aladdin-bigan-anomaly-detection's Introduction

ALADDIN-BiGAN-anomaly-detection

This directory contains the files used for the development of "Unsupervised Anomaly Detection for Underwater Gliders Using Generative Adversarial Networks".

images: anomaly detection using BiGAN for underwater gliders: (a) training using normal data and (b) testing using unseen deployment data,
results: the anomaly detection results of the test deployments presented in the paper,
saved_model: pre-trained neural networks,
anomaly_multi_sensitivity.py: main script to train and test the anomaly detection systems, including a sensitivity study detailed in the paper,
data_processing_sensitivity.py: processes the raw datasets, generating datasets for anomaly detection system training, validation, test and sensitivity study,
model.py: builds the neural networks.
utilities.py: auxiliary functions for training and testing

The datasets are mostly collected from BODC's Glider inventory.

Please cite the paper as below in any resulting publications.

@article{wu2021anomaly,
  title={Unsupervised Anomaly Detection for Underwater Gliders Using Generative Adversarial Networks},
  author={Wu, P. and Harris, C.A. and Salavasidis, G. and Lorenzo-Lopez, A. and Kamarudzaman, I. and Philips, A.B. and Thomas, G. and Anderlini, E.},
  journal={Engineering Applications of Artificial Intelligence},
  volume={104},
  pages={104379},
  year={2021},
  publisher={Elsevier}
}

Recommend Projects