FreeSound 2019 MixMatch + SpecAugment + Noisy

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.

Requirements

• Python 3.6+
• PyTorch 1.1
• torchvision 0.3.0 (older versions are not compatible with this code)
• tensorboardX
• progress
• matplotlib
• numpy
• librosa

Usage

Train

Train the model by using the freesound 2019 curated and noisy data.

./train.sh

Monitoring training progress

tensorboard --logdir=./result

Results (Accuracy)

0.856 lwlwrap using 5% test set till now.

References

@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}
}