Text classification model with online learning

Are there any text classification models out there that allow incremental updates to the model without relearning it from scratch on the whole corpus?

For example, this scenario applies categories (labels) to GitLab issues based on issue text and title.

But when new categories are added/removed, or bad classification is detected, updating the model required downloading the whole issue corpus and training it from scratch. This training can be a quite resource intensive task (and definitely not carbon friendly), so I wonder if there are online learning models there can do incremental updates with new data for the classification task?

I found the list of some online ML resources here GitHub - online-ml/awesome-online-machine-learning: Online machine learning resources but it doesn’t seem to list anything that is related to this specific problem.

Have you seen this ?

1 Like

No. Thanks for the reference. Watching it right now.