Machine Learning을 이용한 자동 돌발상황검지
Automatic Incident Detection Using Machine Learning
0
조회
0
다운로드
3
인용
0
참고문헌
키워드
초록
단속류 및 연속류의 도로상에서 발생하는 돌발상황은 심각한 교통혼잡을 야기할 뿐만 아니라 2차적 교통사고로 이어질 수 있으며, 이로 인해 매우 큰 사회적비용을 초래한다. 따라서 교통관리센타에서는 예측 불가능한 돌발상황에 신속하고 효과적인 대응을 하기 위해서, 돌발상황에 대한 보다 신속하고 정확한 검지가 요구되어 왔다. 특히 연속류 고속도로상에 설치되어있는 검지기에서 수집되는 교통량, 속도, 점유율 등의 교통정보를 활용하여 돌발상황 발생을 자동으로 검지하는 알고리즘에 대한 연구는, 현재 세계적으로 관심이 고조되고 있는 지능형교통체계분야(ITS)의 주요한 연구분야로 인식되고 있으며, 지금까지 다양한 형태의 알고리즘들이 개발되어왔다. 그러나 지금까지 개발된 기존의 많은 알고리즘들은 낮은 검지율 및 높은 오검지율을 보여 왔으며, 따라서 보다 신뢰성이 높은 돌발상황에 대한 검지알고리즘의 개발이 절실히 요구된다. 현재까지 개발된 돌발상황검지를 위한 알고리즘 중 가장 신뢰성이 높고 실시간 검지에 적합한 모형으로 신경망모형이 인식되고 있다. 본 연구에서는 기존의 신경망모형을 이용한 알고리즘보다 높은 검지율 및 낮은 오검지율을 갖는 보다 우수한 실시간 돌발상황검지 알고리즘을 제시하고자 한다. 이를 위해 최근에 패턴인식분야에서 몇몇 연구자들에 의해 전통적인 신경망 모형인 Backpropagation 모형보다 우수한 것으로 평가되고 있는 SVM(Support Vector Machine)을 사용하였다. 본 연구에서는 서울시 내부고속도로에서 수집한 교통상황 데이터 및 돌발상황에 대한 데이터를 이용하여 두 모형, 즉 Backpropagation 과 SVM에 대한 검지능력을 검지율 및 오검지율 측면에서 비교 검토하였으며, 연구결과 SVM 모형이 기존의 Back- propagation 모형보다 돌발상황검지에 더욱 우수한 것으로 나타났다.
Abstract
Incidents on the freeway disrupt traffic flow and the cost of delay caused by incidents is significant. To reduce the impact of an incident, a traffic management center needs to quickly detect and remove it from the freeway. Quick and efficient automatic incident detection has been a main goal of the transportation research for many years. Also many algorithms based on loop detector data have been developed and tested for the Automatic Incident Detection(AID). However, many of them have a limited success in their overall performance in terms of detection rate, false alarm rate, and the mean time to detect an incident. Until recently, the neural network models have been the one of the popular and efficient approach for real-time automatic incident detection and many researches have shown that the neural network models were much more efficient than various other previous models. The purpose of this research is to propose a more efficient and accurate model than the neural network model in the automatic incident detection problem. For this purpose, a machine learning model, Support Vector Machine (SVM) learning which is based on the statistical learning theory, has been used in this paper. The experiments have been done with real world freeway data, and the results show that the SVM could provide better performance in terms of DR(Detection Rate) and FAR(False Alarm Rate) than Backpropagation which is the most popular neural network model.
참고문헌 (26)
- "Application of time-series analysis techniques to freeway incident detection"
- "A SVM classifier to separate false signals from microcalcifications in digital mammograms Physics in Medicine and Biology", 46(6), pp.1651~1663 (2001)
- "Comparison of view-based object recognition algorithms using realistic 3D models Springer Lecture Notes in Computer Science", 1112, pp.251~256 (1996)
- "Automatic Incident Detection on Motorways", pp.221~227 (1990)
- "Smoothing Algorithms for Incident Detection", pp.8~16 (1993)
- "Support Vector Networks", 20, pp.273~297 (1995)
- "Incident Detection on Urban Freeways", pp.12~24 (1974)
- "Another approach to polychotomous classification Department of Statistics", Stanford University
- "On-Line Testing of the McMaster Incident Detection Algorithm Under Recurrent Congestion", pp.1~7 (1993)
- "A comparison of methods for multi-class support vector machines IEEE Transactions on Neural Networks", 13, pp.415~425 (2002)
- "Text Categorization with Support Vector Machine Learning with Many Relevant Features Proceedings of ECML-98 10th European Conference on Machine Learning", pp.137~142 (1998)
- "Normalization Methods for Input and Output Vectors in Backpropagation Neural Networks International Journal of Computer Mathematics", 71(2), pp.161~171 (1999)
- "Standard and Advanced Backpropagation Models for Image Processing Application in Traffic Engineering", 7(3-4), pp.3~4
- "Prediction Performance of Support Vector Machines on Input Vector Normalization Methods International Journal of Computer Mathematics", 81(5), pp.547~554 (2004)
- "University of Wisconsin" (2001)
- "an Application to Face Detection IEEE Conference on Computer Vision and Pattern Recognition", pp.130~136 (1997)
- "Development and Testing of Operational Incident Detection Algorithms US Department of Transportation" (1997)
- "Large margin DAGs for multiclass classification Advances in Neural Information Processing Systems", 12, pp.547~553 (2000)
- "Simulation of Freeway Incident Detection using Artificial Neural Networks", pp.203~217 (1993)
- "Explorations in the Microstructure of Cognition", pp.1~ (1986), MIT Press
- "Support vector machines for phoneme classification School of Artificial Intelligence University of Edinburgh" (2001)
- "Incorporating Invariances in Support Vector Learning Machines Springer Lecture Notes in Computer Science", 1112, pp.47~52 (1996), in C
- "Incident detection using support vector machines", 11, pp.309~328 (2003), Incident detection using support vector machines
- "The Nature of Statistical Learning Theory" (1995), Springer Verlag
- "Department of Computer Science University of London" (1998)
- "Department of Computer Science University of London" (1998)