AAPSO

Deep Feature Selection for Pneumonia Detection

"An adaptive and altruistic PSO-based deep feature selection method for Pneumonia detection from Chest X-rays" published in Applied Soft Computing, Elsevier

Access the journal article: Click Here

Preprint (Arxiv): Abstract & PDF

@article{pramanik2022adaptive,
  title={An adaptive and altruistic PSO-based deep feature selection method for Pneumonia detection from Chest X-rays},
  author={Pramanik, Rishav and Sarkar, Sourodip and Sarkar, Ram},
  journal={Applied Soft Computing},
  volume = {128},
  pages={109464},
  year={2022},
  issn = {1568-4946},
  doi={10.1016/j.asoc.2022.109464},
  url={https://www.sciencedirect.com/science/article/pii/S1568494622005737},
  publisher={Elsevier}
}

Dataset Links

1. Paul Monney's Pneumonia
2. UCI datasets: In this repository
3. Microarray Datasets: In this repository
4. COVID-19 prediction: In this repository

Instructions to run the code:

Required directory structure:

(Note: train and val contains subfolders representing classes in the dataset.)

+-- data
|   +-- .
|   +-- train
|   +-- val
+-- AAPSO.py
+-- main.py

1. Download the repository and install the required packages:

pip3 install -r requirements.txt

2. The main file is sufficient to run the experiments. Then, run the code using linux terminal as follows:

python3 main.py --data_directory "data"

Available arguments:

• --epochs: Number of epochs of training. Default = 10
• --learning_rate: Learning Rate. Default = 0.0001
• --batch_size: Batch Size. Default = 32

3. Please don't forget to edit the above parameters before you start