15th place solution for g2net_gravitational_wave_detection at kaggle.

• For the details of the competition, please check this page -> G2Net Gravitational Wave Detection.

Overview

• Whitening: Using average PSD. Averaging over all noise samples for each site.

• CQT Scaling with filter_scale = 8/bins_per_octave and (fmin, fmax)=(20, 1024). Both abs and angle part were used.

• Augmentation

  • Horizontal/time shift
    • Pad both side and then horizontal random crop to get time shift image. -> ROC +0.002.
  • Mixup, prevent from overfitting

• GeM Fixed power 3 was better than the trainable case. -> ROC +0.001

• Scores

net	spec	height	width	PB score
effnet b0	Log STFT	256	513	0.8760
effnet b0	CQT	181	513	0.8768
effnet b3	CQT	181	1024	0.8797
effnet b3	CQT	273	1024	0.8802

My final score is ensemble of Log STFT/CQT models.

How to run

environment

• Ubuntu 18.04
• Python with Anaconda/Mamba
• NVIDIA GPUx1

Data Preparation

First, download the data, here, and then place it like below.

../input/
    └ g2net-gravitational-wave-detection/

Outputs will be stored under ../working/ through hydra.

Code&Package Installation

# clone project
$PROJECT=kaggle_g2net_gravitational_wave_detection
git clone https://github.com/Fkaneko/$PROJECT

# install project
cd $PROJECT
conda create -n g2_net python==3.8.10
bash install.sh

• This code was for the competition, so some parts of code are not so clean or clear. Please be careful.

Training & Testing

Simply run followings

python train.py

Please check the src/config/config.yaml for the default training configuration. After training, testing will be automatically started with the best validation score checkpoint.

License

Code

Apache 2.0

Dataset

Please check the kaggle page -> https://www.kaggle.com/c/g2net-gravitational-wave-detection/rules