UATR-CMoE

Underwater acoustic target recognition -- Convolution-based Mixture of Experts

This is the PyTorch implementation of the paper:
"Unraveling Complex Data Diversity in Underwater Acoustic Target Recognition through Convolution-based Mixture of Experts",
which has been published on Expert Systems with Applications.

DOI: https://doi.org/10.1016/j.eswa.2024.123431
Arxiv: https://arxiv.org/abs/2402.11919

In addition to the model architecture (cmoe_model.py), this repository offers pre-extracted features of the Shipsear test set, accompanied by corresponding testing code.

Steps of Inference

git clone https://github.com/xy980523/UATR-CMoE.git    
cd UATR-CMoE
pip install -r requirements.txt

1. Download pre-extracted features and pre-trained checkpoint

Download link:
Pre-extracted features: https://github.com/xy980523/UATR-CMoE/releases/download/features/features.zip
Pre-trained checkpoint: https://github.com/xy980523/UATR-CMoE/releases/download/checkpoint/best_model.ckpt

Save features to your own path (/path_features):

mkdir -p /path_features
unzip features.zip -d /path_features

And make sure ``best_model.ckpt'' is in the UATR-CMoE folder.

2. Load models and print results

python test.py /path_features

It will produce the accuracy and confusion matrix on the Shipsear test set.

3. (Optional) Reproduce the confusion matrix

python draw_confusion.py

It will produce the confusion matrix heat map (see Fig. 5 in the paper).

Citation

@article{xie2024unraveling,
  title={Unraveling complex data diversity in underwater acoustic target recognition through convolution-based mixture of experts},
  author={Xie, Yuan and Ren, Jiawei and Xu, Ji},
  journal={Expert Systems with Applications},
  pages={123431},
  year={2024},
  publisher={Elsevier}
}