Google Professional Machine Learning Engineer
Get started today
Ultimate access to all questions.
You are tasked with developing a custom Deep Neural Network in Keras to predict customer purchases from historical transaction data. The project requires evaluating model performance across various architectures, storing training data, and comparing evaluation metrics on a unified dashboard. The solution must be scalable, cost-effective, and comply with data privacy regulations. Which of the following approaches best facilitates this? (Choose two correct options)
You are tasked with developing a custom Deep Neural Network in Keras to predict customer purchases from historical transaction data. The project requires evaluating model performance across various architectures, storing training data, and comparing evaluation metrics on a unified dashboard. The solution must be scalable, cost-effective, and comply with data privacy regulations. Which of the following approaches best facilitates this? (Choose two correct options)
Explanation:
Kubeflow Pipelines (D) is the optimal solution for systematically managing and comparing multiple runs due to its comprehensive features for experiment tracking, pipeline creation, and dashboarding. However, combining AI Platform and Kubeflow Pipelines (E) offers a more scalable and flexible approach, especially for large-scale projects requiring detailed metrics analysis and compliance with data privacy regulations. AutoML Tables (A) simplifies model generation but lacks the flexibility for custom architectures. Cloud Composer (B) is more suited for workflow orchestration rather than model comparison. AI Platform (C) provides a built-in dashboard but lacks the systematic experiment management offered by Kubeflow Pipelines.