deep learning 12/3/24

What is the role of the loss function in training a deep learning model?

• To measure the difference between predicted and actual values

• To initialize the weights of the neural network

• To control the learning rate during optimization

• None of the above

How does backpropagation help in updating the weights of a neural network?

• By calculating the gradient of the loss function with respect to each weight

• By randomly initializing the weights at the start of training

• By applying a fixed learning rate to all weights equally

• None of the above

Which parameter need to be learnt in minimizing objective function in supervised learning

• only weight

• only bias

• both weight and bias

• none

How does a convolutional neural network differ from a fully connected neural network?

• Convolutional neural networks have only one layer

• Convolutional neural networks use convolutional layers for feature extraction

• Fully connected neural networks are more efficient for image recognition tasks

• None of the above

What is the purpose of dropout regularization in deep learning?

• To increase the number of neurons in each layer

• To prevent overfitting by randomly dropping neurons during training

• To speed up the training process by skipping certain layers

• None of the above

How does a recurrent neural network (RNN) differ from a feedforward neural network?

• RNN can only process sequential data while feedforward can process any data

• RNN has connections that form a directed cycle while feedforward has no cycles

• RNN uses convolutional layers for feature extraction while feedforward does not

• None of the above

How does transfer learning benefit deep learning models?

• By randomly initializing the weights at the start of training

• By applying a fixed learning rate to all weights equally

• By leveraging knowledge from pre-trained models to improve performance on new tasks

• None of the above

What is the significance of the learning rate in training a neural network?

• To increase the number of neurons in each layer

• To prevent overfitting by adjusting the weights during training

• To control the step size of weight updates and influence convergence

• None of the above

What is the purpose of the softmax function in the output layer of a neural network?

• To introduce non-linearity and enable complex mappings

• To measure the difference between predicted and actual values

• To convert raw scores into probabilities for multi-class classification

• None of the above

Limitations of RNN is /are

• Exploding Gradient

• Unable to handle Time series data

• Unable to handle sequential data

• Both A & B