Traffic Accident’s Severity Prediction: A Deep-Learning Approach-Based CNN Network

Zheng, Ming; Li, Tong; Zhu, Rui; Chen, Jing; Ma, Zifei; Tang, Mingjing; Zhong-qiang, Cui; Zhan, Wang

doi:10.1109/access.2019.2903319

Cited by 121 publications

(36 citation statements)

References 56 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The performance of BFP-growth-DNNJ48 is measured in terms of accuracy, precision, recall, memory and time utilization. The proposed work is compared with the existing works AdaBoost-SO [15] and TASP-CNN [14].…”

Section: Resultsmentioning

confidence: 99%

“…[ Zheng et al (2019)] proposed a deep learning approach based on Convolutional Neural Network (CNN) for prediction of traffic accident severity. Instead o directly using accident feature into CNN, this method found the feature relationship among the traffic accident features and the weight of feature was assigned based on relationship degree.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Bounded Memory Based Frequent Pattern Growth Approach With Deep Neural Network and Decision Tree for Road Accident Prediction

N¹,

Punithavalli²

2020

INDJCSE

View full text Add to dashboard Cite

Frequent Pattern-growth (FP-growth) was introduced to categorize the accident locations. After the categorization of accident locations, accidents were classified by using J48. FP-growth has poor spatial and temporal locality issues. This is because of the construction process of FP-tree and its access behaviour of the mining algorithm. In this article, the issues of FP-growth are solved by introducing Bounded FPgrowth. It uses only a bounded portion of the primary memory using specialized memory management. When the tree grows out of the allocated memory, it is forced to be partially saved on secondary memory. The secondary memory is accessed in a block-by-block basis so that both temporal and spatial localities of FP-growth are optimized. After the memory management process, FP-growth is applied to categorize the accident locations. The attributes which have high support values are processed in Deep Neural Network J48 (DNNJ48) for classification of accident type. The analysis of experimental results proved that the proposed Bounded FP-growth-Deep Neural Network with J48 classifier (BFP-growth-DNNJ48) achieved higher accuracy (94%) when compared with existing methods like AdaBoost-SO and TASP-CNN.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Bounded Memory Based Frequent Pattern Growth Approach With Deep Neural Network and Decision Tree for Road Accident Prediction

N¹,

Punithavalli²

2020

INDJCSE

View full text Add to dashboard Cite

show abstract

“…As discussed the reason why this method was chosen in this study was this model is similar to MLNN with the difference that the MLNN has been improved by converting a high dimensional data to low dimensional codes through training of MLNN with a small central layer. 24 For the included model, number of epochs was set at 1000 with batch size being adjusted to be as default, 32, and the data were scaled and centered. Scaling was based on min-max normalization to rescale features between 0 and 1.…”

Section: Deep Belief Networkmentioning

confidence: 99%

Application of deep learning techniques in predicting motorcycle crash severity

Rezapour

Nazneen

Ksaibati

2020

Engineering Reports

View full text Add to dashboard Cite

Machine learning (ML) techniques play a crucial role in today's modern world. Over the last years, road traffic safety is one of the applications where ML-methods have been successfully employed to prevent road users from being killed or seriously injured. A reliable data-driven predictive model is essential for this purpose. This could be achieved by successfully applying an intelligent transportation system to identify a driver at a higher risk of crashes. This study investigates the capabilities of different deep learning techniques to predict motorcycle crash severity. This study is based on 2,430 motorcycle crashes in a mountainous area in the United States over a 10-year period. Different deep networks (DNNs), including deep belief network, standard recurrent neural network (RNN), multilayer neural network, and single-layer neural network, were considered and compared in terms of prediction accuracy of motorcycle crash severity. Before conducting any analysis, feature reduction was performed to identify the optimal number of variables to include in the models by minimizing the error rate. Different metrics including the area under the curve and confusion matrix were used to compare the different models. Although the analyses were conducted on a relatively small dataset, the results indicate that almost all the DNN models better perform in predicting the severity of motorcycle crashes, compared with the single layer neural network. Finally, the RNN outperforms the other three neural network models. A comprehensive discussion has been made about the methodological approach implemented in this study.

show abstract

“…Moreover, when compared with the prediction performance of OP, the developed NN model was found to be superior with an accuracy of 74.6%. Zheng et al [ 13 ] investigated the use of convolutional neural networks (CNN) for RTC severity prediction. A comparison was conducted between the proposed CNN model and nine common statistical and ML models such as k-nearest neighbor (KNN), decision trees (DT), SVM, and NN based on crash severity prediction performance.…”

Section: Literature Reviewmentioning

confidence: 99%

Traffic Crash Severity Prediction—A Synergy by Hybrid Principal Component Analysis and Machine Learning Models

Assi

2020

IJERPH

View full text Add to dashboard Cite

The accurate prediction of road traffic crash (RTC) severity contributes to generating crucial information, which can be used to adopt appropriate measures to reduce the aftermath of crashes. This study aims to develop a hybrid system using principal component analysis (PCA) with multilayer perceptron neural networks (MLP-NN) and support vector machines (SVM) in predicting RTC severity. PCA shows that the first nine components have an eigenvalue greater than one. The cumulative variance percentage explained by these principal components was found to be 67%. The prediction accuracies of the models developed using the original attributes were compared with those of the models developed using principal components. It was found that the testing accuracies of MLP-NN and SVM increased from 64.50% and 62.70% to 82.70% and 80.70%, respectively, after using principal components. The proposed models would be beneficial to trauma centers in predicting crash severity with high accuracy so that they would be able to prepare for appropriate and prompt medical treatment.

show abstract

Traffic Accident’s Severity Prediction: A Deep-Learning Approach-Based CNN Network

Cited by 121 publications

References 56 publications

Bounded Memory Based Frequent Pattern Growth Approach With Deep Neural Network and Decision Tree for Road Accident Prediction

Bounded Memory Based Frequent Pattern Growth Approach With Deep Neural Network and Decision Tree for Road Accident Prediction

Application of deep learning techniques in predicting motorcycle crash severity

Traffic Crash Severity Prediction—A Synergy by Hybrid Principal Component Analysis and Machine Learning Models

Contact Info

Product

Resources

About