d-gcc / CAE-Ensemble

Code for the paper Time Series Outlier Detection with Diversity-Driven Convolutional Ensembles

CAE-Ensemble

Code for the paper Time Series Outlier Detection with Diversity-Driven Convolutional Ensembles

How to run the model:

• Execute cae_ensemble.py specifying the model parameters, for example: python cae_ensemble.py --dataset 81 --diversity_factor 1 --beta_transfer 0.5 --rolling_size 16 --ensemble_members 20 --epochs 200
• The complete list of parameters is available at lines 963--1005 in cae_ensemble.py. The model parameters are the ones detailed in 1.
• The data sets number is related to the specification in lines 1064--1169.
• The structure for the data input is defined in data_provider.py.
• Results will be inserted in a database, calculations and connections are managed in metrics_insert.py.

Baselines:

• Use the same structure as the CAE-Ensemble model, just using their specific parameters.

Citation

If you use the code, please cite the following paper:

  
@article{pvldb/Ca22,
  author    = {David Campos and Tung Kieu and Chenjuan Guo and Feiteng Huang and Kai Zheng and 
               Bin Yang and Christian S. Jensen},
  title     = {{Unsupervised Time Series Outlier Detection with Diversity-Driven Convolutional 
               Ensembles}},
  journal   = {{PVLDB}},
  volume    = {15},
  number    = {3},
  pages     = {611--623},
  year      = {2022}
}