Google Professional Machine Learning Engineer Question 35
Single answerGoogle Cloud PlatformYou are developing a customer support chatbot for a retail company. The chatbot must classify customer inquiries into predefined categories (e.g., 'Order Status,' 'Product Inquiry,' 'Returns') and provide responses using natural language understanding. You want to leverage Google Cloud's pre-trained models from Model Garden to get started quickly without custom model training. Which approach should you take?
- A
Use the Google Cloud Natural Language API to classify the inquiries into categories and generate responses.
- B
Deploy a pre-trained text classification model from Model Garden and integrate it with your application to classify the inquiries.
- C
Deploy a pre-trained language model from Model Garden to classify the inquiries and fine-tune it using your customer inquiry dataset.
- D
Build a custom machine learning model using TensorFlow and train it with a labeled dataset of customer inquiries.
Show answer and explanation
Correct answer: B
Explanation
To quickly classify customer inquiries into predefined categories, deploying a pre-trained text classification model from Model Garden is the most efficient approach. These models are ready to use and can be integrated into your application without additional training, meeting the requirements specified in the scenario.
- A. Incorrect.
The Google Cloud Natural Language API can be used to analyze text but does not directly provide pre-trained models for specific classification tasks, such as customer inquiry categorization.
- B. Correct.
Using a pre-trained text classification model from Model Garden allows you to leverage a ready-to-use solution for classification tasks, which matches the requirement of quickly implementing a solution without custom training.
- C. Incorrect.
While deploying and fine-tuning a pre-trained language model from Model Garden is a valid approach, the question specifies that you want to get started quickly without additional model training.
- D. Incorrect.
Building a custom machine learning model from scratch is time-consuming and contradicts the requirement of leveraging pre-trained models from Model Garden to avoid custom training.