This repository tries to solve the multi-label freesound classification using MixMatch: A Holistic Approach to Semi-Supervised Learning and SpecAugment.
Dataset is using the freesound - 2019 kaggle competion.
- Python 3.6+
- PyTorch 1.1
- torchvision 0.3.0 (older versions are not compatible with this code)
- tensorboardX
- progress
- matplotlib
- numpy
- librosa
Train the model by using the freesound 2019 curated and noisy data.
./train.sh
tensorboard --logdir=./result
0.856 lwlwrap using 5% test set till now.
@article{berthelot2019mixmatch,
title={MixMatch: A Holistic Approach to Semi-Supervised Learning},
author={Berthelot, David and Carlini, Nicholas and Goodfellow, Ian and Papernot, Nicolas and Oliver, Avital and Raffel, Colin},
journal={arXiv preprint arXiv:1905.02249},
year={2019}
}