JAWAHARLAL NEHRU TECHNOLOGICAL UNIVERSITY HYDERABAD
JAWAHARLAL NEHRU TECHNOLOGICAL UNIVERSITY HYDERABAD B.Tech IV DEEP LEARNING Previous Question Paper (January/February - 2023 - R18)
Note: i) Question paper consists of Part A, Part B. ii) Part A is compulsory, which carries 25 marks. In Part A, Answer all questions. iii) In Part B, Answer any one question from each unit. Each question carries 10 marks and may have a, b as sub questions.
Part A
Write the heuristics for faster learning.
Answer
Heuristics for faster learning include using mini-batch gradient descent to reduce variance, employing adaptive learning rate algorithms like Adam or RMSProp, utilizing batch normalization to stabilize training, and applying weight initialization techniques such as Xavier or He initialization.
What is the significance of Gradient Descent algorithm?
Answer
Gradient Descent is the fundamental optimization algorithm used to minimize the loss function by iteratively updating model parameters in the direction of the steepest descent. Its significance lies in its ability to navigate high-dimensional error surfaces to find optimal weights, enabling the network to learn complex patterns from data.
What is pooling in CNN?
Answer
Sign in to view the complete answer paper.
What are deep recurrent networks?
Answer
Sign in to view the complete answer paper.
What is difference between object detection and image segmentation?
Answer
Sign in to view the complete answer paper.
How does LSTM work in image captioning?
Answer
Sign in to view the complete answer paper.
What are the applications in word similarity?
Answer
Sign in to view the complete answer paper.
What is word vector representation?
Answer
Sign in to view the complete answer paper.
What is named entity recognition?
Answer
Sign in to view the complete answer paper.
Why is Convolutional Neural Network better for classification?
Answer
Sign in to view the complete answer paper.
Part B
Explain the architecture of feed forward neural network.
Answer
Sign in to view the complete answer paper.
What are computational graphs? Give examples of computational graphs.
Answer
Sign in to view the complete answer paper.
Explain algebraic and graph based representations. Give example to symbol-to-symbol approach to computing derivatives.
Answer
Sign in to view the complete answer paper.
Discuss about the common hidden units and their types.
Answer
Sign in to view the complete answer paper.
Explain operation of Convolutional Neural Network.
Answer
Sign in to view the complete answer paper.
Explain Auto-encoder and DBM Attention and memory models.
Answer
Sign in to view the complete answer paper.
What are Adversarial Generative Networks? Explain their importance.
Answer
Sign in to view the complete answer paper.
List and explain different auto-encoders and their applications.
Answer
Sign in to view the complete answer paper.
Why do we prefer Convolutional Neural networks (CNN) over Artificial Neural networks (ANN) for image processing?
Answer
Sign in to view the complete answer paper.
Explain the process of image generation with generative adversarial networks.
Answer
Sign in to view the complete answer paper.
How does LSTM models work for video to text generation and classification?
Answer
Sign in to view the complete answer paper.
Describe various attention models for computer vision tasks.
Answer
Sign in to view the complete answer paper.
How is deep learning used in natural language processing?
Answer
Sign in to view the complete answer paper.
Explain continuous skip-gram model in detail.
Answer
Sign in to view the complete answer paper.
Discuss about continuous Bag-of-words model.
Answer
Sign in to view the complete answer paper.
Write a technical note on vector space model of semantic.
Answer
Sign in to view the complete answer paper.
Explain the process of Parsing with Recursive Neural Networks.
Answer
Sign in to view the complete answer paper.
What is Chatbot and how it works? Explain.
Answer
Sign in to view the complete answer paper.
How to implement CNN for sentence classification? Explain.
Answer
Sign in to view the complete answer paper.
Explain opinion mining with deep recurrent neural networks.
Answer
Sign in to view the complete answer paper.