Google Professional Machine Learning Engineer Question 56
Single answerGoogle Cloud PlatformYou are tasked with building an image classification model for a retail company that wants to categorize product images into 10 distinct categories. The company has provided you with a labeled dataset. You have limited machine learning expertise on the team and need a solution that minimizes the time spent on model development while still achieving a high level of accuracy. Which approach should you take using Google Cloud AutoML?
- A
Use AutoML Vision to train a custom image classification model by uploading your labeled dataset.
- B
Use BigQuery ML to build a logistic regression model for image classification.
- C
Manually design a convolutional neural network (CNN) using TensorFlow and train it on the dataset.
- D
Use Vertex AI Feature Store to preprocess the dataset and train the model automatically.
Show answer and explanation
Correct answer: A
Explanation
The best approach for this scenario is to use AutoML Vision because it is specifically designed for image classification tasks, requires minimal machine learning expertise, and provides an easy way to build high-performing models by simply uploading labeled data. The other options either do not fit the problem requirements or require advanced expertise, which the team lacks.
- A. Correct.
AutoML Vision is specifically designed for image classification tasks and allows you to train a custom model by simply uploading labeled data. It is suitable for teams with limited ML expertise and provides a streamlined process.
- B. Incorrect.
BigQuery ML is designed for tabular data and does not support image classification tasks. It is not the appropriate tool for this use case.
- C. Incorrect.
Manually designing a CNN requires significant machine learning expertise and time, which contradicts the requirement to minimize development time and cater to a team with limited expertise.
- D. Incorrect.
Vertex AI Feature Store is a tool for managing and serving features for ML models, not for training models directly. It does not provide automated model training.