Google Cloud Professional Machine Learning Engineer — Question 279

You developed a Python module by using Keras to train a regression model. You developed two model architectures, linear regression and deep neural network (DNN), within the same module. You are using the training_method argument to select one of the two methods, and you are using the learning_rate and num_hidden_layers arguments in the DNN. You plan to use Vertex AI's hypertuning service with a budget to perform 100 trials. You want to identify the model architecture and hyperparameter values that minimize training loss and maximize model performance. What should you do?

Answer options

• A. Run one hypertuning job for 100 trials. Set num_hidden_layers as a conditional hyperparameter based on its parent hyperparameter training_method, and set learning_rate as a non-conditional hyperparameter.
• B. Run two separate hypertuning jobs, a linear regression job for 50 trials, and a DNN job for 50 trials. Compare their final performance on a common validation set, and select the set of hyperparameters with the least training loss.
• C. Run one hypertuning job with training_method as the hyperparameter for 50 trials. Select the architecture with the lowest training loss, and further hypertune it and its corresponding hyperparameters tor 50 trials.
• D. Run one hypertuning job for 100 trials. Set num_hidden_layers and learning_rate as conditional hyperparameters based on their parent hyperparameter training_method.

Correct answer: D

Explanation

Option D is the correct choice because it allows both num_hidden_layers and learning_rate to be adjusted based on the selected training_method, ensuring that the hyperparameters are relevant to the chosen model architecture. Option A incorrectly treats learning_rate as non-conditional, which may lead to suboptimal configurations. Option B unnecessarily splits the trials between two jobs, which limits the exploration of hyperparameters. Option C does not utilize the full budget effectively, as it only runs a single trial for the first stage before further tuning.