Fundamentals of Neural Networks - Forward Propagation in RNN

What is the primary function of a recurrent neural network?

• To perform unsupervised learning

• To classify non-sequential data

• To handle sequential data

• To process static images

In the context of RNNs, what does the weight 'WAX' connect?

• The activation A and the output Y

• The input X and the output Y

• The activation A and the bias term

• The input X and the activation A

What is the role of the weight 'WYA' in RNNs?

• It connects the input X to the output Y

• It connects the activation A to the output Y

• It connects the activation A to the bias term

• It connects the input X to the activation A

Why is the sequence initialized with a vector of zeros in RNNs?

• To increase the complexity of the model

• To enhance the learning rate

• To ensure the sequence starts with a neutral state

• To reduce the computational cost

How does the RNN handle the limitation of time series data?

• By dynamically adjusting the sequence length

• By ignoring the sequence length

• By setting a fixed sequence length

• By using infinite loops

What is the significance of shared weights in RNNs?

• They increase the model's flexibility

• They reduce the model's complexity

• They ensure consistency across timestamps

• They enhance the model's speed

What is the purpose of folding the RNN diagram?

• To reduce the number of weights

• To enhance the accuracy of predictions

• To increase the number of neurons

• To simplify the representation

What is the main difference between folded and unfolded RNN diagrams?

• Folded diagrams have more neurons

• Unfolded diagrams are more complex

• Folded diagrams simplify the representation

• Unfolded diagrams have fewer weights

Which activation function is commonly used in RNNs?

• Softmax

• Tanh

• ReLU

• Sigmoid

What is the purpose of backpropagation through time in RNNs?

• To enhance the model's speed

• To handle static data

• To optimize the model's weights

• To process non-sequential data