DL_Unit-4

Complete Roll Number

In the context of RNNs, what is the purpose of unfolding the network?

• To increase model complexity

• To visualize the network architecture

• To extend the sequence length for training

• To decrease the number of parameters

What is the main advantage of using LSTM (Long Short-Term Memory) in RNNs?

• Faster training speed

• Avoidance of the vanishing gradient problem

• Lower computational complexity

• Simplicity of architecture

Which RNN architecture is known for its simplified design compared to LSTM but still addressing the vanishing gradient problem?

• Seq2Seq

• GRU

• Unfolded RNN

• Encoder-Decoder

In an RNN, what is the role of the encoder in the Encoder-Decoder architecture?

• To process input sequences and produce a fixed-size context vector

• To generate output sequences

• To apply regularization to the model

• To unfold the network for visualization

What problem does the use of traditional RNNs face when processing long sequences?

• Overfitting

• Vanishing gradient problem

• Underfitting

• Exploding gradient problem

What is a potential limitation of the unfolded RNN architecture?

• Difficulty in visualizing the network

• Increased computational complexity

• Limited sequence length for training

• Inability to capture long-range dependencies

What issue does the Encoder-Decoder architecture in RNNs aim to address in sequence-to-sequence tasks?

• Overfitting

• Vanishing gradient problem

• Inability to process sequences

• Mismatched input-output lengths

The attention mechanisms in the Encoder-Decoder architecture of RNNs. How do attention mechanisms work, and what benefits do they provide in sequence-to-sequence tasks?

• Attention mechanisms focus on specific parts of the input sequence during decoding, allowing the model to weigh the importance of different elements. This enhances the model's ability to handle long sequences and aligns input-output pairs effectively.

• Attention mechanisms are optional components in RNNs, primarily used for regularization. They adjust the learning rate during training to avoid overfitting. The main benefit is the reduction of computational complexity.

• Attention mechanisms address issues related to weight initialization in the Encoder-Decoder architecture. They ensure that weights are evenly distributed across the network. The primary benefit is increased model interpretability.

• Attention mechanisms are redundant in RNNs, as the model naturally focuses on all parts of the input sequence during decoding. Their inclusion does not significantly impact the performance of sequence-to-sequence tasks.

What is the primary purpose of Seq2Seq RNNs (Sequence-to-Sequence RNNs)?

• • Generating random sequences

• • Converting sequences from one domain to another, such as language translation

• • Exclusively handling image data

• • Performing linear operations