Course Course Course L T P C
21CSE424T DEEP LEARNING FOR DATA ANALYTICS E PROFESSIONAL ELECTIVE
Code Name Category 2 1 0 3
Pre-requisite Co- requisite Progressive
Nil Nil Nil
Courses Courses Courses
Course Offering Department School of Computing Data Book / Codes / Standards Nil
Course Learning Rationale (CLR): The purpose of learning this course is to: Program Outcomes (PO) Program
Specific
CLR-1: understand the theoretical foundations, algorithms, and methodologies of convolutional neural networks 1 2 3 4 5 6 7 8 9 10 11 12 Outcomes
CLR-2: identify and apply appropriate recurrent neural networks for analyzing the data for variety of problems
Individual & Team Work
Engineering Knowledge
Design/development of
Project Mgt. & Finance
Conduct investigations
of complex problems
CLR-3: understand the principles and applications of computer vision
Modern Tool Usage
Life Long Learning
The engineer and
Problem Analysis
Communication
CLR-4: construct Generative Adversarial Networks to solve real - world problems
Environment &
Sustainability
CLR-5: understand different methodologies to create application using deep nets
solutions
society
PSO-1
PSO-2
PSO-3
Ethics
Course Outcomes (CO): At the end of this course, learners will be able to:
CO-1: recognize the characteristics of convolutional neural networks and the use of optimizer - - - - - - - - - - - - 1 - -
CO-2: analyze various recurrent neural networks models - - - - - - - - - - - - 2 - -
CO-3: apply the Deep Learning models for Computer Vision - - - - - - - - - - - - - 2 -
CO-4: solve various real - world problems using Generative Adversarial Networks - - - - - - - - - - - - - - 2
CO-5: build a Deep Reinforcement Learning models for solving various problems - - - - - - - - - - - - - - 1
Unit-1 - Convolutional Neural Networks 9 Hour
Introduction to Convolutional neural networks – Autoencoders: types, denoising, Convolutional autoencoder example, Uses and limitations – Variational autoencoders: Parameters to define a normal distribution,
VAE loss function, Kullback-Leibler divergence, Training the VAE, The reparameterization trick – Applications of autoencoders – Optimizers: Gradient Descent, RMSProp, Adam
Tutorial:
T1. To build a convolutional autoencoder model for the MNIST dataset, T2. To build a convolutional variational autoencoders for the MNIST dataset
Experiment with different learning rates and objective functions for gradient descent
Unit-2 - Recurrent Neural Networks 9 Hour
Recurrent Neural Networks: Overview, Neural Networks with and without Hidden States, RNN-based Character-Level Language Models - Implementation: RNN Model, RNN-based Language Model, Gradient
Clipping, Training, Decoding - Long Short-Term Memory (LSTM), Gated Recurrent Units (GRU)
Tutorial:
T3. To implement the RNN model using the H. G. Wells’ The Time Machine data set. T4. To implement the LSTM model using the H. G. Wells’ The Time Machine data set.
To implement the GRU model using the H. G. Wells’ The Time Machine data set.
Unit-3 - Deep Learning for Computer Vision 9 Hour
Image Augmentation - Object Detection and Bounding Boxes - Anchor Boxes - Multiscale Object Detection - Region-based CNNs: R-CNNs, Fast R-CNN, Faster R-CNN, Mask R-CNN
Tutorial:
T5. To demonstrate object detection model banana detection dataset. T6. To implement the classification model for CIFAR-10 dataset.
To implement the classification model for Dog Breed Identification dataset
485
B.Tech / M.Tech (Integrated) Programmes-Regulations 2021-Volume-11-CSE-Higher Semester Syllabi-Control Copy
Unit-4 - Generative Adversarial Networks 9 Hour
Generative Adversarial Networks: Generator, Discriminator, Loss function, Generator loss, Discriminator loss, Training - Deep Convolutional Generative Adversarial Networks: Wasserstein GAN, BEGAN, CycleGAN
- Conditional GANs: Pix2Pix.
Tutorial:
T7. To build a model using GAN to resemble MNIST digits. , T8. To implement a Deep Convolutional GAN to generate complex color images.
To implement a Deep Convolutional GAN on Fashion-MNIST data set using ReLU as activation function for generator, leaky ReLU as activation function for discriminator.
Unit-5 - Deep Reinforcement Learning 9 Hour
Stateless Algorithms: Na¨ıve Algorithm, ε-Greedy Algorithm, Upper Bounding Methods - The Basic Framework of Reinforcement Learning: Challenges, Simple Reinforcement Learning for Tic-Tac-Toe, Role of Deep
Learning, and a Straw-Man Algorithm
Tutorial:
T9. Self-Learning Robots - Case Study, T10. Building Conversational Systems: Deep Learning for Chatbots - Case Study
Self-Driving Cars - Case Study
1. Iffat Zafar, Giounona Tzanidou, Richard Burton, Nimesh Patel, Leonardo Araujo, Hands-On Convolutional 4. Ian Goodfellow, Yoshua Bengio and Aaron Courville, Deep Learning, the MIT
Neural Networks with TensorFlow, Packt Publishing, 2018. (Unit I) Press, 2017.
Learning
2. Aston Zhang, Zack C. Lipton, Mu Li, Alex J. Smola, Dive into Deep Learning, arXiv preprint, 2021 5. Umberto Michelucci, Applied Deep Learning – A Case-Based Approach to
Resources
(http://d2l.ai/index.html) (Unit II, III & IV) Understanding Deep Neural Networks, Apress, 2018.
3. Charu C. Aggarwal, Neural Networks and Deep Learning - A Text Book, Springer Nature, 2018. (Unit V) 6. François Chollet, Deep Learning with Python, Manning Publications, 2018.
Learning Assessment
Continuous Learning Assessment (CLA)
Summative
Formative Life-Long Learning
Bloom’s Final Examination
CLA-1 Average of unit test CLA-2
Level of Thinking (40% weightage)
(50%) (10%)
Theory Practice Theory Practice Theory Practice
Level 1 Remember 15% - 15% - 15% -
Level 2 Understand 25% - 20% - 25% -
Level 3 Apply 30% - 25% - 30% -
Level 4 Analyze 30% - 25% - 30% -
Level 5 Evaluate - - 10% - - -
Level 6 Create - - 5% - - -
Total 100 % 100 % 100 %
Course Designers
Experts from Industry Experts from Higher Technical Institutions Internal Experts
1. Dr. M. Prakash, SRMIST
2. Dr.G.Vadivu, SRMIST
486
B.Tech / M.Tech (Integrated) Programmes-Regulations 2021-Volume-11-CSE-Higher Semester Syllabi-Control Copy
