Decompose Auto-Transformer Time Series Anomaly Detection for Network Management
Time series anomaly detection through unsupervised methods has been an active research area in recent years due to its enormous potential for networks management. The representation and reconstruction of time series have made extraordinary progress in existing works. However, time series is known to...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-01-01
|
| Series: | Electronics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2079-9292/12/2/354 |
| _version_ | 1797443480297930752 |
|---|---|
| author | Bo Wu Chao Fang Zhenjie Yao Yanhui Tu Yixin Chen |
| author_facet | Bo Wu Chao Fang Zhenjie Yao Yanhui Tu Yixin Chen |
| author_sort | Bo Wu |
| collection | DOAJ |
| description | Time series anomaly detection through unsupervised methods has been an active research area in recent years due to its enormous potential for networks management. The representation and reconstruction of time series have made extraordinary progress in existing works. However, time series is known to be complex in terms of their temporal dependency and stochasticity, which makes anomaly detection difficult. To this end, we propose a novel approach based on a decomposition auto-transformer networks(DATN) for time series anomaly detection. The time series is decomposed into seasonal and trend components, and renovated as a basic inner block deep model. With this design, transformers can decompose complex time series in a progressive manner. We also design an auto-transfomer block that determines dependencies and representation aggregation at the sub-series level based on series seasonal and trend components. Moreover, the complex transformer decoder is replaced by a simple linear decoder, which makes the model more efficient. Extensive experiments on various public benchmarks demonstrate that our method has achieved state-of-the-art performance. |
| first_indexed | 2024-03-09T12:56:39Z |
| format | Article |
| id | doaj.art-2aa1b2d77bcc4d6f9a7852b94e94541e |
| institution | Directory Open Access Journal |
| issn | 2079-9292 |
| language | English |
| last_indexed | 2024-03-09T12:56:39Z |
| publishDate | 2023-01-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Electronics |
| spelling | doaj.art-2aa1b2d77bcc4d6f9a7852b94e94541e2023-11-30T21:59:18ZengMDPI AGElectronics2079-92922023-01-0112235410.3390/electronics12020354Decompose Auto-Transformer Time Series Anomaly Detection for Network ManagementBo Wu0Chao Fang1Zhenjie Yao2Yanhui Tu3Yixin Chen4School of Software, Nanchang Hangkong University, Nanchang 330063, ChinaPurple Mountain Laboratories, Nanjing 211111, ChinaPurple Mountain Laboratories, Nanjing 211111, ChinaPurple Mountain Laboratories, Nanjing 211111, ChinaPurple Mountain Laboratories, Nanjing 211111, ChinaTime series anomaly detection through unsupervised methods has been an active research area in recent years due to its enormous potential for networks management. The representation and reconstruction of time series have made extraordinary progress in existing works. However, time series is known to be complex in terms of their temporal dependency and stochasticity, which makes anomaly detection difficult. To this end, we propose a novel approach based on a decomposition auto-transformer networks(DATN) for time series anomaly detection. The time series is decomposed into seasonal and trend components, and renovated as a basic inner block deep model. With this design, transformers can decompose complex time series in a progressive manner. We also design an auto-transfomer block that determines dependencies and representation aggregation at the sub-series level based on series seasonal and trend components. Moreover, the complex transformer decoder is replaced by a simple linear decoder, which makes the model more efficient. Extensive experiments on various public benchmarks demonstrate that our method has achieved state-of-the-art performance.https://www.mdpi.com/2079-9292/12/2/354networks managementtime seriesanomaly detectionseries decomposetransformer |
| spellingShingle | Bo Wu Chao Fang Zhenjie Yao Yanhui Tu Yixin Chen Decompose Auto-Transformer Time Series Anomaly Detection for Network Management Electronics networks management time series anomaly detection series decompose transformer |
| title | Decompose Auto-Transformer Time Series Anomaly Detection for Network Management |
| title_full | Decompose Auto-Transformer Time Series Anomaly Detection for Network Management |
| title_fullStr | Decompose Auto-Transformer Time Series Anomaly Detection for Network Management |
| title_full_unstemmed | Decompose Auto-Transformer Time Series Anomaly Detection for Network Management |
| title_short | Decompose Auto-Transformer Time Series Anomaly Detection for Network Management |
| title_sort | decompose auto transformer time series anomaly detection for network management |
| topic | networks management time series anomaly detection series decompose transformer |
| url | https://www.mdpi.com/2079-9292/12/2/354 |
| work_keys_str_mv | AT bowu decomposeautotransformertimeseriesanomalydetectionfornetworkmanagement AT chaofang decomposeautotransformertimeseriesanomalydetectionfornetworkmanagement AT zhenjieyao decomposeautotransformertimeseriesanomalydetectionfornetworkmanagement AT yanhuitu decomposeautotransformertimeseriesanomalydetectionfornetworkmanagement AT yixinchen decomposeautotransformertimeseriesanomalydetectionfornetworkmanagement |