Non-Contrastive Self-Supervised Learning with UNO Process for Respiratory Sound Analysis

This repository is the official implementation of Non-Contrastive Self-Supervised Learning with Unpredictable Neuron Operation (UNO) Process for Respiratory Sound Analysis.

Conda Environment

0. Install conda environment:

conda env create -f env.yml

1. Activate the environment:

conda activate uno_pl

Dataset

Please refer to icbhi_data/ for the data preparation and preprocessing.

Acknowledgements

1. AST model: https://github.com/YuanGongND/ast/blob/master/src/models/ast_models.py
2. ResNet model: https://github.com/hche11/VGGSound/blob/master/models/resnet.py
3. Non-Constrastive Self-Supervised Learning: https://github.com/lucidrains/byol-pytorch