Understanding Regularization Techniques

What is L1 regularization also known as?

• Ridge regularization

• Lasso regularization

• Dropout regularization

• Elastic Net regularization

How does L2 regularization differ from L1 regularization?

• L1 regularization promotes sparsity by driving some coefficients to zero, while L2 regularization distributes weights more evenly without leading to sparsity.

• L1 regularization is used for classification tasks only, while L2 is for regression tasks.

• L2 regularization eliminates all coefficients, while L1 regularization keeps them all.

• L1 regularization is always more effective than L2 regularization.

What is the primary purpose of regularization in machine learning?

• To increase model complexity.

• To reduce training time.

• To enhance data accuracy.

• To prevent overfitting and improve generalization.

How can overfitting be identified in a model?

• The model performs equally well on both training and validation datasets.

• The model has a high accuracy on the validation dataset only.

• Overfitting can be identified by a significant performance gap between training and validation datasets.

• The model's complexity is reduced without affecting performance.

What role do hyperparameters play in regularization?

• Hyperparameters are used to select the model architecture.

• Hyperparameters control the strength and type of regularization, helping to prevent overfitting.

• Hyperparameters only affect the learning rate of the model.

• Hyperparameters determine the number of training epochs.

Which regularization technique tends to produce sparse models?

• L1 regularization (Lasso)

• L2 regularization (Ridge)

• Dropout regularization

• Early stopping

What is the effect of increasing the regularization parameter in L2 regularization?

• It reduces model complexity and helps prevent overfitting.

• It improves the model's accuracy on the training set.

• It has no effect on the model's performance.

• It increases model complexity and leads to overfitting.

How can feature selection improve model performance?

• Feature selection always leads to worse model accuracy.

• Feature selection improves model performance by reducing overfitting, enhancing generalization, and speeding up training.

• Feature selection has no impact on model training time.

• Feature selection increases the number of features used in the model.

What is a common method for hyperparameter tuning?

• Neural architecture search

• Grid search

• Random search

• Bayesian optimization

Why is it important to balance bias and variance in a model?

• To increase the amount of training data available.

• To ensure the model runs faster during training.

• To minimize the complexity of the model.

• It is important to balance bias and variance to achieve optimal model performance and generalization.