AL-503 (B) RGPV B.Tech
Deep Learning (DL)
Previous year papers and the official syllabus are free to read, no account needed.
Previous Year Papers
5 papers, latest Dec 2025
- AL-503 (B) paper, Dec 2025
- AL-503 (B) paper, Jun 2025
- AL-503 (B) paper, Dec 2024
- AL-503 (B) paper, Nov 2023
- AL-503 (B) paper, Nov 2022
Syllabus, notes and important questions for AL-503 (B)
Syllabus units
Topics as the official RGPV syllabus lists them.
Unit 1 Introduction 12 topics
- Introduction and History of Deep Learning
- McCulloch–Pitts neuron model
- Multilayer Perceptrons (MLPs)
- Representation power of MLPs
- Sigmoid neurons
- Feed‑forward neural networks
- Back‑propagation algorithm
- Weight initialization methods
- Batch normalization
- Representation learning
- GPU implementation of neural networks
- Dimensionality reduction – Principal Component Analysis (PCA) and Singular Value Decomposition (SVD)
Unit 2 Deep Feed‑Forward Networks and Optimization 16 topics
- Deep feed‑forward neural networks
- Gradient Descent (GD)
- Momentum‑based GD
- Nesterov accelerated GD
- Stochastic GD
- AdaGrad optimizer
- Adam optimizer
- RMSProp optimizer
- Auto‑encoders
- Regularization in auto‑encoders
- Denoising auto‑encoders
- Sparse auto‑encoders
- Contractive auto‑encoders
- Variational auto‑encoder (VAE)
- Relationship between auto‑encoders and PCA/SVD
- Dataset augmentation techniques
Unit 3 Convolutional Neural Networks 8 topics
- Introduction to Convolutional Neural Networks (CNN) and architectures
- Convolutional terminology – ReLU activation, stride, padding, pooling, convolution operations
- Convolutional kernels
- Layer types – convolutional, pooling, fully‑connected
- Visualization of CNNs
- Canonical CNN models – LeNet, AlexNet, ZF‑Net, VGGNet, GoogLeNet, ResNet, R‑CNN
- Deep Dream and Deep Art applications
- Regularization techniques – dropout, drop‑connect, unit pruning, stochastic pooling, artificial data injection, early stopping, weight decay, limiting the number of parameters
Unit 4 Recurrent Neural Networks 11 topics
- Introduction to deep Recurrent Neural Networks (RNN) and architectures
- Back‑Propagation Through Time (BPTT)
- Vanishing and exploding gradient problems
- Truncated BPTT
- Gated Recurrent Units (GRUs)
- Long Short‑Term Memory (LSTM) networks
- Mitigating vanishing gradients with LSTMs
- Encoder–decoder frameworks for RNNs
- Attention mechanisms – standard attention, visual attention, hierarchical attention
- Directed graphical models
- Applications of deep RNNs in image processing, natural language processing, speech recognition, and video analytics
Unit 5 Generative Models 7 topics
- Introduction to deep generative models
- Restricted Boltzmann Machines (RBMs) and Gibbs sampling
- Deep belief networks
- Markov networks and Markov chains
- Autoregressive models – NADE, MADE, PixelRNN
- Generative Adversarial Networks (GANs)
- Applications of deep generative models in object detection, speech and image recognition, video analysis, NLP, and medical science
Exam insights
Counted from 5 past AL-503 (B) papers.
Marks asked per paper, by unit
-
Unit 1 Introduction 22 marks
-
Unit 2 Deep Feedforward Neural Networks 18 marks
-
Unit 3 Introduction to Convolutional neural Networks (CNN) and its architectures 22 marks
-
Unit 4 Introduction to Deep Recurrent Neural Networks and its architectures 22 marks
-
Unit 5 Introduction to Deep Generative Models 13 marks
Average marks of the questions from each unit in one paper. You answer only some of them, so the total is more than the paper’s maximum.
Asked again and again
-
Discuss regularization in autoencoders, including techniques and its role/impact on training.
Asked 4 times: 2025, 2023, 2022
-
Explain the fundamental concepts and architectures of Deep Recurrent Neural Networks with a neat diagram.
Asked 3 times: 2025, 2023, 2022
-
Explain feed-forward neural networks and sigmoid neurons with suitable diagram.
Asked 2 times: 2025