Deep Learning Unit 4

Which activation function is commonly used in the hidden layers of Recurrent Neural Networks?

• ReLU

• Sigmoid

• Tanh

• Linear

What is the main limitation of basic RNNs when dealing with long sequences?

• Vanishing gradients

• Exploding gradients

• Overfitting

• Underfitting

Which of the following sequences can be best modeled using Recurrent Neural Networks?

• Images

• Text

• Audio

• All the options

Which component of the LSTM unit controls the flow of information into the cell state?

• Input gate

• Output gate

• Forget gate

• Cell state

What is the role of the cell state in an LSTM unit?

• To store long-term information

• To control the flow of information

• To compute the hidden state

• To compute the output gate

Which of the following statements about LSTM networks is true?

• LSTM networks have no internal memory.

• LSTM networks are incapable of handling sequential data.

• LSTM networks are a type of feedforward neural network.

• LSTM networks can handle vanishing gradient problems in RNNs.

Which of the following gates controls the flow of information from the previous hidden state?

• Update gate

• Reset gate

• Forget gate

• Input gate

Which activation function is commonly used in the update gate of a GRU unit?

• ReLU

• Sigmoid

• Tanh

• Linear

What is the purpose of the reset gate in a GRU unit?

• To reset the hidden state to its initial value

• To reset the cell state to its initial value

• To reset the previous hidden state to its initial value

• To reset the current hidden state to its initial value

In natural language processing (NLP), what task involves predicting the next word in a sequence given the previous words?

• Sentiment analysis

• Named entity recognition

• Part-of-speech tagging

• Language modeling