Deep Learning and Natural Language Processing Question Bank

Unit 1 Introduction to Deep Learning

1. Explain the historical development of Deep Learning and the factors responsible for its rapid growth.

2. Explain the structure and working of an Artificial Neural Network. Compare a single-layer perceptron and a multilayer perceptron.

3. Why can a single-layer perceptron not solve the XOR and XNOR classification problems? Explain how a multilayer neural network can solve them.

4. Explain feed-forward networks and deep feed-forward networks with suitable diagrams. Discuss the role of hidden units.

5. Explain stochastic gradient-based learning and backpropagation in the training of a neural network.

6. Discuss the key considerations in neural network architecture design, including the number of hidden layers, hidden units, and choice of activation functions.

Unit 2 Convolution Neural Network

1. Explain the basic architecture and working of a CNN. Discuss its applications in computer vision.

2. Explain sigmoid, tanh, ReLU, and Softmax activation functions and their roles in a CNN.

3. Explain the convolution operation and different types of pooling layers with suitable examples.

4. Describe the architectures of VGG16 and VGG19 and compare their main features.

5. Explain a residual block and the basic architecture of ResNet50. How do residual connections help train deep networks?

6. Describe the steps involved in training a CNN using TensorFlow. Explain the roles of Dropout, normalization, and data augmentation.

 

Unit 3 Recurrent Neural Network

1. Explain the architecture and working of an RNN. Discuss its applications in sequential data analysis.

2. Explain Backpropagation Through Time (BPTT) and how it is used to train an RNN.

3. Explain the vanishing gradient problem in RNNs and its effect on learning long-term dependencies.

4. What is gradient clipping? Explain how it helps control exploding gradients during RNN training.

5. Explain the architecture of an LSTM network, including its cell state and input, forget, and output gates. Compare LSTM and GRU.

6. Explain Bidirectional RNNs and Bidirectional LSTMs. Compare them with their unidirectional counterparts and discuss suitable applications.

Unit 4 Natural Language Processing

1. Define Natural Language Processing and explain its major applications or use cases with suitable examples.

2. Explain the major components of NLP and their roles in understanding and generating human language.

3. Describe the main steps in an NLP workflow, from raw text input to the required output, using a suitable example.

4. Explain words and their components. Differentiate between lexemes and morphemes with suitable examples.

5. What is morphology? Discuss the major problems encountered in morphological processing with suitable examples.

6. Explain linguistic typology and morphological typology. Describe the main morphological language types with suitable examples.

Unit 5 Natural Language Processing with Python NLTK Package

1. Explain word tokenization and sentence tokenization. Write Python code using NLTK to demonstrate both.

2. Explain stop-word filtering and stemming. Write Python code using NLTK to perform these tasks on sample text.

3. Explain lemmatization and compare it with stemming. Illustrate lemmatization using Python and NLTK.

4. What is Part-of-Speech tagging? Write Python code using NLTK to tokenize and tag a sentence.

5. Explain chunking, chinking, and Named Entity Recognition. Illustrate their purposes with suitable examples and describe how they can be performed using NLTK.

6. Explain Term Frequency and Inverse Document Frequency (TF-IDF). Calculate TF-IDF for a selected term in a small document collection and discuss its role in text representation.