I am trying to train a model that should automatically classify into specific categories some free text, but I have a low volume of data to start with. I have tried to give more context adding an header to each text piece and it worked, but now I am stuck again.
Do you have any suggestion to overcome the issue of a small training sample on AI Builder?
Thanks.
Unfortunatelly, training data is required to get a good quality model.
What you can do is to build a model with the data you have so far.
Then put your model in production with a manual review process.
Then use the user's edits, when the model didn't predicted properly, to increase you training set size and train a new version that you will deploy and iterate again on same principles.