Google Professional Machine Learning Engineer
Get started today
Ultimate access to all questions.
As a Google Professional Machine Learning Engineer, you are tasked with enhancing a visual search engine for an online retail company. The ML pipeline is designed to classify images based on the presence of the company's products. With the anticipation of new product launches, a retraining feature has been integrated into the pipeline. To ensure the model maintains high accuracy, AI Platform's continuous evaluation service will be utilized. Considering the need for a robust evaluation strategy that accommodates new products without compromising the assessment of existing product classifications, what is the most effective approach to manage your test dataset? Choose the best option.
As a Google Professional Machine Learning Engineer, you are tasked with enhancing a visual search engine for an online retail company. The ML pipeline is designed to classify images based on the presence of the company's products. With the anticipation of new product launches, a retraining feature has been integrated into the pipeline. To ensure the model maintains high accuracy, AI Platform's continuous evaluation service will be utilized. Considering the need for a robust evaluation strategy that accommodates new products without compromising the assessment of existing product classifications, what is the most effective approach to manage your test dataset? Choose the best option.
Explanation:
Extending the test dataset with images of newer products ensures that the model's accuracy is evaluated against a comprehensive set of data that includes both existing and new products. This approach maintains the model's relevance and effectiveness as the product line evolves. Replacing the test dataset or keeping it unchanged could lead to inadequate evaluation of the model's performance on the full range of products, while waiting for performance metrics to drop below a threshold might delay necessary updates to the test dataset.