Deep Learning PYQ 2024-25 AKTU (KCS078) Question Paper

AKTU · BTECH · Semester 7 · Deep Learning · Session 2024-25 · PYQ

AKTU Deep Learning (KCS078) previous year question paper 2024-25 for B.Tech Semester 7. Covers Introduction to Neural Networks, Deep Networks…

Open the interactive reader to study this resource on AcademicArk.

Deep Learning AKTU syllabus

  1. Unit 1: Introduction to Neural Networks
  2. Unit 2: Deep Networks
  3. Unit 3: Dimensionality Reduction
  4. Unit 4: Optimization & Generalization
  5. Unit 5: Case Studies & Applications

Questions in Deep Learning AKTU PYQ 2024-25

  1. Q1a. Discuss the differences between a perceptron and a support vector machine in terms of decision boundary formation. (2 marks, 2024-25)
  2. Q1b. Discuss how the choice of a loss function affects model performance for a classification problem. (2 marks, 2024-25)
  3. Q1c. Trace the historical development of deep learning that enabled deep networks to outperform traditional machine learning models. (2 marks, 2024-25)
  4. Q1d. How does weight sharing in CNNs improve efficiency and performance? (2 marks, 2024-25)
  5. Q1e. Explain the impact of weight initialization on the performance of deep networks. (2 marks, 2024-25)
  6. Q1f. Compare performance of autoencoders with traditional dimensionality reduction techniques. (2 marks, 2024-25)
  7. Q1g. Provide examples of optimization algorithms used for deep networks. (2 marks, 2024-25)
  8. Q1h. How do techniques address the challenges of stochastic optimization in deep learning? (2 marks, 2024-25)
  9. Q1i. Describe the architecture and training of WaveNet for audio generation. (2 marks, 2024-25)
  10. Q1j. Discuss the role of Word2Vec in natural language processing. (2 marks, 2024-25)
  11. Q2a. Derive the gradient of the logistic regression loss function with respect to its parameters. How does regularization impact its optimization process? (10 marks, 2024-25)
  12. Q2b. Define VC dimension and discuss its implications for understanding the generalization capabilities of deep networks. (10 marks, 2024-25)
  13. Q2c. Explain the role of manifold learning in dimensionality reduction. How does it compare to linear methods? (10 marks, 2024-25)
  14. Q2d. Explain the concept of generalization in neural networks. How do techniques like dropout and weight regularization improve generalization performance? (10 marks, 2024-25)
  15. Q2e. Discuss the pipeline of scene understanding using deep learning models. How does this pipeline handle challenges like occlusion and varying lighting conditions? (10 marks, 2024-25)
  16. Q3a. Prove that neural networks can act as universal function approximators. Discuss any constraints or assumptions involved in this theorem. (10 marks, 2024-25)
  17. Q3b. Discuss the relationship between the choice of activation function in a shallow neural network and its ability to model non-linear relationships. (10 marks, 2024-25)
  18. Q4a. Analyze the trade-offs between increasing network depth and the risk of vanishing/exploding gradients. What role does skip connections in ResNet play in addressing these issues? (10 marks, 2024-25)
  19. Q4b. How does semi-supervised learning differ from supervised and unsupervised learning? Discuss its application in training deep networks with limited labeled data. (10 marks, 2024-25)
  20. Q5a. Explain the concept of metric learning in the context of dimensionality reduction. How does it differ from traditional techniques like PCA (10 marks, 2024-25)
  21. Q5b. Analyze the architectural differences between AlexNet, VGG, Inception, and ResNet. How do these architectures address the vanishing gradient problem? (10 marks, 2024-25)
  22. Q6a. Discuss the implications of non-convex optimization in designing deep learning models. How do saddle points influence the training process? (10 marks, 2024-25)
  23. Q6b. Provide a detailed overview of computational neuroscience principles applied in designing artificial neural networks. (10 marks, 2024-25)
  24. Q7a. Provide a detailed overview of deep learning-based techniques for audio detection. How does WaveNet outperform traditional methods? (10 marks, 2024-25)
  25. Q7b. Describe the process of joint detection and captioning in image captioning systems. How does attention enhance this process? (10 marks, 2024-25)

AKTU paper codes: BAI701, KCS078, KDS078, KOT076

Deep Learning previous year papers

More Deep Learning resources

Browse all notes · Semester 7 notes