AI-102 Practice Question: Implement knowledge mining and document intelligence solutions
A company uses Azure Document Intelligence to process custom forms. They have trained a custom model using labeled data. They need to improve the model's accuracy for a specific field that is frequently misrecognized. The field appears in a consistent location but has varying formats. Which two actions should they take? (Choose two.)
⚠ Common exam trap
The trap here is thinking that adjusting confidence thresholds or changing the model type can fix recognition errors, when the real solution lies in improving the training data.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Add more labeled samples that include variations of the field's format to the training dataset.
Improving a custom model's accuracy requires enhancing the training data. Adding labeled samples with format variations exposes the model to diverse representations, while correcting low-confidence instances ensures the training set is accurate. These actions directly address the model's learning process, unlike threshold adjustments or architectural changes.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✓
Add more labeled samples that include variations of the field's format to the training dataset.
Why this is correct
Adding more labeled samples with format variations helps the model learn to generalize and recognize the field regardless of format changes. The custom model learns from the labeled examples, so increasing diversity in the training data directly improves accuracy for that field. This is a fundamental step in improving model performance.
- ✗
Use the prebuilt model for that field type instead of the custom model.
Why it's wrong here
Prebuilt models are designed for common document types (e.g., invoices, receipts) and may not include the specific field or format required. Switching to a prebuilt model would likely reduce accuracy for the custom field. Custom models are tailored to the specific forms and fields, so they should be improved rather than replaced.
- ✓
Use the model's confidence scores to identify low-confidence instances and correct their labels in the training set.
Why this is correct
Reviewing low-confidence predictions and correcting mislabeled or unlabeled instances in the training set helps the model learn from its mistakes. This iterative feedback loop is a best practice for improving custom model accuracy. By ensuring the training data is accurate and representative, the model can better distinguish the field's patterns.
- ✗
Retrain the model using the same dataset but with a different neural network architecture.
Why it's wrong here
Azure Document Intelligence custom models use a fixed architecture; you cannot choose a different neural network. Retraining with the same data would yield similar results. Improving accuracy requires better or more diverse training data, not architectural changes, which are not exposed to the user.
- ✗
Increase the model's confidence threshold to reduce false positives for that field.
Why it's wrong here
Adjusting the confidence threshold affects post-processing decisions (e.g., whether to accept a prediction) but does not improve the model's underlying accuracy. It may reduce false positives at the cost of more false negatives. The goal is to improve the model's ability to recognize the field correctly, not just filter its output.
Go deeper
Related to this question
About these practice questions
Courseiva writes every AI-102 question from scratch — 761 in total, each with an explanation and a wrong-answer breakdown. None are copied from real exams or dumps. Learn why practice questions differ from exam dumps →
JA
Written and reviewed by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
Last reviewed September 2026 · checked against the official Microsoft exam blueprint
This AI-102 practice question is part of Courseiva's free Microsoft certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the AI-102 exam.