Gurpreet Kaur, Yasir Malik, Hamman W. Samuel, Fehmi Jaafar
Cross-site scripting (XSS) is a scripting attack targeting web applications by injecting malicious scripts into web pages. Blind XSS is a subset of stored XSS, where an attacker blindly deploys malicious payloads in web pages that are stored in a persistent manner on target servers. Most of the XSS detection techniques used to detect the XSS vulnerabilities are inadequate to detect blind XSS attacks. In this research, we present machine learning based approach to detect blind XSS attacks. Testing results help to identify malicious payloads that are likely to get stored in databases through web applications.
{"title":"Detecting Blind Cross-Site Scripting Attacks Using Machine Learning","authors":"Gurpreet Kaur, Yasir Malik, Hamman W. Samuel, Fehmi Jaafar","doi":"10.1145/3297067.3297096","DOIUrl":"https://doi.org/10.1145/3297067.3297096","url":null,"abstract":"Cross-site scripting (XSS) is a scripting attack targeting web applications by injecting malicious scripts into web pages. Blind XSS is a subset of stored XSS, where an attacker blindly deploys malicious payloads in web pages that are stored in a persistent manner on target servers. Most of the XSS detection techniques used to detect the XSS vulnerabilities are inadequate to detect blind XSS attacks. In this research, we present machine learning based approach to detect blind XSS attacks. Testing results help to identify malicious payloads that are likely to get stored in databases through web applications.","PeriodicalId":340004,"journal":{"name":"International Conference on Signal Processing and Machine Learning","volume":"139 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131427974","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}
Jiahui Wang, Kun Yang, Jianhai Zhang, N. Zhang, Bin Chen
The present research was aimed to find out EEG potentials related to movement in exercise-induced fatigue task using brain function network analysis, so that future researchers can find more accurate mutual informations between these potentials to detect fatigue to make healthy people exercise better and especially improve the effectiveness of rehabilitation in patients with motor dysfunction. EEG signals from 32 electrode sites of 20 subjects(10 adults (5 females and 5 males) and 10 children (6 females and 4 males) were recorded. We applied network topologies extracted from brain function networks constructed by phase synchronization to identify movement-related electrode sites. We first found that there were significant differences on the global network topologies of subjects of different ages and genders, and the difference between subjects of different ages was greater, so adults and children in the subjects were separated to discuss potential selection related to movement. The following finding illustrated that local network topologies of some electrode sites correlated significantly with the degree of fatigue, we thought and selected such electrode sites to be movement-related. Results showed that 17 potentials in adults, 6 most relevant potentials as important potentials(CP5,C3,AF4,CZ,PZ,C4), and 4 potentials (F4,F8,F3,FC5) in children were selected as movement-related EEG potentials associated with exercise-induced fatigue in rotating the forearm repetitively task. We demonstrated that the credibility of our selections by observing the classification accuracy of local network topologies of non-fatigue state and fatigue state in our selected electrode sites was higher than that of local network topologies of non-fatigue state and fatigue state in our unselected electrode sites, which suggested that our selected movement-related electrode sites were more able to detect non-fatigue state and fatigue state.
{"title":"Brain Function Networks Reveal Movement-related EEG Potentials Associated with Exercise-induced Fatigue","authors":"Jiahui Wang, Kun Yang, Jianhai Zhang, N. Zhang, Bin Chen","doi":"10.1145/3297067.3297074","DOIUrl":"https://doi.org/10.1145/3297067.3297074","url":null,"abstract":"The present research was aimed to find out EEG potentials related to movement in exercise-induced fatigue task using brain function network analysis, so that future researchers can find more accurate mutual informations between these potentials to detect fatigue to make healthy people exercise better and especially improve the effectiveness of rehabilitation in patients with motor dysfunction. EEG signals from 32 electrode sites of 20 subjects(10 adults (5 females and 5 males) and 10 children (6 females and 4 males) were recorded. We applied network topologies extracted from brain function networks constructed by phase synchronization to identify movement-related electrode sites.\u0000 We first found that there were significant differences on the global network topologies of subjects of different ages and genders, and the difference between subjects of different ages was greater, so adults and children in the subjects were separated to discuss potential selection related to movement. The following finding illustrated that local network topologies of some electrode sites correlated significantly with the degree of fatigue, we thought and selected such electrode sites to be movement-related. Results showed that 17 potentials in adults, 6 most relevant potentials as important potentials(CP5,C3,AF4,CZ,PZ,C4), and 4 potentials (F4,F8,F3,FC5) in children were selected as movement-related EEG potentials associated with exercise-induced fatigue in rotating the forearm repetitively task. We demonstrated that the credibility of our selections by observing the classification accuracy of local network topologies of non-fatigue state and fatigue state in our selected electrode sites was higher than that of local network topologies of non-fatigue state and fatigue state in our unselected electrode sites, which suggested that our selected movement-related electrode sites were more able to detect non-fatigue state and fatigue state.","PeriodicalId":340004,"journal":{"name":"International Conference on Signal Processing and Machine Learning","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131202418","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}
Hongli Lin, Zhenzhen Kong, Weisheng Wang, K. Liang, Jun Chen
With the development of artificial intelligence, pedestrian detection has become an important research topic in the field of intelligent video surveillance. Fish-eye camera is a useful tool for video monitoring. However, due to the edge distortion of the fish-eye image, which puts higher requirements and challenges on the pedestrian detection technology of fish-eye images. In this paper, an effective method is proposed by rotating cutting to address the problem, a fish-eye image is divided into an edge portion and a center portion. The effectiveness and performance of our method is verified by the traditional pedestrian detection method HOG+SVM and the Faster R-CNN based on convolutional neural network. The experimental results demonstrate the efficacy of the proposed approach, and Faster R-CNN achieves better performance than traditional method.
{"title":"Pedestrian Detection in Fish-eye Images using Deep Learning: Combine Faster R-CNN with an effective Cutting Method","authors":"Hongli Lin, Zhenzhen Kong, Weisheng Wang, K. Liang, Jun Chen","doi":"10.1145/3297067.3297069","DOIUrl":"https://doi.org/10.1145/3297067.3297069","url":null,"abstract":"With the development of artificial intelligence, pedestrian detection has become an important research topic in the field of intelligent video surveillance. Fish-eye camera is a useful tool for video monitoring. However, due to the edge distortion of the fish-eye image, which puts higher requirements and challenges on the pedestrian detection technology of fish-eye images. In this paper, an effective method is proposed by rotating cutting to address the problem, a fish-eye image is divided into an edge portion and a center portion. The effectiveness and performance of our method is verified by the traditional pedestrian detection method HOG+SVM and the Faster R-CNN based on convolutional neural network. The experimental results demonstrate the efficacy of the proposed approach, and Faster R-CNN achieves better performance than traditional method.","PeriodicalId":340004,"journal":{"name":"International Conference on Signal Processing and Machine Learning","volume":"49 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121729387","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}
Feature selection is an important preprocessing stage in signal processing and machine learning. Feature selection methods choose the most informative feature subset for classification. Mutual information and conditional mutual information are used extensively in feature selection methods. However, mutual information suffers from an overestimation problem, with conditional mutual information suffering from a problem of underestimation. To address the issues of overestimation and underestimation, we introduce a new measure named part mutual information that could accurately quantify direct association among variables. The proposed method selects the maximal value of cumulative summation of the part mutual information between candidate features and class labels when each selected feature is known. To evaluate the classification performance of the proposed method, our method is compared with four state-of the-art feature selection methods on twelve real-world data sets. Extensive studies demonstrate that our method outperforms the four compared methods in terms of average classification accuracy and the highest classification accuracy.
{"title":"Feature Selection by Maximizing Part Mutual Information","authors":"Wanfu Gao, Liang Hu, Ping Zhang","doi":"10.1145/3297067.3297068","DOIUrl":"https://doi.org/10.1145/3297067.3297068","url":null,"abstract":"Feature selection is an important preprocessing stage in signal processing and machine learning. Feature selection methods choose the most informative feature subset for classification. Mutual information and conditional mutual information are used extensively in feature selection methods. However, mutual information suffers from an overestimation problem, with conditional mutual information suffering from a problem of underestimation. To address the issues of overestimation and underestimation, we introduce a new measure named part mutual information that could accurately quantify direct association among variables. The proposed method selects the maximal value of cumulative summation of the part mutual information between candidate features and class labels when each selected feature is known. To evaluate the classification performance of the proposed method, our method is compared with four state-of the-art feature selection methods on twelve real-world data sets. Extensive studies demonstrate that our method outperforms the four compared methods in terms of average classification accuracy and the highest classification accuracy.","PeriodicalId":340004,"journal":{"name":"International Conference on Signal Processing and Machine Learning","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116923773","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 depth estimation can facilitate classification of surface ships or water-column targets thus reducing the false rates in active surveillance systems. Active sonar mainly determines the distance of the target by measuring the roundtrip time of the transmitted signal to the received echo, but it can't determine the depth of the target. For the long distance sound field, the echo is regarded as a point source sound field emitted from the reflector, the distance-depth space is divided into grids, and the sound field at each grid point is calculated according to the parameters of the ocean environment, and then matched with the received echoes, the best match point is the distance and depth of the target. In the active matched field depth-estimation algorithm, the pulse signal generated by the active sonar is sent to the transmitter to generate sound wave, at the same time, it is sent to the emission model to calculate the copy field of the hypothetical target point, and then the reflected sound field of the hypothetical target is calculated through the reflection model, finally, calculate the total copy vector at the receiving hydrophone array. The active matched field processor matches the received echo signal with the calculated total copy vector and outputs an ambiguity surface, it can be seen that the active matched field processing makes full use of the ocean environment. Since the active sonar has estimated the distance of the target according to the arrival time of the echo, the matched field depth estimation is to search for the target depth in a small range so as to determine the depth of the target. Sea trial data show that under good hydrological conditions, when the SNR of target echo is relatively high, the low-frequency active towed array sonar has good depth estimation capability.
{"title":"Target-depth Estimation for Active Towed Array Sonar in Shallow Sea base on Matched Field Processing","authors":"Jun Wang, Fuchen Liu","doi":"10.1145/3297067.3297087","DOIUrl":"https://doi.org/10.1145/3297067.3297087","url":null,"abstract":"Target depth estimation can facilitate classification of surface ships or water-column targets thus reducing the false rates in active surveillance systems. Active sonar mainly determines the distance of the target by measuring the roundtrip time of the transmitted signal to the received echo, but it can't determine the depth of the target. For the long distance sound field, the echo is regarded as a point source sound field emitted from the reflector, the distance-depth space is divided into grids, and the sound field at each grid point is calculated according to the parameters of the ocean environment, and then matched with the received echoes, the best match point is the distance and depth of the target. In the active matched field depth-estimation algorithm, the pulse signal generated by the active sonar is sent to the transmitter to generate sound wave, at the same time, it is sent to the emission model to calculate the copy field of the hypothetical target point, and then the reflected sound field of the hypothetical target is calculated through the reflection model, finally, calculate the total copy vector at the receiving hydrophone array. The active matched field processor matches the received echo signal with the calculated total copy vector and outputs an ambiguity surface, it can be seen that the active matched field processing makes full use of the ocean environment. Since the active sonar has estimated the distance of the target according to the arrival time of the echo, the matched field depth estimation is to search for the target depth in a small range so as to determine the depth of the target. Sea trial data show that under good hydrological conditions, when the SNR of target echo is relatively high, the low-frequency active towed array sonar has good depth estimation capability.","PeriodicalId":340004,"journal":{"name":"International Conference on Signal Processing and Machine Learning","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129305764","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}
H. Cao, Chao Wang, Ping Wang, Qingquan Zou, Xiao Xiao
In this paper, we present an unsupervised learning based approach to conduct depth estimation for monocular camera video images. Our system is formed by two convolutional neural networks (CNNs). A Depth-net is applied to estimate the depth information of objects in the target frame, and a Pose-net tends to estimate the relative motion of the camera from multiple adjacent video frames. Different from most previous works, which normally assume that all objects captured by the images are static so that a frame-level camera pose is generated by the Pose-net, we take into account of the motions of all objects and require the Pose-net to estimate the pixel-level relative pose. The outputs of the two networks are then combined to formulate a synthetic view loss function, through which the two CNNs are optimized to provide accurate depth estimation. Experimental test results show that our method can provide better performance than most conventional approaches.
{"title":"Unsupervised Depth Estimation from Monocular Video based on Relative Motion","authors":"H. Cao, Chao Wang, Ping Wang, Qingquan Zou, Xiao Xiao","doi":"10.1145/3297067.3297094","DOIUrl":"https://doi.org/10.1145/3297067.3297094","url":null,"abstract":"In this paper, we present an unsupervised learning based approach to conduct depth estimation for monocular camera video images. Our system is formed by two convolutional neural networks (CNNs). A Depth-net is applied to estimate the depth information of objects in the target frame, and a Pose-net tends to estimate the relative motion of the camera from multiple adjacent video frames. Different from most previous works, which normally assume that all objects captured by the images are static so that a frame-level camera pose is generated by the Pose-net, we take into account of the motions of all objects and require the Pose-net to estimate the pixel-level relative pose. The outputs of the two networks are then combined to formulate a synthetic view loss function, through which the two CNNs are optimized to provide accurate depth estimation. Experimental test results show that our method can provide better performance than most conventional approaches.","PeriodicalId":340004,"journal":{"name":"International Conference on Signal Processing and Machine Learning","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123231843","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}
David R Ha, Hideyuki Watanabe, Yuya Tomotoshi, Emilie Delattre, S. Katagiri
We proposed a novel method that selects an optimal classifier model's parameter status through the uncertainty measure evaluation of the estimated class boundaries instead of an estimation of the classification error probability. A key feature of our method is its potential to perform a classifier parameter status selection without a separate validation sample set that can be easily applied to any reasonable type of classifier model, unlike traditional approaches that often need a validation sample set or are sometimes less practical. In this paper, we first summarize our method and its experimental evaluation results and introduce the mathematical formalization for the posterior probability estimation procedure adopted in it. Then we show the convergence property of the estimation procedure and finally demonstrate our method's optimality in a practical situation where only a finite number of training samples are available.
{"title":"Optimality Analysis of Boundary-Uncertainty-Based Classifier Model Parameter Status Selection Method","authors":"David R Ha, Hideyuki Watanabe, Yuya Tomotoshi, Emilie Delattre, S. Katagiri","doi":"10.1145/3297067.3297076","DOIUrl":"https://doi.org/10.1145/3297067.3297076","url":null,"abstract":"We proposed a novel method that selects an optimal classifier model's parameter status through the uncertainty measure evaluation of the estimated class boundaries instead of an estimation of the classification error probability. A key feature of our method is its potential to perform a classifier parameter status selection without a separate validation sample set that can be easily applied to any reasonable type of classifier model, unlike traditional approaches that often need a validation sample set or are sometimes less practical. In this paper, we first summarize our method and its experimental evaluation results and introduce the mathematical formalization for the posterior probability estimation procedure adopted in it. Then we show the convergence property of the estimation procedure and finally demonstrate our method's optimality in a practical situation where only a finite number of training samples are available.","PeriodicalId":340004,"journal":{"name":"International Conference on Signal Processing and Machine Learning","volume":"171 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117302536","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}
As new gadgets that interact with the user through voice become accessible, the importance of not only the content of the speech increases, but also the significance of the way the user has spoken. Even though many techniques have been developed to indicate emotion on speech, none of them can fully grasp the real emotion of the speaker. This paper presents a neural network model capable of predicting emotions in conversations by analyzing transcriptions and raw audio waveforms, focusing on feature extraction using convolutional layers and feature combination. The model achieves an accuracy of over 71% across four classes: Anger, Happiness, Neutrality and Sadness. We also analyze the effect of audio and textual features on the classification task, by interpreting attention scores and parts of speech. This paper explores the use of raw audio waveforms, that in the best of our knowledge, have not yet been used deeply in the emotion classification task, achieving close to state of art results.
{"title":"Speech Emotion Classification using Raw Audio Input and Transcriptions","authors":"Gabriel Lima, Jinyeong Bak","doi":"10.1145/3297067.3297089","DOIUrl":"https://doi.org/10.1145/3297067.3297089","url":null,"abstract":"As new gadgets that interact with the user through voice become accessible, the importance of not only the content of the speech increases, but also the significance of the way the user has spoken. Even though many techniques have been developed to indicate emotion on speech, none of them can fully grasp the real emotion of the speaker. This paper presents a neural network model capable of predicting emotions in conversations by analyzing transcriptions and raw audio waveforms, focusing on feature extraction using convolutional layers and feature combination. The model achieves an accuracy of over 71% across four classes: Anger, Happiness, Neutrality and Sadness. We also analyze the effect of audio and textual features on the classification task, by interpreting attention scores and parts of speech. This paper explores the use of raw audio waveforms, that in the best of our knowledge, have not yet been used deeply in the emotion classification task, achieving close to state of art results.","PeriodicalId":340004,"journal":{"name":"International Conference on Signal Processing and Machine Learning","volume":"82 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133684582","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}
Autonomous vehicles are widely accepted as one of the most potential technologies in alleviating traffic problems. In most existing autonomous vehicles for object detection and distance measurement, compared with radar or LIDAR which obviously increases the cost, camera combined with Convolutional Neural Network (CNN) has advantage in accuracy and low cost. However, most object detection methods applied on camera cannot perform distance measurement. In this paper, we simultaneously carry out real-time object detection and distance measurement (DDM) in one system by utilizing CNN on a binocular camera. Firstly, a binocular camera is used to acquire disparity maps. Secondly, a set of high-quality region proposals is generated by those disparity maps and the number of region proposals is reduced. Thirdly, CNN is utilized to classify those region proposals and get the bounding box of detected objects. Consequently, those reduced region proposals generated by disparity maps lead to improved computational efficiency. Finally, the object distance is measured by the disparity map and the bounding box. The experiment results show that the proposed method can achieve an accuracy of 87.2% on KITTI dataset and an accuracy of 68% in the real environment for object detection. The average relative error of the distance measurement is 0.85% within 10 meters in real environment. The operation time of the whole DDM system is less than 80 ms.
{"title":"Object Detection Based on Binocular Vision with Convolutional Neural Network","authors":"Zekun Luo, Xia Wu, Qingquan Zou, Xiao Xiao","doi":"10.1145/3297067.3297081","DOIUrl":"https://doi.org/10.1145/3297067.3297081","url":null,"abstract":"Autonomous vehicles are widely accepted as one of the most potential technologies in alleviating traffic problems. In most existing autonomous vehicles for object detection and distance measurement, compared with radar or LIDAR which obviously increases the cost, camera combined with Convolutional Neural Network (CNN) has advantage in accuracy and low cost. However, most object detection methods applied on camera cannot perform distance measurement. In this paper, we simultaneously carry out real-time object detection and distance measurement (DDM) in one system by utilizing CNN on a binocular camera. Firstly, a binocular camera is used to acquire disparity maps. Secondly, a set of high-quality region proposals is generated by those disparity maps and the number of region proposals is reduced. Thirdly, CNN is utilized to classify those region proposals and get the bounding box of detected objects. Consequently, those reduced region proposals generated by disparity maps lead to improved computational efficiency. Finally, the object distance is measured by the disparity map and the bounding box. The experiment results show that the proposed method can achieve an accuracy of 87.2% on KITTI dataset and an accuracy of 68% in the real environment for object detection. The average relative error of the distance measurement is 0.85% within 10 meters in real environment. The operation time of the whole DDM system is less than 80 ms.","PeriodicalId":340004,"journal":{"name":"International Conference on Signal Processing and Machine Learning","volume":"67 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124931248","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 development of society, the number of vehicles increases rapidly. The vehicle plays an important role in people's life, however the problem of traffic safety caused by vehicles has also become increasingly prominent. In China, the high crash rate and casualty rate on expressways have always troubled traffic management department. So crash prediction on expressway becomes vital. Conventionally, crash prediction is based on traffic flow data. These data do not contain all the necessary factors. In this paper, we propose a method of prediction using real-world data, including historical accident data, road geometry data, vehicle speed data, and weather data. We treat the crash prediction problem as a binary classification problem. For classification, sample imbalanced is a great challenge in practice. Modifying sample weights is applied to handle this challenge. Three machine learning classification techniques, namely Random Forest (RF), Gradient Boosting Decision Tree (GBDT) and Xgboost, are considered to carry out the crash prediction task respectively. The best recall and precision rate of these models are respectively 0.764253 and 0.01062. The proposed method can be integrated into urban traffic control systems toward police dispatch and crash prevention.
{"title":"Expressway Crash Prediction based on Traffic Big Data","authors":"Hailang Meng, Xinhong Wang, X. Wang","doi":"10.1145/3297067.3297093","DOIUrl":"https://doi.org/10.1145/3297067.3297093","url":null,"abstract":"With the development of society, the number of vehicles increases rapidly. The vehicle plays an important role in people's life, however the problem of traffic safety caused by vehicles has also become increasingly prominent. In China, the high crash rate and casualty rate on expressways have always troubled traffic management department. So crash prediction on expressway becomes vital. Conventionally, crash prediction is based on traffic flow data. These data do not contain all the necessary factors. In this paper, we propose a method of prediction using real-world data, including historical accident data, road geometry data, vehicle speed data, and weather data. We treat the crash prediction problem as a binary classification problem. For classification, sample imbalanced is a great challenge in practice. Modifying sample weights is applied to handle this challenge. Three machine learning classification techniques, namely Random Forest (RF), Gradient Boosting Decision Tree (GBDT) and Xgboost, are considered to carry out the crash prediction task respectively. The best recall and precision rate of these models are respectively 0.764253 and 0.01062. The proposed method can be integrated into urban traffic control systems toward police dispatch and crash prevention.","PeriodicalId":340004,"journal":{"name":"International Conference on Signal Processing and Machine Learning","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114248185","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：