Class-Balanced Loss Based on Effective Number of Samples

tensorflow / addons

Useful extra functionality for TensorFlow 2.x maintained by SIG-addons

Apache License 2.0

1.69k stars 613 forks source link

Class-Balanced Loss Based on Effective Number of Samples #2732

Closed ashutosh1919 closed 1 year ago

ashutosh1919 commented 2 years ago

Currently, TensorFlow has Focal loss which can be one of the options to handle the class imbalance issues. But I myself have faced problems in the case of some datasets like RecSys. IMO, TensorFlow should consist of Class-balanced loss (paper).

I am interested in working on this issue. If you think we should add this, then please assign the issue to me and I can raise PR.

cc @bhack

bhack commented 2 years ago

As the focal loss Is going to be duplicated in Keras-cv can you open a ticket in keras-cv so that if they are interested we avoid another potential duplication?

ashutosh1919 commented 2 years ago

Created #16735

seanpmorgan commented 1 year ago

TensorFlow Addons is transitioning to a minimal maintenance and release mode. New features will not be added to this repository. For more information, please see our public messaging on this decision: TensorFlow Addons Wind Down

Please consider sending feature requests / contributions to other repositories in the TF community with a similar charters to TFA: Keras Keras-CV Keras-NLP