Data Science Deep Learning and Neural Networks 2

Question 1

Which type of neural network architecture is specifically designed to handle sequential data such as time series or text?

Accepted Answer

Recurrent Neural Network (RNN)

Answer

Convolutional Neural Network (CNN)

Answer

Autoencoder

Answer

Generative Adversarial Network (GAN)

Question 2

What is the key innovation of Long Short-Term Memory (LSTM) networks compared to standard RNNs?

Accepted Answer

They include gating mechanisms to control information flow and address long-range dependencies

Answer

They use convolutional layers for faster computation

Answer

They require no backpropagation

Answer

They replace the hidden state with attention scores

Question 3

In the context of deep learning, what does 'transfer learning' mean?

Accepted Answer

Using a pre-trained model's learned representations as a starting point for a new task

Answer

Moving a trained model to a different hardware platform

Answer

Transferring data between training and test sets

Answer

Sharing weights between parallel training jobs

Question 4

What is the purpose of batch normalization in a neural network?

Accepted Answer

To normalize layer inputs to reduce internal covariate shift and speed up training

Answer

To increase the number of trainable parameters

Answer

To randomly drop batches during training

Answer

To apply L2 regularization to each batch

Question 5

Which loss function is most appropriate for a multi-class classification problem with a softmax output layer?

Accepted Answer

Categorical Cross-Entropy

Answer

Mean Squared Error

Answer

Binary Cross-Entropy

Answer

Hinge Loss

Question 6

What does a Generative Adversarial Network (GAN) consist of?

Accepted Answer

A generator and a discriminator trained in opposition

Answer

An encoder and a decoder

Answer

A policy network and a value network

Answer

A feature extractor and a classifier