Deep Learning CS-641

Deep Learning (CS-641)

Course Type: Programme Elective
Batch: 1st Year M.Tech-AI & Ph.D
Course Credits: 04

To introduce the principles of deep learning and its applications.
To enable the students in practical skills to design, implement, and train practical deep learning systems.
To provide a structured approach covering Fundamentals of Machine Learning, Neural Networks, Modern Deep Learning, and Applications and other advanced topics.

Probability, Statistics, Variable Calculus and Linear Algebra

Conference Hall, Second Floor, Department of Computer Science and Engineering

Monday: 10:00 AM - 11:00 AM
Wednesday: 10:00 AM - 11:00 AM
Thursday: 10:00 AM - 11:00 AM
Friday: 10:00 AM - 11:00 AM

Introduction to Deep Learning: Overview of neural networks and deep learning, Historical perspective and key milestones in deep learning, Deep learning applications in computer vision, natural language processing, and reinforcement learning.
Deep Neural Networks: Early Models, Perceptron Learning, Multilayer Perceptron (MLPs) and feedforward neural networks, Backpropagation, Initialization, Training & Validation.
Parameter Estimation: MLE, MAP, Bayesian Estimation, Activation functions, loss functions, and optimization techniques, Regularization methods and hyperparameter tuning.
Convolutional Neural Networks (CNNs): Architecture of CNNs and convolutional layers, Pooling layers, batch normalization, and dropout, Applications of CNNs in image recognition and computer vision tasks.
Recurrent Neural Networks (RNNs): Introduction to RNNs and Long Short-Term Memory (LSTM) networks, Sequence modeling, text generation, and time series prediction, Attention mechanisms and Transformer models.
Generative Adversarial Networks (GANs): Fundamentals of GANs and adversarial training, Conditional GANs, StyleGAN, and CycleGAN, Applications of GANs in image generation, style transfer, and data augmentation.
Advanced Topics in Deep Learning: Transfer learning and domain adaptation, Autoencoders, Variational Autoencoders (VAEs), and unsupervised learning, Reinforcement learning with deep neural networks, Transformers, LLMs.

Upon successful completion of the course, the students will be able to:

CO1: Demonstrate the ability to apply concepts from linear algebra, probability, optimization, and machine learning to solve complex problems in the context of deep learning.
CO2: Evaluate the advantages and disadvantages of deep learning neural network architectures and compare them with other approaches in the context of case studies.
CO3: Design, implement, and train deep learning models using convolutional, recurrent, and other neural network architectures to address real-world problems effectively.
CO4: Analyze, design, and implement solutions to real-world computer vision problems, NLP and other problems using deep learning techniques.

Deep Learning by Ian Goodfellow, Yoshua Bengio, and Aaron Courville, MIT Press.
Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow by Aurélien Géron, O’Reilly Media.
Neural Networks and Deep Learning by Michael Nielsen, Determination Press.
Deep Learning for Computer Vision by Rajalingappaa Shanmugamani, Packt Publishing.
Generative Deep Learning by David Foster, O'Reilly Media.