# [8th place solution] : SpecMix and warm-up pipeline

Hello,
I released on a
[github repository](https://github.com/ebouteillon/freesound-audio-tagging-2019)
my solution. It is highly inspired from
[@daisukelab](/daisukelab)
notebooks for preprocessing and
[@mhiro2](/mhiro2)
for his simple CNN model.
Keys points of this solution, 2 techniques I imagined (maybe they exist under another name 😄) 
-
warm-up pipeline
: try to use noisy dataset as pretrain model and as semi-supervised to increase diversity in generating a set of models. In another words, first use noisy set to warmup model training, then fine tune model with curated set and then use noisy set again in a semi-supervised way.
-
SpecMix
: my new data-augmentation technique which takes what I consider the best from SpecAugment and mixup. It generates new samples by applying frequency replacement and time replacement on inputs and compute a weighted average on targets. More details in the github
[README](https://github.com/ebouteillon/freesound-audio-tagging-2019)
Inference is done using an ensemble of 2 models (
[@mhiro2](/mhiro2)
simple CNN model + VGG16) on 10 folds CV.
More details are available in the github
[README](https://github.com/ebouteillon/freesound-audio-tagging-2019)
, as well code and weights.