AI-102 Question 275
Select 3You are working on optimizing a language understanding model built using Azure Language Understanding (LUIS). The model has a high rate of misclassified intents for user queries. Which of the following actions should you take to improve the model's performance?
- A
Add more labeled example utterances for intents that are frequently misclassified.
- B
Reduce the number of intents in your model by combining similar intents where possible.
- C
Enable active learning to identify and label ambiguous or frequently misunderstood utterances.
- D
Increase the number of entities defined in your model to improve intent classification.
- E
Train the model without validating it to save processing time.
Show answer and explanation
Correct answers: A, B, C
Explanation
To optimize a language understanding model, it is important to provide sufficient labeled data, simplify the model structure when possible, and leverage active learning to address ambiguous cases. These steps help the model generalize better and improve accuracy in recognizing intents.
- A. Correct.
Adding more labeled example utterances helps the model learn to recognize patterns more effectively for specific intents, especially when they are frequently misclassified.
- B. Correct.
Reducing the number of intents and combining similar ones can reduce confusion for the model, improving overall classification accuracy.
- C. Correct.
Enabling active learning allows the system to suggest utterances that are difficult to classify, which can then be labeled to improve the model's understanding.
- D. Incorrect.
Increasing the number of entities does not directly improve intent classification, as entities are primarily used for extracting specific information from utterances rather than distinguishing between intents.
- E. Incorrect.
Training the model without validating it can lead to overfitting or poor generalization, which is counterproductive to optimizing the model.