Deep Learning Batch 2

What is the primary goal of optimization in deep learning?

• A) To minimize the training error

• B) To maximize the generalization error

• C) To increase the model complexity

• ) To reduce the number of layers

Which of the following is a common issue with high variance in a model?

• A) Underfitting

• B) Overfitting

• C) High bias

• D) Low accuracy on training data

What is the main advantage of Mini-Batch Gradient Descent over Batch Gradient Descent?

• A) It requires less memory

• B) It converges faster for large datasets

• C) It is less noisy

• D) It always reaches the global minimum

Which of the following is true about the bias-variance trade-off?

• A) Increasing model complexity reduces bias but increases variance

• B) Increasing model complexity reduces both bias and variance

• C) Decreasing model complexity reduces bias but increases variance

• D) Decreasing model complexity reduces both bias and variance

What is the primary purpose of early stopping in deep learning?

• A) To reduce the number of layers in the network

• B) To stop training when the validation error starts increasing

• C) To increase the learning rate

• D) To reduce the number of parameters in the model

Which of the following is true about the Adagrad optimizer?

• A) It uses a fixed learning rate for all parameters

• B) It is less efficient than SGD

• C) It adapts the learning rate based on the history of gradients

• D) It does not use momentum

What is the main challenge in deep learning related to data?

• A) Lack of computational power

• B) Overfitting due to small datasets

• C) The need for large amounts of data

• D) The complexity of neural networks

Which of the following is a common technique for hyperparameter optimization?

• A) Grid search

• B) Early stopping

• C) Dropout

• D) Data augmentation

What is the primary purpose of data augmentation in deep learning?

• A) To reduce the number of layers in the network

• B) To increase the size of the training dataset artificially

• C) To reduce the learning rate

• D) To increase the model complexity

Which of the following is true about the RMSprop optimizer?

• A) It uses a fixed learning rate for all parameters

• B) It adapts the learning rate based on the magnitude of recent gradients

• C) It is less efficient than SGD

• D) It does not use momentum