In the era of big data, the requirement of mass storage and fast access of data makes solid state disk(SSD) based on NAND flash be widely used. However, increasing flash memory capacity imposes huge SRAM consumption for logical-physical translation table in a page-level flash translation layer(FTL). Existing FTL schemes selectively cache the on-demand address mappings to quicken the address translation, while keeping all address mappings in flash memory. But the page-level catching mechanism causes a certain degree of cache pollution. In this paper, we manage page-level address translation information at hybrid-level mapping scheme and use two-level buffer for map groups to decrease SRAM consumption while reducing the cache pollution. What's more, an efficient replacement policy is designed. We can increase the cache hit ratio and reduce the write backs of evicted dirty entries and decrease garbage collection operations by these means. The performance and lifetime of the flash memory is improved. Experimental results show that the proposed scheme increases cache hit ratio by up to 28% and decreases the average response time by up to 23% compared with the existing FTL schemes.
{"title":"Improving Flash Translation Layer Performance by Using Log Block Mapping Scheme and Two-level Buffer for Address Translation Information","authors":"Yinxia Xu","doi":"10.1145/3366715.3366746","DOIUrl":"https://doi.org/10.1145/3366715.3366746","url":null,"abstract":"In the era of big data, the requirement of mass storage and fast access of data makes solid state disk(SSD) based on NAND flash be widely used. However, increasing flash memory capacity imposes huge SRAM consumption for logical-physical translation table in a page-level flash translation layer(FTL). Existing FTL schemes selectively cache the on-demand address mappings to quicken the address translation, while keeping all address mappings in flash memory. But the page-level catching mechanism causes a certain degree of cache pollution. In this paper, we manage page-level address translation information at hybrid-level mapping scheme and use two-level buffer for map groups to decrease SRAM consumption while reducing the cache pollution. What's more, an efficient replacement policy is designed. We can increase the cache hit ratio and reduce the write backs of evicted dirty entries and decrease garbage collection operations by these means. The performance and lifetime of the flash memory is improved. Experimental results show that the proposed scheme increases cache hit ratio by up to 28% and decreases the average response time by up to 23% compared with the existing FTL schemes.","PeriodicalId":425980,"journal":{"name":"Proceedings of the 2019 International Conference on Robotics Systems and Vehicle Technology - RSVT '19","volume":"83 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122898689","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}
With the advent of the Internet era, the content of network information has largely increased, hence information extraction has became significant. As an important sub-task of information extraction, entity relationship extraction is also paid more and more attention. Most current entity relationship extraction methods not only require manual annotation, but the quality of annotation also cannot be guaranteed, besides the evaluation criteria has not been unified yet. Therefore, this paper proposes an entity relationship extraction method based on the combination of dependency syntax analysis and rules. The method does not need to annotate the input text manually, dependency parsing is used to determine the sentence components and the relationships among them. Meanwhile, a semantic triple representing entity relations is formed and output by combining rules. The experiment results shows that the method proposed in this paper has a good effect and saves labor cost. The average accuracy in corpus reaches 63.04%, the average output time of triples is shortened as well.
{"title":"Entity Relationship Extraction Method Based on Dependency Syntax Analysis and Rules","authors":"Xiaolin Li, Jiaying Fan","doi":"10.1145/3366715.3366740","DOIUrl":"https://doi.org/10.1145/3366715.3366740","url":null,"abstract":"With the advent of the Internet era, the content of network information has largely increased, hence information extraction has became significant. As an important sub-task of information extraction, entity relationship extraction is also paid more and more attention. Most current entity relationship extraction methods not only require manual annotation, but the quality of annotation also cannot be guaranteed, besides the evaluation criteria has not been unified yet. Therefore, this paper proposes an entity relationship extraction method based on the combination of dependency syntax analysis and rules. The method does not need to annotate the input text manually, dependency parsing is used to determine the sentence components and the relationships among them. Meanwhile, a semantic triple representing entity relations is formed and output by combining rules. The experiment results shows that the method proposed in this paper has a good effect and saves labor cost. The average accuracy in corpus reaches 63.04%, the average output time of triples is shortened as well.","PeriodicalId":425980,"journal":{"name":"Proceedings of the 2019 International Conference on Robotics Systems and Vehicle Technology - RSVT '19","volume":"83 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126278028","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 this paper, we introduce an improved A* path planning algorithm for indoor intelligent robot. Aiming at the problem of intelligent robot car path planning in complex indoor environment with obstacles such as wall. Firstly, the indoor environment is divided into grids and we can get the connected topology. The connectivity between grids is characterized by adjacency matrices. Then we study the influence of different heuristic functions on the efficiency of path planning algorithm. Based on the traditional A* algorithm, the direction factor is introduced. Moreover we also consider the impact of distance and direction on search efficiency. Finally, the algorithm is simulated by Matlab. The experimental results show that compared with the traditional A* algorithm, the proposed algorithm has a significant improvement in path search efficiency.
{"title":"An Improved A* Path Planning Algorithm for Indoor Intelligent Robot","authors":"Shiyun Qian, Yajie Ma, Doudou Hong","doi":"10.1145/3366715.3366718","DOIUrl":"https://doi.org/10.1145/3366715.3366718","url":null,"abstract":"In this paper, we introduce an improved A* path planning algorithm for indoor intelligent robot. Aiming at the problem of intelligent robot car path planning in complex indoor environment with obstacles such as wall. Firstly, the indoor environment is divided into grids and we can get the connected topology. The connectivity between grids is characterized by adjacency matrices. Then we study the influence of different heuristic functions on the efficiency of path planning algorithm. Based on the traditional A* algorithm, the direction factor is introduced. Moreover we also consider the impact of distance and direction on search efficiency. Finally, the algorithm is simulated by Matlab. The experimental results show that compared with the traditional A* algorithm, the proposed algorithm has a significant improvement in path search efficiency.","PeriodicalId":425980,"journal":{"name":"Proceedings of the 2019 International Conference on Robotics Systems and Vehicle Technology - RSVT '19","volume":"12 2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114383911","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}
Yanwen Jiang, Hua Sun, Gang Jin, Jiaping Chen, X. Qian
Blackheads are a subtype of acne. As a cosmetic problem, it seriously affects patient's facial appearance and psychological condition. So it has attracted more and more public attention in recent years. However due to the evaluation standards are not uniformed, the grade methods of existing acne still are lack of objective quantitative standard. Even for professionals with long training, there remains great variability among the evaluators. The experimental shows that the new intelligent method is similar to the results of professional dermatologists in terms of blackheads counting and has high efficiency advantages. What's more, it achieves the leap from qualitative to quantitative analysis in blackheads identification field.
{"title":"One Intelligent Facial Blackheads Identification Method Based on Computer Vision Technology","authors":"Yanwen Jiang, Hua Sun, Gang Jin, Jiaping Chen, X. Qian","doi":"10.1145/3366715.3366736","DOIUrl":"https://doi.org/10.1145/3366715.3366736","url":null,"abstract":"Blackheads are a subtype of acne. As a cosmetic problem, it seriously affects patient's facial appearance and psychological condition. So it has attracted more and more public attention in recent years. However due to the evaluation standards are not uniformed, the grade methods of existing acne still are lack of objective quantitative standard. Even for professionals with long training, there remains great variability among the evaluators. The experimental shows that the new intelligent method is similar to the results of professional dermatologists in terms of blackheads counting and has high efficiency advantages. What's more, it achieves the leap from qualitative to quantitative analysis in blackheads identification field.","PeriodicalId":425980,"journal":{"name":"Proceedings of the 2019 International Conference on Robotics Systems and Vehicle Technology - RSVT '19","volume":"221 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122517995","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}
People usually perceive things from a three-dimensional real world, requiring the computer to automatically reconstruct the corresponding three-dimensional shape model using two-dimensional human face images. The three-dimensional face reconstruction based on binocular vision is to use SIFT algorithm for feature point matching and disparity calculation to obtain three-dimensional coordinates. The three-dimensional coordinates are applied directly to the face deformation model, which overcomes the traditional method of finding two-dimensional and its coordinates, convert the relationship, the disadvantage of low precision, thus reconstruct a three-dimensional face model with a strong sense of reality.The experimental results show that the algorithm can be stably matched under illumination, blur, noise, near and far, and the experimental error of Zhang zheng you's calibration method is controlled within 0.4 pixel, and accuracy to meet the general application requirements.The three-dimensional reconstruction of the indoor scene can also learn from this method.
{"title":"Research On 3D Reconstruction of Face Based on Binocualr Stereo Vision","authors":"Ziwei Wang, Haihui Wang, Jianing Li","doi":"10.1145/3366715.3366745","DOIUrl":"https://doi.org/10.1145/3366715.3366745","url":null,"abstract":"People usually perceive things from a three-dimensional real world, requiring the computer to automatically reconstruct the corresponding three-dimensional shape model using two-dimensional human face images. The three-dimensional face reconstruction based on binocular vision is to use SIFT algorithm for feature point matching and disparity calculation to obtain three-dimensional coordinates. The three-dimensional coordinates are applied directly to the face deformation model, which overcomes the traditional method of finding two-dimensional and its coordinates, convert the relationship, the disadvantage of low precision, thus reconstruct a three-dimensional face model with a strong sense of reality.The experimental results show that the algorithm can be stably matched under illumination, blur, noise, near and far, and the experimental error of Zhang zheng you's calibration method is controlled within 0.4 pixel, and accuracy to meet the general application requirements.The three-dimensional reconstruction of the indoor scene can also learn from this method.","PeriodicalId":425980,"journal":{"name":"Proceedings of the 2019 International Conference on Robotics Systems and Vehicle Technology - RSVT '19","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122802271","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}
Collective adaptive systems (CAS) have been attracting increasing attention in the field of artificial intelligence (AI), in which collaboration of agents plays a key role. These systems aim to accomplish a certain goal though collaborating between a variety of agents with different tasks, which adapt to changes of environment to be of adaptability. To solve the issue of collaboration of agents in an uncertain and highly dynamic environment, our research team had proposed a Goal-Capability-Commitment (GCC) based mediation for multi-agent collaboration, which generates the collaboration planning driven by capability based on global context states in real-time dynamic environment. As a case study for the application of GCC model, this paper adopts GCC to model the RoboCup Rescue Simulation System (RCRSS). As the result of modelling, the GCC domain model is applied to RCRSS where an efficiently quantitative evaluation is provided.
{"title":"Research on Dynamic Decision-making of Post-disaster Rescue Based on GCC Framework","authors":"Longlong Xu, Wei Liu, J. Guo","doi":"10.1145/3366715.3366743","DOIUrl":"https://doi.org/10.1145/3366715.3366743","url":null,"abstract":"Collective adaptive systems (CAS) have been attracting increasing attention in the field of artificial intelligence (AI), in which collaboration of agents plays a key role. These systems aim to accomplish a certain goal though collaborating between a variety of agents with different tasks, which adapt to changes of environment to be of adaptability. To solve the issue of collaboration of agents in an uncertain and highly dynamic environment, our research team had proposed a Goal-Capability-Commitment (GCC) based mediation for multi-agent collaboration, which generates the collaboration planning driven by capability based on global context states in real-time dynamic environment. As a case study for the application of GCC model, this paper adopts GCC to model the RoboCup Rescue Simulation System (RCRSS). As the result of modelling, the GCC domain model is applied to RCRSS where an efficiently quantitative evaluation is provided.","PeriodicalId":425980,"journal":{"name":"Proceedings of the 2019 International Conference on Robotics Systems and Vehicle Technology - RSVT '19","volume":"42 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134639903","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}
This paper proposes an algorithm about real-time estimation of queue length using the data from both connected vehicle (CV) and a detector for both under-saturated and over-saturated situations. None of the penetration ratio, signal timing plan or traffic volume is needed as input, making the model more applicable. The resolution reaches second level, depending on the sampling rate of devices. The detector is placed a certain distance away from the stop line so that vehicle's queuing behavior is more predictable. It greatly improved the accuracy especially when there is few CVs. To make the results more robust and accurate, the upper bound of the queue length is estimated using the data of moving CVs and car following model for the first time. The estimation algorithm is verified by the simulation in VISSIM. The relationship between estimation accuracy and market penetration ratio, traffic volume is also analyzed. Results show that only 10% CVs are needed in under-saturated traffic flow and 30% CVs are needed in over-saturated traffic flow.
{"title":"Real-time Estimation of Queue Length Based on Fused Data Using Connected Vehicle Technology and A Detector","authors":"Wenqiang Jin, Xi Zhang, Kaijiong Zhang","doi":"10.1145/3366715.3366719","DOIUrl":"https://doi.org/10.1145/3366715.3366719","url":null,"abstract":"This paper proposes an algorithm about real-time estimation of queue length using the data from both connected vehicle (CV) and a detector for both under-saturated and over-saturated situations. None of the penetration ratio, signal timing plan or traffic volume is needed as input, making the model more applicable. The resolution reaches second level, depending on the sampling rate of devices. The detector is placed a certain distance away from the stop line so that vehicle's queuing behavior is more predictable. It greatly improved the accuracy especially when there is few CVs. To make the results more robust and accurate, the upper bound of the queue length is estimated using the data of moving CVs and car following model for the first time. The estimation algorithm is verified by the simulation in VISSIM. The relationship between estimation accuracy and market penetration ratio, traffic volume is also analyzed. Results show that only 10% CVs are needed in under-saturated traffic flow and 30% CVs are needed in over-saturated traffic flow.","PeriodicalId":425980,"journal":{"name":"Proceedings of the 2019 International Conference on Robotics Systems and Vehicle Technology - RSVT '19","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123183484","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}
It is found that the prediction model has great influence on the performance of stock index. The traditional ensemble learning model has some problems such as limited use of high performance basic classifiers in stock index regression prediction. In this paper, it is found that there is a certain degree of complementarity between basic classifiers. In order to make use of the complementarity of different models, this paper proposes a dynamic weighted ensemble learning model for stock index prediction. The experimental results show that the dynamic weighted ensemble learning model is more accurate than the single basic classifier and is suitable for the regression prediction of different stock indexes.
{"title":"Stock Index Prediction Method Based on Dynamic Weighted Ensemble Learning","authors":"Datao You, Xiangyu Yao, Xudong Geng, Xuyang Fang, Shenming Qu","doi":"10.1145/3366715.3366727","DOIUrl":"https://doi.org/10.1145/3366715.3366727","url":null,"abstract":"It is found that the prediction model has great influence on the performance of stock index. The traditional ensemble learning model has some problems such as limited use of high performance basic classifiers in stock index regression prediction. In this paper, it is found that there is a certain degree of complementarity between basic classifiers. In order to make use of the complementarity of different models, this paper proposes a dynamic weighted ensemble learning model for stock index prediction. The experimental results show that the dynamic weighted ensemble learning model is more accurate than the single basic classifier and is suitable for the regression prediction of different stock indexes.","PeriodicalId":425980,"journal":{"name":"Proceedings of the 2019 International Conference on Robotics Systems and Vehicle Technology - RSVT '19","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114132494","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}
Traditional traffic accident identification methods have the problems of complex detection process, poor detection performance and poor real-time performance so far. In this paper, we propose a new type of traffic accident identification method based on target detection algorithm Single Shot MultiBox Detector (SSD). We collect and simulate traffic accident data sets in different scenarios and compare the detection performance of different target detection algorithms, aiming at the problems of traffic accident detection existing in the original SSD, the idea of multi-feature fusion and adaptive default box selection algorithm are proposed to improve it. Finally, we present an evaluation on the collected data, the improved SSD_A method shows considerable performance, which can reach 97% mAP (mean average precision) at the speed of 32 FPS (frames per second).
{"title":"Fast Traffic Accident Identification Method Based on SSD Model","authors":"Haiyang Jiang, Yuning Wang, Yong Yang","doi":"10.1145/3366715.3366721","DOIUrl":"https://doi.org/10.1145/3366715.3366721","url":null,"abstract":"Traditional traffic accident identification methods have the problems of complex detection process, poor detection performance and poor real-time performance so far. In this paper, we propose a new type of traffic accident identification method based on target detection algorithm Single Shot MultiBox Detector (SSD). We collect and simulate traffic accident data sets in different scenarios and compare the detection performance of different target detection algorithms, aiming at the problems of traffic accident detection existing in the original SSD, the idea of multi-feature fusion and adaptive default box selection algorithm are proposed to improve it. Finally, we present an evaluation on the collected data, the improved SSD_A method shows considerable performance, which can reach 97% mAP (mean average precision) at the speed of 32 FPS (frames per second).","PeriodicalId":425980,"journal":{"name":"Proceedings of the 2019 International Conference on Robotics Systems and Vehicle Technology - RSVT '19","volume":"57 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121243310","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 order to solve the problem of Chinese medical image text being missed and misdetected under the CTPN model, a new convolutional neural network DVNet based on the fusion of VGG convolutional neural network and DenseNet dense network was proposed. DVNet takes the first two layers of VGG network for deep feature extraction, and then connects DenseNet dense modules. Using the idea of feature reuse, the features of the front convolutional layer and the features of the back convolutional layer are output together. During post-processing, NMS is used to filter out redundant text boxes. In the Chinese medical text data set provided, three different networks, VGG, DenseNet and DVNet, were used to detect the text. The experimental results showed that the precision rate of DVNet were improved by 2%-3% compared with VGG and DenseNet.
{"title":"Medical Image Text Area Detection Based on Feature Reuse Convolutional Neural Network","authors":"Yang Liu, Jun Liu, S. Sun, Zhuang Du","doi":"10.1145/3366715.3366738","DOIUrl":"https://doi.org/10.1145/3366715.3366738","url":null,"abstract":"In order to solve the problem of Chinese medical image text being missed and misdetected under the CTPN model, a new convolutional neural network DVNet based on the fusion of VGG convolutional neural network and DenseNet dense network was proposed. DVNet takes the first two layers of VGG network for deep feature extraction, and then connects DenseNet dense modules. Using the idea of feature reuse, the features of the front convolutional layer and the features of the back convolutional layer are output together. During post-processing, NMS is used to filter out redundant text boxes. In the Chinese medical text data set provided, three different networks, VGG, DenseNet and DVNet, were used to detect the text. The experimental results showed that the precision rate of DVNet were improved by 2%-3% compared with VGG and DenseNet.","PeriodicalId":425980,"journal":{"name":"Proceedings of the 2019 International Conference on Robotics Systems and Vehicle Technology - RSVT '19","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129255505","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：