Skip to content

Information-Fusion-Lab-Umass/Kaggle-Pawpularity

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

pawpularity

Hybrid CNN on Pawpularity Kaggle Contest

Kaggle contest link

Attempt to use the shortfused hybrid conv layer to improve a baseline VGG model with multi-modal inputs

Usage

Repo is located on gypsum at /mnt/nfs/work1/mfiterau/genglinliu/Kaggle-Pawpularity

Use any gpu node, the main script is src/main.py

Notes:

After setting the correct data directory, run main.py using a TitanX gpu on gypsum - this works without error.

For the competition, the performance is not very satisfactory, with a public score of rmse = 22.7 on the baseline. Possible reaons include

  • the outdated VGG model
  • high memory complexity for more structured covariates to be used
  • training from scratch on a dataset that's not large enough might not be better than transfer learning.

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published