Exam AI-102 topic 5 question 6 discussion

1. Enable active learning 2. Validate the utterances logged for review and modify the model 3. Train and republish the language understanding model not sure how the pre built model can improve accuracy....

Absolutely wrong. To improve the ability of a Language Understanding service that's frequently responding with "Sorry, I don't understand that," I need to select three actions in the correct sequence. The appropriate sequence is: 1. Enable log collection by using Log Analytics. 2. Validate the utterances logged for review and modify the model. 3. Train and republish the Language Understanding model. This approach follows a logical improvement cycle: First, you need data about what users are actually saying that the model doesn't understand, which Log Analytics will provide by collecting user interactions. Second, you examine these logged utterances to identify patterns in what's being missed, then modify your model to address these gaps (adding intents, entities, or example utterances as needed). Finally, after making these improvements, you train the updated model to incorporate the changes and republish it so users can benefit from the improvements. This sequence creates a data-driven approach to iteratively improving the model based on real user interactions.

1) Enable active learning is CORRECT. This step allows the Language Understanding service to collect and suggest new utterances that are not well understood. Active learning helps identify gaps in the model. 2) Validate the utterances logged for review and modify the model is CORRECT. Review the utterances that the system flagged as not well understood and use these to improve the model. Modify the intents and entities to better capture the user's queries. 3) Train and republish the Language Understanding model is CORRECT. After making modifications based on the reviewed utterances, train the model again to incorporate the changes. Then, republish the model so that the updates take effect and the agent can respond more accurately to user requests.

a) Add prebuilt domain models as required. Validate the utterances logged for review and modify the model. (2) b) Migrate authoring to an Azure resource authoring key. c) Enable active learning. d) Enable log collection by using Log Analytics. (1) e) Train and republish the Language (3) f) Understanding model.

Copilot says To improve the ability of the agent to respond to requests, you should perform the following actions in sequence: Enable log collection using Log Analytics: This will allow you to collect and analyze the utterances that the agent is not understanding. Enable active learning: Active learning will help the model to learn from the utterances it didn't understand and improve over time. Train and republish the language understanding model: After enabling active learning and collecting more data, you should retrain the model and publish it again. This will incorporate the new learnings into the model and should improve its ability to understand user requests.

1. Enable active learning 2. Validate the utterances logged for review and modify the model 3. Train and republish the language understanding model

Final Answer: 1. Enable active learning 2. Validate the utterances logged for review and modify the model 3. Train and republish the language understanding model

given answer is NOT Correct! why: Enable Active Learning: LUIS's Active Learning feature automatically identifies vague or low-confidence user inputs that may require further training. Review and Add Utterances Logged for Review: Utterances collected through Active Learning are marked for review. Regularly reviewing and categorizing these utterances into the relevant intents is a crucial step in enhancing model performance. Train and Republish the Language Understanding Model: After adding new utterances and potentially adjusting intents and entities, you need to retrain the LUIS model to incorporate these updates.

Same as Question 4. https://www.examtopics.com/discussions/microsoft/view/57673-exam-ai-102-topic-5-question-4-discussion

1. Enable active learning 2. Validate the utterances logged for review and modify the model 3. Train and republish the language understanding model