A framework for end-to-end deep learning-based anomaly detection in transportation networks. (May 2020)

• Record Type: Journal Article
• Authors: Davis, Neema; Raina, Gaurav; Jagannathan, Krishna
• Is Part Of: Transportation research interdisciplinary perspectives. Volume 5(2020)
• Publication Date: 2020-05
• Subjects: End-to-end anomaly detection -- LSTM -- Extreme value theory; Transportation -- Periodicals; 388.05
• DOI: 10.1016/j.trip.2020.100112
• Languages: English
• ISSNs: 2590-1982

Abstract

Abstract: We develop an end-to-end deep learning-based anomaly detection model for temporal data in transportation networks. The proposed EVT-LSTM model is derived from the popular LSTM (Long Short-Term Memory) network and adopts an objective function that is based on fundamental results from EVT (Extreme Value Theory). We compare the EVT-LSTM model with some established statistical, machine learning, and hybrid deep learning baselines. Experiments on seven diverse real-world data sets demonstrate the superior anomaly detection performance of our proposed model over the other models considered in the comparison study. Highlights: An end-to-end deep learning-based anomaly detection model for temporal data in transportation networks is developed. The proposed EVT-LSTM model is based on the Long Short-Term Memory (LSTM) network, with an objective function derived from Extreme Value Theory (EVT) principles. Our model is compared with established statistical, machine learning, and hybrid deep learning baseline models. The superior anomaly detection performance of the EVT-LSTM model is observed across seven diverse data sets with F1-score as the metric.