A. Sparse coding removes the need for input data B. Sparse coding is only a file I/O technique C. Sparse coding encourages representations where only a small number of units are active D. Sparse coding forces every hidden unit to be active
Correct Answer: C. Sparse coding encourages representations where only a small number of units are active
A. Computer vision requires no evaluation metrics B. CNNs are unrelated to computer vision C. Deep learning can be applied to image classification, detection, and segmentation D. Deep learning cannot process images
Correct Answer: C. Deep learning can be applied to image classification, detection, and segmentation
A. A test set is used to update weights every epoch B. A test set should be identical to the training set C. A test set is used for data normalization only D. A test set estimates performance on unseen data
Correct Answer: D. A test set estimates performance on unseen data
A. A deep belief network is the same as a spreadsheet B. A deep belief network has no hidden variables C. A deep belief network is only a sorting algorithm D. A deep belief network is built using stacked probabilistic latent-variable layers
Correct Answer: D. A deep belief network is built using stacked probabilistic latent-variable layers
A. Speech recognition is impossible with neural networks B. Speech recognition uses only database joins C. Deep models cannot process audio signals D. Deep learning can map acoustic or sequential features to speech units or text
Correct Answer: D. Deep learning can map acoustic or sequential features to speech units or text
A. A validation set supports model selection and hyperparameter tuning B. A validation set is always larger than the training set C. A validation set replaces the loss function D. A validation set stores GPU kernels
Correct Answer: A. A validation set supports model selection and hyperparameter tuning
A. A restricted Boltzmann machine has visible and hidden units with no connections within the same layer B. An RBM has connections only among visible units C. An RBM is a deterministic decision tree D. An RBM cannot model probability distributions
Correct Answer: A. A restricted Boltzmann machine has visible and hidden units with no connections within the same layer
A. Deep learning is used for language modeling, translation, and text classification B. NLP models cannot use embeddings C. Text classification is not a machine learning task D. NLP cannot involve sequence models
Correct Answer: A. Deep learning is used for language modeling, translation, and text classification
A. Generalization can only be measured on training loss B. Generalization is the ability to perform well on unseen examples C. Generalization means memorizing every training sample D. Generalization is unrelated to overfitting
Correct Answer: B. Generalization is the ability to perform well on unseen examples
A. An RNN has no hidden state B. A recurrent neural network processes sequential data using connections across time steps C. An RNN ignores ordering in sequences completely D. An RNN can only classify still images
Correct Answer: B. A recurrent neural network processes sequential data using connections across time steps
A. Loss is never computed in deep learning B. Evaluation metrics such as accuracy, precision, recall, and loss help compare models C. Evaluation metrics are irrelevant after training D. Accuracy is always sufficient for imbalanced data
Correct Answer: B. Evaluation metrics such as accuracy, precision, recall, and loss help compare models
A. Overfitting always improves test accuracy B. Overfitting occurs only in linear regression C. Overfitting occurs when a model learns noise or specific training patterns too strongly D. Overfitting means the model is too simple for the data
Correct Answer: C. Overfitting occurs when a model learns noise or specific training patterns too strongly
