EDAD

This repository is the implementation of: An Encode-then-Decompose Approach for Unsupervised Time Series Anomaly Detection. We propose the EDAD framework for unsupervised anomaly detection and evaluate its performance on nine open-source datasets.

Get Started

1. Install Python 3.10, PyTorch >= 2.0.0, Wandb. Then run the following command.

   pip install -r requirements.txt

2. Before running EDAD, download the publicly available dataset from the link, unzip it and place it in the /dataset directory.

3. Use the following example to run the algorithm.

   python main-all.py --lr 0.0005 \
   --input_c 25 \
   --output_c 25 \
   --dataset PSM \
   --win_size 100 \
   --d_model 512 \
   --critic sep \
   --batch_size 256 \
   --l_intra_s 1 \
   --l_intra_r 1 \
   --l_mi 1 

4. In order to run EDAD on other data sets, you need to prepare the data set and put it in the /dataset directory, and add the read operation of the dataset in the dataloder.