Course Curriculam

Course Code: AIML203 Credit Units L T P/S SW AS/DS FW No. of PSDA Total Credit Unit
Course Level UG 3 0 2 0 0 0 0 4
Course Title Deep Neural Networks

Course
Description :

Course Objectives :

SN
Objectives
.
To introduce the fundamental set of techniques and algorithms that constitute deep learning algorithms, To understand the problem settings, and
1 various deep learning applications to solve real world problems. Understand and apply various deep learning models to solve various domain’s
problems.

Pre-Requisites : General

SN. Course Code Course Name

Course Contents / Syllabus :

SN. Module Descriptors / Topics Weightage

Module I:
• Introduction and Applications of deep learning in healthcare, time-series analysis, computer vision, social media,
Introduction to
marketing, banking and other sectors • Biological inspiration behind neural networks, Neuron model- Perceptron •
1 Deep Learning 20.00
Multilayer Perceptron (MLP) and the feedforward architecture • Activation Functions • Training neural
and its
networksGradient Descent (Stochastic and Batch Gradient Descent) • Backpropagation algorithm.
applications
• Loss functions and performance measures for regression and classification tasks • Overfitting and underfitting
Module II: Model
issues, Bias-variance trade-off • Data Augmentation, Early Stopping, Regularization- L1, L2, Dropout • Optimizers
2 Parameters 20.00
(RMSprop, Adam, etc.) • Batch Normalization • Hyperparameter tuning (learning rate, batch size, number of
Optimization
hidden layers.)
Module III:
Convolutional Basics of convolution operation, kernels, 1D, 2D, 3D convolutions • Parameter sharing, receptive field, pooling,
3 Neural Networks stride, padding. • Popular CNN architectures such as LeNet, AlexNet, VGG-Net, ResNet, DenseNet, GoogleNet • 20.00
and Transfer Transfer Learning and Fine-tuning
Learning
Module IV:
Sequential
• Introduction to RNNs and their applications • Long Short-Term Memory (LSTM) networks and Gated Recurrent
models &
4 Units (GRUs) • Sequence-to-sequence models for natural language processing • Applications of RNNs in speech 20.00
Recurrent
recognition, machine translation, and video analysis • Attention mechanism. • Transformers
Neural Networks
(RNNs)
Module V:
• Autoencoders • Variational Autoencoders (VAEs) and Generative Adversarial Networks (GANs) • GAN
Autoencoders
5 architectures (e.g., DCGAN, CycleGAN) • Image data generation with GANs • Applications of generative models in 20.00
and Generative
computer vision and data augmentation
networks

Course Learning Outcomes :

SN. Course Learning Outcomes
1 • To Understand the principles, advantages, limitations, and possible applications of popular deep learning approaches
2 • Identify the deep learning algorithms which are more appropriate for various types of learning tasks in various domains.
3 • Implement deep learning algorithms and solve real-world problems.
4 • Implement Memory based deep learning algorithms.
5 Able to formulate appropriate optimization methods for deep learning.
6 • Able to design applications of deep learning models and chatbots.

Pedagogy for Course Delivery :

SN. Pedagogy Methods

The class will be taught using theory and applications. In addition to discussing real life applications, the course instructor will spend considerable
1 time in understanding the concept. The instructor will cover the ways to think innovatively liberally using thinking techniques. Delivery of course will
be covered using E-content based on 4-Quadrant approach.

Theory /VAC / Architecture Assessment (L,T & Self Work): 75.00 Max : 100

Attendance+CE+EE : 5+35+60

SN. Type Component Name Marks

1 Attendance 5.00
2 End Term Examination (OMR) 60.00
3 Internal MID TERM EXAM 15.00
4 Internal CLASS QUIZ 10.00
5 Internal HOME ASSIGNMENT 10.00

Lab/ Practical/ Studio/Arch. Studio/ Field Work Assessment : 25.00 Max : 100

Attendance+CE+EE : 5+35+60

SN. Type Component Name Marks

1 Attendance 5.00
2 External Viva 30.00
3 External Major Experiment (Practical) 30.00
4 Internal PERFORMANCE 10.00
5 Internal Lab record including Attendance,Ethics,Precautions,Interactions 15.00
6 Internal Viva 10.00

Lab/ Practical details, if applicable :

SN
Lab / Practical Details
.
1. Build an Artificial Neural Network to implement Regression task using the Back-propagation algorithm and test the same using appropriate data
1
sets.
2. Build an Artificial Neural Network to implement Binary Classification task using the Back-propagation algorithm and test the same using appropriate
2
data sets.
 3. Build an Artificial Neural Network to implement Multi-Class Classification task using the Back-propagation algorithm and test the same using
3
appropriate data sets.
4 4. Implement an image classification task using pre-trained models like AlexNet, VGGNet, InceptionNet and ResNet and compare the results.
5. Design a CNN architecture to implement the image classification task over an image dataset. Perform the Hyper-parameter tuning and record the
5
results
6 6. Implement a deep neural network (Ex: UNet / SegNet) for Image Segmentation task.
7 7. Train a Recurrent Neural Network (RNN) on the IMDB large movie review dataset for sentiment analysis.
8. Download the tweets data from twitter and run an RNN/LSTM to classify the sentiment as Hate/ Non-hate speech. Compare the results of both the
8
models.
9 9. Implement GAN architecture on MNIST dataset to recognize the handwritten digits.

List of Professional skill development activities :

No.of PSDA : 0
SN. PSDA Point

Text & References :

SN. Type Title/Name Description ISBN/ URL

1. "Deep Learning" by Ian Goodfellow,
1 Book Yoshua Bengio, and Aaron Courville,
published by MIT Press in
2. "Hands-On Machine Learning with Scikit-
2 Book Learn, Keras, and TensorFlow" by Aurélien
Géron, published
3. Deep Learning with Python" by François
3 Book Chollet, published by Manning Publications
in 2018.
4. "Python Machine Learning" by Sebastian
4 Book Raschka and Vahid Mirjalili, published by
Packt Publishing
5. "Applied Deep Learning: A Case-Based
5 Book Approach to Understanding Deep Neural
Networks" by Umberto M
6. "Deep Learning for Computer Vision" by
6 Book Rajalingappaa Shanmugamani, published
by Packt Publishing
7. "Reinforcement Learning: An
7 Book Introduction" by Richard S. Sutton and
Andrew G. Barto, published by
8. "Deep Learning Illustrated" by Jon Krohn,
8 Book published by Addison-Wesley Professional
in 2019.
9. "Hands-On Deep Learning for Images
9 Book with TensorFlow" by Will Ballard, Luis
Capelo, and Ankit Jain,