚框架的多尺度目标检测算法作为解决方案。首先，从特征融合的方向，提出了空间注意融合模块(SAFM)，该模块设计了基于Bi-FPN的同尺度变换(SST)，增强了相邻特征层之间的有价值信息，抑制了干扰特征，提高了不同尺度目标的检测精度和分辨能力。然后，从无锚帧检测的方向，提出了基于热图的多尺度检测模块(HMDM);通过引入SDM (scale distribution mechanism)和HIOU (Heatmap-IOU)损失函数，将不同的目标分配到不同的对应特征映射中，使模型收敛更快、更准确。通过在MS COCO数据集上的实验，我们的方法达到了40.5%的平均精度(AP)，大、中、小目标的平均精度分别为24.5%、44.1%和53.9%。
{"title":"AFMTD: Anchor-free Frame for Multi-scale Target Detection","authors":"Xueting Liu, Jingrou Xu, Ruoxi Lin, Jinyang Pan, Junying Mao, Guangqiang Yin","doi":"10.1109/CCISP55629.2022.9974392","DOIUrl":"https://doi.org/10.1109/CCISP55629.2022.9974392","url":null,"abstract":"Target detection task plays the most fundamental and important role in computer vision. The appearance of deep learning method has produced a positive effect on target detection, but multi-scale target detection is poor. The reasons could be attributed to two aspects; the first one is that the small target tends to contain less semantic information, which leads algorithm be hard to detect it; the other is that the sample distribution in the practical application scenarios is random, and the different-scaled target features will interfere with each other, which poses negative effect on multi-scale target detection. Based on existing technical issues, we propose an anchor-free frame for the multi-scale target detection (AFMTD) algorithm as solution. First, from the direction of feature fusion, we propose a spatial attention fusion module (SAFM), which designs same scale transformation (SST) based on Bi-FPN, strengthens the valuable information between adjacent feature layers, and suppresses interference features, improving the detection accuracy and resolution ability of targets of different scales. Then, from the direction of anchor-free frame detection, the heatmap-based multi-scale detection module (HMDM) is proposed; by introducing a scale distribution mechanism (SDM) and Heatmap-IOU (HIOU) loss function, the module allocates different targets to different corresponding feature maps, which makes the model converge faster and more accurately. Through experiments on the MS COCO dataset, our approach achieved 40.5% average precision (AP), and the AP of large, medium, and small-scale targets is 24.5%, 44.1%, and 53.9%, respectively.","PeriodicalId":431851,"journal":{"name":"2022 7th International Conference on Communication, Image and Signal Processing (CCISP)","volume":"214 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114846197","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-11-01DOI: 10.1109/CCISP55629.2022.9974518
Li Liu, Ke Zhang, Linjun Liu, Le Zhang, Jun Zhang
Artificial intelligence (AI) has become an important means of network anomaly detection and fault root cause analysis (RCA), but most applications are only for a certain segment of the network. In the process of our research on the end-to-end experience analysis of mobile Internet services, we have summarized a set of network end-to-end root cause analysis methods, mainly using non-orthogonal random forest modeling method, AI-based dynamic threshold adjustment and indicator feature extraction, classification modeling and cross-validation of the poor quality of the entire network and the poor quality of the cells. This method has been verified in practice in the production network. The results of root cause analysis are consistent with the actual situation of the production network up to 96%. Practice has proved that this method has played a positive role in supporting the network operation, and greatly improved the production efficiency.
{"title":"Root cause analysis of network fault based on random forest","authors":"Li Liu, Ke Zhang, Linjun Liu, Le Zhang, Jun Zhang","doi":"10.1109/CCISP55629.2022.9974518","DOIUrl":"https://doi.org/10.1109/CCISP55629.2022.9974518","url":null,"abstract":"Artificial intelligence (AI) has become an important means of network anomaly detection and fault root cause analysis (RCA), but most applications are only for a certain segment of the network. In the process of our research on the end-to-end experience analysis of mobile Internet services, we have summarized a set of network end-to-end root cause analysis methods, mainly using non-orthogonal random forest modeling method, AI-based dynamic threshold adjustment and indicator feature extraction, classification modeling and cross-validation of the poor quality of the entire network and the poor quality of the cells. This method has been verified in practice in the production network. The results of root cause analysis are consistent with the actual situation of the production network up to 96%. Practice has proved that this method has played a positive role in supporting the network operation, and greatly improved the production efficiency.","PeriodicalId":431851,"journal":{"name":"2022 7th International Conference on Communication, Image and Signal Processing (CCISP)","volume":"52 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122276289","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-11-01DOI: 10.1109/CCISP55629.2022.9974377
Idrees Fazili, A. Achuthan, M. Mustapha, B. Belaton
Diffusion Tensor Imaging (DTI) allows us to reconstruct the brain white matter (WM) pathways in-vivo. Generating a diffusion tractograph from raw MRI data involves multiple layers of processes. Each set of processes that produces a particular analysis is called a pipeline. An extensive collection of software tools have been developed over the years for each layer of tractograph generation, giving researchers the freedom to choose the tools of their preference for different processes. However, this has resulted in the establishment of various pipelines aimed towards the same task, and depending upon an analysis, one pipeline may be more suitable than the other. This creates a hurdle for the clinical application of the DTI tools, as the clinicians and neuroscience researchers are not usually conversant with the technical aspects of the DTI tools. This study proposes an automated decision model for selection of tractography pipelines that will allow researchers and clinicians to select best of the possible DTI pipelines for a particular Analysis.
{"title":"Conceptual Framework For Optimized Pipeline Selection For Brain Tractography Using Multi-Criteria Decision Analysis","authors":"Idrees Fazili, A. Achuthan, M. Mustapha, B. Belaton","doi":"10.1109/CCISP55629.2022.9974377","DOIUrl":"https://doi.org/10.1109/CCISP55629.2022.9974377","url":null,"abstract":"Diffusion Tensor Imaging (DTI) allows us to reconstruct the brain white matter (WM) pathways in-vivo. Generating a diffusion tractograph from raw MRI data involves multiple layers of processes. Each set of processes that produces a particular analysis is called a pipeline. An extensive collection of software tools have been developed over the years for each layer of tractograph generation, giving researchers the freedom to choose the tools of their preference for different processes. However, this has resulted in the establishment of various pipelines aimed towards the same task, and depending upon an analysis, one pipeline may be more suitable than the other. This creates a hurdle for the clinical application of the DTI tools, as the clinicians and neuroscience researchers are not usually conversant with the technical aspects of the DTI tools. This study proposes an automated decision model for selection of tractography pipelines that will allow researchers and clinicians to select best of the possible DTI pipelines for a particular Analysis.","PeriodicalId":431851,"journal":{"name":"2022 7th International Conference on Communication, Image and Signal Processing (CCISP)","volume":"321 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131985372","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：