Learning Robust Deep State Space for Unsupervised Anomaly Detection in Contaminated Time-Series
The DOI matches its Crossref record; 10 of the 10 references it cites were found in a registry or an index.
Abstract
Anomalies are ubiquitous in real-world time-series data which call for effective and timely detection, especially in an unsupervised setting for labeling cost saving. In this paper, we develop an unsupervised density reconstruction model for multi-dimensional time-series anomaly detection. In particular, it directly handles an important realistic setting that the detection is achieved towards raw time-series contaminated with noise for training, in contrast to most existing anomaly detection works that assume the training data is in general clean i.e. not contaminated with anomaly. It extends recent advancements in deep generative models and state space models to achieve robust anomaly detection. Our approach comprises of a novel state space based generative model, a filtering based inference model, together with a carefully-designated emission model based on robust statistics theory. Extensive experimental results are conducted to show that our approach can adapt to complex patterns even given severely contaminated training data. We also develop visualization techniques to help better understand the behavior of the anomaly detection models. Empirical results show that our method outperforms state-of-the-arts on both synthetic and real-world datasets.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Adam: A Method for Stochastic Optimization
Kingma, D. P., & Ba, J. (2014) · arXiv · doi:10.48550/arxiv.1412.6980 · Cited by 98,946
Saito, T., & Rehmsmeier, M. (2015) · PLOS ONE · doi:10.1371/journal.pone.0118432 · Cited by 5,309
An Introduction to Variational Autoencoders
Kingma, D. P., & Welling, M. (2019) · Foundations and Trends® in Machine Learning · doi:10.1561/2200000056 · Cited by 2,609 · cited 1×
“In this regard, the most common models are Variational Auto-Encoders (VAEs) [22], [23] and Generative Adversarial Networks (GANs) [24], also known as deep generative models.”
MethodRobust Anomaly Detection for Multivariate Time Series through Stochastic Recurrent Neural Network
Su, Y., Zhao, Y., Niu, C., Liu, R., Sun, W., & Pei, D. (2019) · Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining · doi:10.1145/3292500.3330672 · Cited by 1,804 · cited 5×
“Models such as Omnianomaly [10] and USAD [11], although assuming clean training data, can still be found experimentally to remain robust in noisy training data due to the implicit robustness contained in the probabilistic models.”
BackgroundMethodInfluentialDetecting Spacecraft Anomalies Using LSTMs and Nonparametric Dynamic Thresholding
Hundman, K., Constantinou, V., Laporte, C., Colwell, I., & Soderstrom, T. (2018) · Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining · doi:10.1145/3219819.3219845 · Cited by 1,771 · cited 1×
“The Soil Moisture Active Passive satellite (SMAP) and Mars Science Laboratory rover (MSL) are two datasets published by NASA [69], which is called Spacecraft Telemetry as a whole.”
MethodUSAD: UnSupervised Anomaly Detection on Multivariate Time Series
Audibert, J., Michiardi, P., Guyard, F., Marti, S., & Zuluaga, M. A. (2020) · Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining · doi:10.1145/3394486.3403392 · Cited by 1,094 · cited 5×
“Models such as Omnianomaly [10] and USAD [11], although assuming clean training data, can still be found experimentally to remain robust in noisy training data due to the implicit robustness contained in the probabilistic models.”
BackgroundMethodInfluentialLSTM-based Encoder-Decoder for Multi-sensor Anomaly Detection
Malhotra, P., Ramakrishnan, A., Anand, G., Vig, L., Agarwal, P., & Shroff, G. (2016) · arXiv · doi:10.48550/arxiv.1607.00148 · Cited by 819 · cited 9×
“To generate multiple training samples f xx i 1: T g Ni ¼ 1 , we follow the protocol used in Donut [9] and LSTM-ED [4], we use the shingling [63] technique to convert a long multi-variate time-series into many short time-series chunks.”
BackgroundMethodInfluentialImportance Weighted Autoencoders
Burda, Y., Grosse, R., & Salakhutdinov, R. (2015) · arXiv · doi:10.48550/arxiv.1509.00519 · Cited by 644
Deep Structured Energy Based Models for Anomaly Detection
Zhai, S., Cheng, Y., Lu, W., & Zhang, Z. (2016) · arXiv · doi:10.48550/arxiv.1605.07717 · Cited by 334 · cited 3×
“In particular, a majority line of efforts e.g., LSTM-based models [2], [3], LSTM-based autoencoders [4], DSEBMs [5], Convolutional and recurrent encoder-decoders (MSCRED) [6] have been recently put to the setting when the training data is assumed totally clean 1 [2], [4], [5], [7], or the…”
BackgroundKrishnan, R. G., Shalit, U., & Sontag, D. (2015) · arXiv · doi:10.48550/arxiv.1511.05121 · Cited by 206 · cited 1×
“Deep Kalman Filters (DKFs) [50], [51] further allow exogenous input in the state space models, and they incorporate a regularizer to penalize state transitions between time steps.”
Background
Cited by
Deep Learning for Time Series Anomaly Detection: A Survey
Zamanzadeh Darban, Z., Webb, G. I., Pan, S., Aggarwal, C., & Salehi, M. (2025) · ACM Computing Surveys · doi:10.1145/3691338 · Cited by 613
DCdetector: Dual Attention Contrastive Representation Learning for Time Series Anomaly Detection
Yang, Y., Zhang, C., Zhou, T., Wen, Q., & Sun, L. (2023) · Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining · doi:10.1145/3580305.3599295 · Cited by 387
Self-citationNavigating the metric maze: a taxonomy of evaluation metrics for anomaly detection in time series
Sørbø, S., & Ruocco, M. (2024) · Data Mining and Knowledge Discovery · doi:10.1007/s10618-023-00988-8 · Cited by 58
“14, where the event-based metrics prioritize the cluster of three events over the single long one.”
DiffLoad: Uncertainty Quantification in Electrical Load Forecasting With the Diffusion Model
Wang, Z., Wen, Q., Zhang, C., Sun, L., & Wang, Y. (2025) · IEEE Transactions on Power Systems · doi:10.1109/tpwrs.2024.3449032 · Cited by 34
“Restrictions apply. which is not heavy-tailed, we suggest using the heavy-tailed Cauchy distribution to make the model more robust to outliers and mutation according to robust statistics [27], [35].”
Self-citation
Metadata
- Authors
- Li, L., Yan, J., Wen, Q., Jin, Y., & Yang, X.
- Venue
- IEEE Transactions on Knowledge and Data Engineering
- Published
- 2022
- Pages
- 1-1
- ISSN
- 1041-4347 (ISSN-L)
- 1558-2191
- 2326-3865
- OpenAlex
- W4285109203
- Cited by
- 55
- References
- 10
- FWCI
- 4.12
- Type
- journal article
- Language
- en
- Open access
- closed
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.