Skip to main content
Note: You can use a separate document set to train your Deep Learning activity. To do so, select the Deep Learning activity from the drop-down list next to the skill name. Then, in the drop-down list to the left of the Upload button, select the necessary document set or click Create Set… to create a new one. You can upload, delete, and rotate documents on this tab as described in the Documents section.

​
Setup Steps

To set up a Deep Learning activity:

​
Step 1: Add the Activity

On the Activities tab, add a Deep Learning activity for semi-structured documents to your document processing flow.

​
Step 2: Select Fields

In the Activity Properties pane, select the fields to be trained using this activity. You can select one of the following:
  • Up to 50 fields of type Text, regardless of their nesting level.
  • One table with up to 32 columns.
If you need to train more fields, you can add more Deep Learning activities and use them to select additional fields. For example, if you need to train several text fields and a table, create two Deep Learning activities.
Note: The following fields cannot be trained:
  • Fields of type other than Text
  • Groups with multiple items, tables, or text fields with multiple items nested in a group with multiple items
  • Tables with more than 32 columns
You will need to set up extraction of such fields using other activities, e.g. an Extraction Rules activity.

​
Step 3: Label Documents

Click Activity Editor and go to the Fields tab to label your documents. The labeling process in the Activity Editor is identical to the regular document labeling process. The following guidelines will help you decide on the size of the document set:
  • If the training set contains fewer than 100 documents, you will need to upload and label more documents before you can start the training process.
  • If the training set contains between 100 and 1,000 documents, you will be able to start training your activity, but Advanced Designer will display a warning saying that you should label at least 1,000 documents to achieve good extraction quality.
  • If the training set contains between 1,000 and 10,000 documents, you will be able to start training your activity right away. This is the recommended number of documents to have in your training set.
  • If the training set contains more than 10,000 documents, Advanced Designer will display a warning saying that the skill may become unstable.

​
Step 4: Train the Activity

Once you have uploaded and labeled your documents, click Train Activity.

​
Step 5: Monitor Training Progress

Go to the Results tab to evaluate the training progress. If necessary, adjust the training length or stop the training. For more information, see Monitoring and adjusting activity training.

​
Post-Training Steps

Once the activity has been trained, activity testing will start automatically. If you stop the training, you will be prompted to start testing the activity manually. When the testing is complete, analyze the field extraction results in the Activity Test Results section of the Results tab. Statistics for this activity are identical to the general statistics for the skill displayed on the Results tab. If you are not satisfied with the quality of field extraction, you have the following options:
  • Add more documents to the training set and resume the training process. The training results obtained so far will be preserved and the neural network will be additionally trained using the updated document set.
  • Adjust the labeling and restart training. The training results obtained so far will be discarded and the neural network will be trained from scratch.
  • Create a Hypothesis Filtering container with an Extraction Rules activity, which will allow you to set conditions for the output of the Deep Learning activity.
The activity can only be trained and tested using documents with confirmed labeling. Documents have unconfirmed labeling if the reference labeling was generated automatically based on the predicted labeling, unless you copy predicted labeling to reference using the corresponding option in the document context menu. You can check the labeling status for each document on the Documents tab. To confirm labeling for a document, you should review it on the Fields tab.
Note: Beginning with Advanced Designer v. 2.3.1, the field limitations for the Deep Learning activity have changed. If your skill uses a trained Deep Learning activity that extracts more than 50 fields, you can continue processing documents with that skill. However, when you open such a skill for editing, the existing Deep Learning activity will be split into several Deep Learning activities, which you may need to retrain. You will also have to route the activities in the document processing workflow.