KAN-AD: Time Series Anomaly Detection with Kolmogorov–Arnold Networks

Quan Zhou, Changhua Pei, Fei Sun, Han Jing, Zhengwei Gao, Haiming Zhang, Gaogang Xie, Dan Pei, Jianhui Li
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:79136-79149, 2025.

Abstract

Time series anomaly detection (TSAD) underpins real-time monitoring in cloud services and web systems, allowing rapid identification of anomalies to prevent costly failures. Most TSAD methods driven by forecasting models tend to overfit by emphasizing minor fluctuations. Our analysis reveals that effective TSAD should focus on modeling "normal" behavior through smooth local patterns. To achieve this, we reformulate time series modeling as approximating the series with smooth univariate functions. The local smoothness of each univariate function ensures that the fitted time series remains resilient against local disturbances. However, a direct KAN implementation proves susceptible to these disturbances due to the inherently localized characteristics of B-spline functions. We thus propose KAN-AD, replacing B-splines with truncated Fourier expansions and introducing a novel lightweight learning mechanism that emphasizes global patterns while staying robust to local disturbances. On four popular TSAD benchmarks, KAN-AD achieves an average 15% improvement in detection accuracy (with peaks exceeding 27%) over state-of-the-art baselines. Remarkably, it requires fewer than 1,000 trainable parameters, resulting in a 50% faster inference speed compared to the original KAN, demonstrating the approach’s efficiency and practical viability.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-zhou25u, title = {{KAN}-{AD}: Time Series Anomaly Detection with Kolmogorov–Arnold Networks}, author = {Zhou, Quan and Pei, Changhua and Sun, Fei and Jing, Han and Gao, Zhengwei and Zhang, Haiming and Xie, Gaogang and Pei, Dan and Li, Jianhui}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {79136--79149}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/zhou25u/zhou25u.pdf}, url = {https://proceedings.mlr.press/v267/zhou25u.html}, abstract = {Time series anomaly detection (TSAD) underpins real-time monitoring in cloud services and web systems, allowing rapid identification of anomalies to prevent costly failures. Most TSAD methods driven by forecasting models tend to overfit by emphasizing minor fluctuations. Our analysis reveals that effective TSAD should focus on modeling "normal" behavior through smooth local patterns. To achieve this, we reformulate time series modeling as approximating the series with smooth univariate functions. The local smoothness of each univariate function ensures that the fitted time series remains resilient against local disturbances. However, a direct KAN implementation proves susceptible to these disturbances due to the inherently localized characteristics of B-spline functions. We thus propose KAN-AD, replacing B-splines with truncated Fourier expansions and introducing a novel lightweight learning mechanism that emphasizes global patterns while staying robust to local disturbances. On four popular TSAD benchmarks, KAN-AD achieves an average 15% improvement in detection accuracy (with peaks exceeding 27%) over state-of-the-art baselines. Remarkably, it requires fewer than 1,000 trainable parameters, resulting in a 50% faster inference speed compared to the original KAN, demonstrating the approach’s efficiency and practical viability.} }
Endnote
%0 Conference Paper %T KAN-AD: Time Series Anomaly Detection with Kolmogorov–Arnold Networks %A Quan Zhou %A Changhua Pei %A Fei Sun %A Han Jing %A Zhengwei Gao %A Haiming Zhang %A Gaogang Xie %A Dan Pei %A Jianhui Li %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-zhou25u %I PMLR %P 79136--79149 %U https://proceedings.mlr.press/v267/zhou25u.html %V 267 %X Time series anomaly detection (TSAD) underpins real-time monitoring in cloud services and web systems, allowing rapid identification of anomalies to prevent costly failures. Most TSAD methods driven by forecasting models tend to overfit by emphasizing minor fluctuations. Our analysis reveals that effective TSAD should focus on modeling "normal" behavior through smooth local patterns. To achieve this, we reformulate time series modeling as approximating the series with smooth univariate functions. The local smoothness of each univariate function ensures that the fitted time series remains resilient against local disturbances. However, a direct KAN implementation proves susceptible to these disturbances due to the inherently localized characteristics of B-spline functions. We thus propose KAN-AD, replacing B-splines with truncated Fourier expansions and introducing a novel lightweight learning mechanism that emphasizes global patterns while staying robust to local disturbances. On four popular TSAD benchmarks, KAN-AD achieves an average 15% improvement in detection accuracy (with peaks exceeding 27%) over state-of-the-art baselines. Remarkably, it requires fewer than 1,000 trainable parameters, resulting in a 50% faster inference speed compared to the original KAN, demonstrating the approach’s efficiency and practical viability.
APA
Zhou, Q., Pei, C., Sun, F., Jing, H., Gao, Z., Zhang, H., Xie, G., Pei, D. & Li, J.. (2025). KAN-AD: Time Series Anomaly Detection with Kolmogorov–Arnold Networks. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:79136-79149 Available from https://proceedings.mlr.press/v267/zhou25u.html.

Related Material