What is the core idea behind Bayesian optimization for hyperparameter tuning?

To build a probabilistic model of the objective function and use it to decide where to sample next

To randomly sample hyperparameters from a uniform distribution

To use gradient descent to find the optimal hyperparameters

To exhaustively search all possible hyperparameter combinations

Why is random search often more efficient than grid search when only a few hyperparameters are actually important?

Because it uses fewer total evaluations to cover the same range of values for each hyperparameter

Because it evaluates each hyperparameter independently

Because it avoids evaluating any hyperparameter combination more than once

Because it always finds the global optimum

Why is cross-validation used during hyperparameter tuning?

To evaluate hyperparameter configurations on multiple data splits and reduce the risk of overfitting to a single validation set

To increase the size of the training set

To eliminate the need for a test set

To speed up the training process by using only a subset of data

What is the purpose of early stopping in hyperparameter tuning?

To stop using the test set during tuning

To stop the tuning process after a fixed number of hyperparameter combinations

To stop training a model once its performance on a validation set stops improving

To stop the model from learning any more features

A data science team is tuning a deep neural network for image classification. Each training run takes 30 minutes, and they have a budget of 20 hours. They need to find a good set of hyperparameters (learning rate, batch size, dropout rate) with as few runs as possible. Which approach is most appropriate?

Bayesian optimization with 40 trials

Grid search with 3 values per hyperparameter

Random search with 40 iterations

SnowflakeSnowPro Advanced — Data Scientist

Domain 4Objective 3

Tune Model Hyperparameters SNOWPRO-ADVANCED-DATA-SCIENTIST Practice Questions (Page 6)

Part of the Model Development domain, which makes up ~19% of our current practice bank.

32questions here

7free pages

10concepts

Questions 26–30

26foundation · easy
What is the core idea behind Bayesian optimization for hyperparameter tuning?
Select an answer first
27foundation · easy
Why is random search often more efficient than grid search when only a few hyperparameters are actually important?
Select an answer first
28foundation · easy
Why is cross-validation used during hyperparameter tuning?
Select an answer first
29foundation · easy
What is the purpose of early stopping in hyperparameter tuning?
Select an answer first
30expert · hard
A data science team is tuning a deep neural network for image classification. Each training run takes 30 minutes, and they have a budget of 20 hours. They need to find a good set of hyperparameters (learning rate, batch size, dropout rate) with as few runs as possible. Which approach is most appropriate?
Select an answer first

Finished these 5 questions?

Review the revealed explanations, or continue through the curriculum.

Next 5 questions →Back to SNOWPRO-ADVANCED-DATA-SCIENTIST

Free Basic Practice is a study aid with revealable answers — not a scored exam. Examers.io is independent and not affiliated with or endorsed by Snowflake. “SNOWPRO-ADVANCED-DATA-SCIENTIST” is a trademark of its owner, used for identification only.