This code is the official PyTorch implementation of our NeurIPS'25 Paper: CrossAD: Time Series Anomaly Detection with Cross-scale Associations and Cross-window Modeling.

If you find this project helpful, please don't forget to give it a ⭐ Star to show your support. Thank you!

Introduction

We propose CrossAD, a novel framework for time series Anomaly Detection that takes Cross-scale associations and Cross-window modeling into account. We propose a cross-scale reconstruction that reconstructs fine-grained series from coarser series, explicitly capturing cross-scale associations. Furthermore, we design a query library and incorporate global multi-scale context to overcome the limitations imposed by fixed window sizes. Extensive experiments conducted on seven real-world datasets using nine evaluation metrics validate the effectiveness of CrossAD, demonstrating state-of-the-art performance in anomaly detection.

Quickstart

Folder structure

.
|-- README.md
|-- configs
|   |-- GECCO
|   |-- MSL
|   |-- PSM
|   |-- SMAP
|   |-- SMD
|   |-- SWAN
|   |-- SWAT
|   |-- TSB-AD-U
|   `-- UCR
|-- data_provider
|   |-- __init__.py
|   |-- data_provider.py
|   `-- data_provider_UCR.py
|-- dataset
|   |-- DETECT_META.csv
|   |-- TSB-AD
|   |-- UCR_Anomaly_FullData
|   `-- data
|-- exp
|   |-- exp_TSBAD.py
|   `-- exp_anomaly_detection.py
|-- figs
|   |-- affiliation.png
|   |-- main.png
|   `-- vus.png
|-- models
|   `-- CrossAD
|-- requirements.txt
|-- run.py
|-- run_TSBAD.py
|-- run_UCR.py
|-- scripts
|   |-- test
|   `-- train
|-- test_results
|   |-- GECCO
|   |-- MSL
|   |-- PSM
|   |-- SMAP
|   |-- SMD
|   |-- SWAN
|   |-- SWAT
|   |-- TSB-AD-U
|   `-- UCR
`-- ts_ad_evaluation
    |-- __init__.py
    |-- accomplish_UCR
    |-- affiliation
    |-- auc_vus
    |-- evaluator.py
    |-- f1
    |-- pate
    `-- spot.py

Installation

pip install -r requirements.txt

Data preparation

Prepare Data. You can obtained the well pre-processed datasets at https://drive.google.com/file/d/1YU_d9kIaP2EubyUhGWOwSKAhJxNntB8W/view?usp=sharing. Then place the downloaded data under the folder ./dataset/.

dataset
|   |-- DETECT_META.csv
|   |-- TSB-AD
|   |-- UCR_Anomaly_FullData
|   `-- data

Train and evaluate model

1. The model structure of CrossAD under the folder ./models/CrossAD/
2. We provide the scripts for CrossAD under the folder ./scripts/. For example you can reproduce a experiment result as the following:

sh ./scripts/test/GECCO.sh

If there is no model weight, please execute the following script:

sh ./scripts/train/GECCO.sh

Results

Results in the seven real-world datasets. The V-R and V-P are the VUS-ROC and VUS-PR, that higher indicate better performance. The best ones are in bold, and the second ones are underlined.

The affiliation metric results for the five real-world datasets are as follows.

Citation

@inproceedings{CrossAD,
  title     = {{CrossAD}: Time Series Anomaly Detection with Cross-scale Associations and Cross-window Modeling},
  author    = {Li, Beibu and Shentu, Qichao and Shu, Yang and Zhang, Hui and Li, Ming and Jin, Ning and Yang, Bin and Guo, Chenjuan},
  booktitle = {NeurIPS},
  year      = {2025}
}

@inproceedings{DADA,
title      = {Towards a General Time Series Anomaly Detector with Adaptive Bottlenecks and Dual Adversarial Decoders},
author     = {Shentu, Qichao and Li, Beibu and Zhao, Kai and Shu, Yang and Rao, Zhongwen and Pan, Lujia and Yang, Bin and Guo, Chenjuan},
booktitle  = {ICLR},
year       = {2025}
}

Community Support | Acknowledgements

This project is built on the shoulders of the open-source community.
Special thanks to the authors and contributors of the following repositories:

• decisionintelligence/TAB – large-scale time-series anomaly benchmark with standardized data, metrics, and reproducible experiments.
• TheDatumOrg/TSB-AD – a comprehensive benchmark and evaluation protocol for time-series anomaly detection.
• thuml/Time-Series-Library – comprehensive and reproducible deep-learning baselines for time-series tasks.