AKTU Deep Learning (KDS078) Question Paper 2024-25

Note: Attempt all Sections. In case of any missing data; choose suitably.

Section A

Q1. Attempt all questions in brief · 2 × 10 = 20 marks

1. Discuss the differences between a perceptron and a support vector machine in 1 K terms of learning.
2. Define and explain the role of a loss function in neural networks.
3. Discuss the role of convolutional layers in Convolutional Neural Networks.
4. Compare deep and shallow networks in terms of representational capacity and computational complexity.
5. Discuss how batch normalization impacts the training of convolutional neural networks.
6. How autoencoders can be used to learn low-dimensional representations of data?
7. Discuss the role of non-convex optimization in deep learning. 4 K
8. Analyze the challenges of stochastic optimization in deep learning. 4 K
9. How deep learning models have revolutionized computer vision tasks?
10. What are the key challenges in modeling audio signals?

Section B

Q2. Attempt any three of the following · 10 × 3 = 20 marks

1. Explain the mathematical foundation of the Support Vector Machine (SVM) algorithm and describe its kernel trick for non-linear classification problems. How does it differ from logistic regression?
2. Derive and explain the role of batch normalization in accelerating deep neural 2 K network training. How does it help in reducing internal covariate shifts?
3. Describe the process of hyperparameter optimization in training a ConvNet. What are the key hyperparameters, and how do they influence network performance?
4. Compare and contrast LSTMs and traditional recurrent neural networks 4 K (RNNs). How do LSTMs mitigate the vanishing gradient problem in sequence modeling?
5. Analyze the applications of deep learning in bioinformatics. Provide examples of how neural networks are used for tasks like protein structure prediction.

Section C

Q3

Attempt any one part of the following · 10 × 1 = 10 marks

1. Explain the concept of stochastic gradient descent (SGD). How does it differ from batch and mini-batch gradient descent in terms of computational efficiency and convergence?
2. Compare and contrast the theoretical underpinnings of perceptrons and logistic regression in the context of binary classification problems.

Q4

Attempt any one part of the following · 10 × 1 = 10 marks

1. Explain the architecture and training process of a Generative Adversarial 2 K Network. What challenges arise during training, and how can they be mitigated?
2. Discuss the probabilistic theory behind deep learning models. How do 2 K Bayesian principles apply to the optimization and uncertainty estimation in deep networks?

Q5

Attempt any one part of the following · 10 × 1 = 10 marks

1. Compare Principal Component Analysis and Linear Discriminant Analysis terms of their objectives and mathematical formulations. Provide examples of their use in classification tasks.
2. Derive the reconstruction loss of an autoencoder. How does this loss relate to the dimensionality reduction capabilities of the network?

Q6

Attempt any one part of the following · 10 × 1 = 10 marks

1. Explain how deep reinforcement learning combines reinforcement learning principles with deep network architectures.
2. Derive the architecture of a recurrent neural network language model. Discuss its strengths and limitations compared to word-level RNNs.

Q7

Attempt any one part of the following · 10 × 1 = 10 marks

1. Discuss the limitations of deep learning models in real-world applications. How 5 K can interpretability and ethical considerations be integrated into these systems?
2. Explain the architecture of a deep learning model used for face recognition. 5 K How does it ensure robustness against variations in pose and illumination?

Short on time? See the most repeated AKTU Deep Learning questions across all years.

Other years

More AKTU Deep Learning papers on LearnAIFast.

This paper is reproduced for study purposes only. Copyright belongs to AKTU. Source: the question paper bank of ABES Institute of Technology, an AKTU-affiliated college. If you own the copyright and want it removed, email learnwithskillnest@gmail.com and we will take it down.