CoAD

🌉 Bridging Classification and Reconstruction: Cooperative Time Series Anomaly Detection

Table of Contents

• 🏄‍ Overall Framework
• 📄 Main Results
• 📊 Case Studies
• ⚙️ Setup
• 🗄️ Prepare datasets
• 🔁 Reproduce the main results
• 🔁 Reproduce the baseline results
• 🔁 Reproduce the ablation study results
• 🔁 Reproduce the parameter study results
• 📚 References

🏄‍ Overall Framework

📄 Main Results

Evaluation results on reliable datasets (KDD21 [1] and TSB-AD [2]) using rigorous evaluation protocols [2].

📊 Case Studies

Visualizes the detection results of COAD on several challenging cases.

⚙️ Setup

Installation

conda create -n CoAD python=3.11
conda activate CoAD
pip install -r requirements.txt

🗄️ Prepare datasets

• A visual illustration of the flaws present in the datasets (SMD, PSM, SWaT, SMAP, MSL, and NAB) can be found at the following anonymous link: flawed datasets. The PDF is sourced from a presentation given at ITISE 2023.

• Download the dataset from the anonymous link dataset and extract it to the dataset folder.

• The 64 subsets of TSB-AD used in the paper can be found in the runners/run_all_CoAD.sh file, including MGAB, SED, SVDB, IOPS and TODS benchmark datasets.

├─dataset
├───TSB-AD
│   ├───raw
│   │   ├───001_NAB_id_1_Facility_tr_1007_1st_2014.csv
│   │   ├───...
├───UCR(KDD21)
│   ├───processed
│   │   ├───train
│   │   ├───test
│   │   ├───label
│   ├───all_period.csv

🔁 Reproduce the main results

sh runners/run_all_CoAD.sh

🔁 Reproduce the baseline results

# reproduce the deep learning based methods
sh runners/run_deep_baseline.sh
# reproduce the data mining based methods
sh runners/run_dm_baseline.sh

🔁 Reproduce the ablation study results

The details of the ablation versions are shown in the runners/run_ablation_CoAD.sh file.

sh runners/run_ablation_CoAD.sh

🔁 Reproduce the parameter study results

sh runners/parameter.sh

📚 References

1. “Multidataset time series anomaly detection competition,” 2021, https://compete.hexagon-ml.com/practice/competition/39/.

2. “The elephant in the room: Towards a reliable time-series anomaly detection benchmark,” in The 38th Conference on Neural Information Processing Systems Datasets and Benchmarks Track, 2024