Detecting anomalies in high-dimensional time series data from transportation systems is challenging due to complex multivariate distributions and temporal dependencies. This work is innovative because it combines normalizing flows with sequence models to tractably perform density estimation in high-dimensional spaces. The probabilistic approach provides interpretable anomaly scores and enables diagnosis of anomaly causes.
This paper proposes a generative anomaly detection framework for time series data using normalizing flows and LSTM encoder-decoder models. The approach performs multi-variate anomaly detection through conditional density estimation on time series data, enabling both detection of anomalies and diagnosis of their causes. The method is demonstrated on traffic network data for detecting and characterizing traffic incidents.
The generative anomaly detection approach was evaluated on traffic data from Nashville and demonstrated superior performance in detecting anomalies at both timestep and segment levels. The method successfully identified anomalies in multiple road segments simultaneously and provided interpretable analysis of anomaly causes. The conditional density estimation approach showed better sensitivity and precision compared to alternative methods.
@inproceedings{kang2022generative,
author = {Kang, Zhuangwei and Mukhopadhyay, Ayan and Gokhale, Aniruddha and Wen, Shijie and Dubey, Abhishek},
booktitle = {2022 IEEE 25th International Conference on Intelligent Transportation Systems (ITSC)},
title = {Traffic Anomaly Detection Via Conditional Normalizing Flow},
year = {2022},
pages = {2563-2570},
abstract = {Traffic congestion anomaly detection is of paramount importance in intelligent traffic systems. The goals of transportation agencies are two-fold: to monitor the general traffic conditions in the area of interest and to locate road segments under abnormal congestion states. Modeling congestion patterns can achieve these goals for citywide roadways, which amounts to learning the distribution of multivariate time series (MTS). However, existing works are either not scalable or unable to capture the spatial-temporal information in MTS simultaneously. To this end, we propose a principled and comprehensive framework consisting of a data-driven generative approach that can perform tractable density estimation for detecting traffic anomalies. Our approach first clusters segments in the feature space and then uses conditional normalizing flow to identify anomalous temporal snapshots at the cluster level in an unsupervised setting. Then, we identify anomalies at the segment level by using a kernel density estimator on the anomalous cluster. Extensive experiments on synthetic datasets show that our approach significantly outperforms several state-of-the-art congestion anomaly detection and diagnosis methods in terms of Recall and F1-Score. We also use the generative model to sample labeled data, which can train classifiers in a supervised setting, alleviating the lack of labeled data for anomaly detection in sparse settings.},
contribution = {lead},
doi = {10.1109/ITSC55140.2022.9922061},
keywords = {anomaly detection, generative models, normalizing flows, time series, traffic networks, LSTM}
}