1. . Deep learning is:
(A) Compressing datasets only
(B) Encrypting neural networks only
(C) A subset of machine learning that uses neural networks with multiple layers to model complex patterns
(D) Backup only
2. . The main advantage of deep learning over traditional machine learning is:
(A) Compressing data
(B) Encrypting features only
(C) Ability to automatically learn features from raw data
(D) Backup only
3. . A neuron in a neural network is:
(A) Compressing unit only
(B) Encrypting unit only
(C) A computational unit that receives input, applies weights, sums them, and passes through an activation function
(D) Backup only
4. . Weights in a neural network represent:
(A) Backup only
(B) Encrypting weights
(C) Compressing weights
(D) The importance of each input in computing the output
5. . Bias in a neuron helps to:
(A) Encrypt bias
(B) Shift the activation function to better fit the data
(C) Compress bias
(D) Backup only
6. . Activation functions introduce:
(A) Compressing non-linearity
(B) Encrypting non-linearity
(C) Non-linearity to neural networks allowing them to learn complex patterns
(D) Backup only
7. . Common activation functions include:
(A) Compressing functions only
(B) Encrypting functions only
(C) Sigmoid, Tanh, ReLU, Leaky ReLU, and Softmax
(D) Backup only
8. . The output layer in a neural network:
(A) Encrypts the output
(B) Produces the final prediction of the network
(C) Compresses the output
(D) Backup only
9. . Loss function in deep learning measures:
(A) Backup only
(B) Encrypting errors
(C) Compressing errors
(D) The difference between predicted and actual outputs
10. . Common loss functions include:
(A) Compressing losses only
(B) Encrypting losses only
(C) Mean Squared Error (MSE), Cross-Entropy Loss, Hinge Loss
(D) Backup only
11. . Backpropagation is:
(A) Backup only
(B) Encrypting gradients
(C) Compressing gradients
(D) An algorithm to compute gradients of the loss function with respect to weights for learning
12. . Optimizers in deep learning help to:
(A) Backup only
(B) Encrypt weights
(C) Compress weights
(D) Update network weights to minimize the loss function
13. . Common optimizers include:
(A) Encrypting optimizers
(B) Gradient Descent, Stochastic Gradient Descent (SGD), Adam, RMSprop
(C) Compressing optimizers
(D) Backup only
14. . Epoch in deep learning is:
(A) Encrypting data pass
(B) One complete pass of the entire training dataset through the network
(C) Compressing epoch
(D) Backup only
15. . Batch size refers to:
(A) Encrypting batch
(B) The number of training samples processed before updating weights
(C) Compressing batch
(D) Backup only
16. . Overfitting occurs when:
(A) Compressing models
(B) Encrypting training data
(C) The model learns training data too well and fails to generalize on new data
(D) Backup only
17. . Dropout in deep learning is used to:
(A) Randomly deactivate neurons during training to prevent overfitting
(B) Encrypt neurons
(C) Compress activations
(D) Backup only
18. . Convolutional Neural Networks (CNNs) are mainly used for:
(A) Encrypting images
(B) Image and video data processing
(C) Compressing images
(D) Backup only
19. . Recurrent Neural Networks (RNNs) are suitable for:
(A) Compressing sequences
(B) Encrypting sequences
(C) Sequential data such as time series or text
(D) Backup only
20. . The main purpose of deep learning fundamentals is to:
(A) Compress all features
(B) Encrypt all data
(C) Build models that can automatically learn complex patterns from data for prediction or classification tasks
(D) Backup only