AI-102 Question 412
Select 4You are tasked with automating the processing of invoices for your company using Azure AI. You decide to create a custom document intelligence model using Azure Form Recognizer. After uploading a dataset of sample invoices, what steps should you take to train, test, and publish the model?
- A
Label the fields in the sample documents to train the model.
- B
Directly publish the model without training or testing, as Azure Form Recognizer automatically detects document fields.
- C
Train the model using the labeled data within the Azure Form Recognizer studio.
- D
Test the model by submitting new sample invoices and verifying the extracted data.
- E
Deploy the model using an Azure Cognitive Services endpoint to make it accessible for integration.
Show answer and explanation
Correct answers: A, C, D, E
Explanation
To create a custom document intelligence model using Azure Form Recognizer, you need to label the fields in your sample documents, train the model with this labeled data, and test it to ensure it accurately extracts the needed information. Once validated, you can deploy the model using an Azure Cognitive Services endpoint to make it available for integration with other systems. Skipping any of these steps can lead to incomplete or inaccurate results.
- A. Correct.
Correct: Labeling fields is a necessary step to help the model understand the structure and key-value pairs in the documents.
- B. Incorrect.
Incorrect: While Azure Form Recognizer provides pre-trained functionality, a custom model requires training and testing to ensure it meets specific needs.
- C. Correct.
Correct: Training the model with labeled data is required to teach it to recognize patterns and extract the correct fields.
- D. Correct.
Correct: Testing the model ensures it performs accurately on unseen documents and meets the requirements before publishing.
- E. Correct.
Correct: Deploying the model via an endpoint is essential for making it accessible to other systems or applications.