When creating a training dataset, what is the recommended number of samples for the Classification fields?
According to the UiPath documentation, the recommended number of samples for the classification fields depends on the number of document types and layouts that you want to classify. The more document types and layouts you have, the more samples you need to cover the diversity of your data.However, a general guideline is to have at least 20-50 document samples from each class, as this would provide enough data for the classifiers to learn from12.A large number of samples per layout is not mandatory, as the classifiers can generalize from other layouts as well3.
References:1: Document Classification Training Overview2: Document Classification Training Related Activities3: Training High Performing Models
Limited Time Offer
25%
Off
Currently there are no comments in this discussion, be the first to comment!
Currently there are no comments in this discussion, be the first to comment!