Additional categories

In the lab, i attempted to add additional categories (synonyms such as ‘infection’, ‘nodule’) to the list of categories. However, this made no difference to the outcome. What is the proper way of doing this?

Hi @Isaac_Lam

Good to see you on the forum again.

I may not be able to specifically tell you wrt to this lab, since I am not sure exactly how you went about this, particulary relabeling the existing dataset or expanding it.

In general if you change the classes but maintain the dataset size, you will at least need to retrain the model.

(For instance in a neural network model, you may have to update (remove and re-add) the classification layer, potentially freezing the weights and then retrain the model. )

And if you have expanded the dataset i.e. added more data, you may have to do exploratory data analysis and preprocess the dataset again and then retrain the model.

Hope this helps, some.

Best Regards.

– Jaidev