Which of the following types of data is used to tune hyperparameters?

The correct answer is A. Validation. In the standard machine learning workflow, datasets are split into three roles: training data fits the model's parameters, validation data is used to evaluate performance during development and tune hyperparameters (e.g., learning rate, depth, regularization strength), and test…

AI Security Design and Implementation

Question

Options

AValidation
BConfiguration
CTraining
DTest

How the community answered

(33 responses)

A
91% (30)
B
6% (2)
D
3% (1)

Explanation

In the standard machine learning workflow, datasets are split into three roles: training data fits the model's parameters, validation data is used to evaluate performance during development and tune hyperparameters (e.g., learning rate, depth, regularization strength), and test data provides a final unbiased evaluation after all tuning is complete. Validation data (A) is correct because it acts as a proxy for unseen data during the iterative tuning process without contaminating the final test evaluation. Using training data (C) for tuning would cause overfitting. Test data (D) must remain untouched until final evaluation to ensure an honest performance estimate. 'Configuration data' (B) is not a standard ML dataset category.

Topics

#Machine Learning Fundamentals#Hyperparameter Tuning#Data Splitting#Model Development

Community Discussion

No community discussion yet for this question.

Full AAISM Practice