Time Series Anomaly Detection Solutions for KDD2021 UCR dataset
https://compete.hexagon-ml.com/practice/competition/39/
- Download ucr data from here, and create a directory named ".data" in this repo directory, unzip the downloaded data to ".data".
- Enter a directory, such as Augment, install package according to "requirements.txt", and execute "python main.py".
| solution name | score | directory | resources |
|---|---|---|---|
| genta | 83.6 | genta | Video |
| Augment (3way) | 73.2 | Augment | 参考Paper |
| DeepSAD | 64.8 | DeepSAD | 参考Code |
| DeepSVDD | 54.4 | DeepSVDD | 参考Code |
| matrix-profile-1 | 52.8 | matrixprofile-1 | 参考Page |
| LOF | 42.8 | LOF | 参考Doc |
| vanilla_ae | 29.2 | ae-torch/ae_conv.py | |
| vanilla_vae | 27.6 | ae-tf/vae.py | 参考Doc |
| Isolation Forest | 23.6 | IF | 参考Doc |