Deep Learning (DL) SPPU Insem Question Paper Aug 2025
All 4 questions from this SPPU insem paper, as printed. Read them here or open the PDF.
- Subject code
- 414443
- Pattern
- 2019 pattern
- Exam
- Insem
- Session
- Aug 2025
- Max. marks
- 30
- Time
- 1 Hour
- Semester
- VII
- Paper code
- [6580]-602
Printed for: B.E. (Information Technology) (Insem)
Instructions to the candidates
- Answer Q.1 or Q.2, Q.3 or Q.4
- Assume suitable data, jf Necessary.
- Neat diagrams must be drawn wherever necessary.
- Figures to the right indicate full marks.
Questions
Q1
- a)[5]
Draw & explain diagram of an artificial neuron in a multilayer perceptron neural network. Write mathematical expressions for all activation functions.
- b)[5]
Explain in detail the concept of mini-batch Differentiate between SGD and mini-batch .
- c)[5]
Why regularization is required? Differentiate between L1 and L2 regularization.
OR- a)
Q2
- a)[5]
Define the term hyperparameter and list all hyperparameters and their usage range of values with units.
- b)[5]
What is a gradient? Explain mathematically the conditions for vanishing or exploding gradient?
- c)[5]
List commonly used optimization algorithms. Explain any one in detail with suitable diagrams. 2
- a)
Q3
- a)[10]
Explain layered architecture of Alex-Net with relevant details.
- b)[5]
This part has equations that don’t show as text. Open the PDF to read it.
OR- a)
Q4
- a)[5]
What is parameter sharing related to CNN? How is it done?
- b)[5]
An input image has been converted into a matrix of size 16 × 16 along with a filter of size 3 × 3 with a Stride of 1. Determine the size of the convoluted matrix.
- c)[5]
This part has equations that don’t show as text. Open the PDF to read it.
- a)