Why is my custom AI model categorizing some of my documents incorrectly?

Review the training progress and overall accuracy score of your document categories to know which category needs additional training.

If you have more than one category in one of your document Workflows and find that your custom AI model has categorized a document incorrectly: 

  1. Check the Training Progress and Overall Accuracy score of the categorization Training module that belongs to the Workflow in which the categorization error is occurring by clicking the Menu button IDP Menu Button on the top right corner, then clicking Edit Workflow.

    Workflow - Overview@2x
    Then click the Categorize step of your workflow setup and manually add more documents to categories that still show a Not Fully Trained status or a Low Overall Accuracy score. Doing this will improve your A.I. model's accuracy score and auto-categorization feature.

    Categorize-modify
  2. Make sure that you have manually uploaded your training documents to their correct categories while setting up the Workflow in question. Do this by clicking the Vertical Elipsismore_vertbutton of any category you want to check, then click Modify. If you have accidentally uploaded a document from, e.g. Vendor A in the category of Vendor B, delete the file and add it to the correct category. Consistency is critical when training your custom A.I. model.


Find steps to improve a Workflow's custom AI model here:

How do I upload documents into the training module and improve my AI document categorization and extraction accuracy?