mlsanityData to 14 Oct 2026

A Comparative Study of Adaptation Strategies for Time Series Foundation Models in Anomaly Detection

arXiv 2601.00446 ↗First posted 1 Jan 2026cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

Time series anomaly detection is essential for the reliable operation of complex systems, but most existing methods require extensive task-specific training. We explore whether time series foundation models (TSFMs), pretrained on large heterogeneous data, can serve as universal backbones for anomaly detection. Through systematic experiments across multiple benchmarks, we compare zero-shot inference, full model adaptation, and parameter-efficient fine-tuning (PEFT) strategies. Our results demonstrate that TSFMs outperform task-specific baselines, achieving notable gains in AUC-PR and VUS-PR, particularly under severe class imbalance. Moreover, PEFT methods such as LoRA, OFT, and HRA not only reduce computational cost but also match or surpass full fine-tuning in most cases, indicating that TSFMs can be efficiently adapted for anomaly detection, even when pretrained for forecasting. These findings position TSFMs as promising general-purpose models for scalable and efficient time series anomaly detection.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 26 of its 39 references that are in the map. First posted on arXiv 1 Jan 2026.

Builds on

26 of 39 references.

References · 26

  1. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,599
  2. 2022TimesNet: Temporal 2D-Variation Modeling for General Time Series Analysiscited by 1,058
  3. 2022Few-Shot Parameter-Efficient Fine-Tuning is Better and Cheaper than In-Context Learningcited by 896
  4. 2021Towards a Unified View of Parameter-Efficient Transfer Learningcited by 872
  5. 2024Chronos: Learning the Language of Time Seriescited by 759
  6. 2024Unified Training of Universal Time Series Forecasting Transformerscited by 614
  7. 2024MOMENT: A Family of Open Time-series Foundation Modelscited by 482
  8. 2021TS2Vec: Towards Universal Representation of Time Seriescited by 471
  9. 2018Detecting Spacecraft Anomalies Using LSTMs and Nonparametric Dynamic Thresholdingcited by 410
  10. 2021Anomaly Transformer: Time Series Anomaly Detection with Association Discrepancycited by 394
  11. 2021Graph Neural Network-Based Anomaly Detection in Multivariate Time Seriescited by 301
  12. 2024Time-MoE: Billion-Scale Time Series Foundation Models with Mixture of Expertscited by 287
14 more
  1. 2022TranAD: Deep Transformer Networks for Anomaly Detection in Multivariate Time Series Datacited by 214
  2. 2017A Multimodal Anomaly Detector for Robot-Assisted Feeding Using an LSTM-based Variational Autoencodercited by 209
  3. 2023Controlling Text-to-Image Diffusion by Orthogonal Finetuningcited by 208
  4. 2019MAD-GAN: Multivariate Anomaly Detection for Time Series Data with Generative Adversarial Networkscited by 176
  5. 2020Multivariate Time-series Anomaly Detection via Graph Attention Networkcited by 167
  6. 2023TimeGPT-1cited by 158
  7. 2021Learning Graph Structures with Transformer for Multivariate Time Series Anomaly Detection in IoTcited by 104
  8. 2024Moirai-MoE: Empowering Time Series Foundation Models with Sparse Mixture of Expertscited by 95
  9. 2024Position: What Can Large Language Models Tell Us about Time Series Analysiscited by 47
  10. 2017Unsupervised and Semi-supervised Anomaly Detection with LSTM Neural Networkscited by 40
  11. 2024TOTEM: TOkenized Time Series EMbeddings for General Time Series Analysiscited by 39
  12. 2024Bridging The Gap between Low-rank and Orthogonal Adaptation via Householder Reflection Adaptationcited by 15
  13. 2025VUS: Effective and Efficient Accuracy Measures for Time-Series Anomaly Detectioncited by 13
  14. 2024Time Series Foundational Models: Their Role in Anomaly Detection and Predictioncited by 4

Built on by

Not cited yet.