BEL-483: Deep Learning

Course Schedule

• Instructor: Jitin Singla (email: jsingla@bt.iitr.ac.in)

• Lectures:

  • • Time: TBD

    • Venue: TBD

• Tutorial:

  • • Time: TBD

    • Venue: TBD

• Office Hours:

  • • Time: TBD

    • Venue: TBD

• Teams:

• Piazza:

Announcements

• 15-July-2026: Course Announcements will be posted here regularly. Email notifications will only be sent if information is urgent

Course Objectives

This course serves as an introduction to deep learning, providing students with a solid foundation in one of the most transformative areas of machine learning.

• Establish Mathematical Foundations: Gain a comprehensive understanding of the core mathematical principles that underpin deep learning models and algorithms.

• Explore Key Architectures: Study widely adopted techniques and neural network architectures that have become standard in applications such as image analysis, language modeling, and beyond.

• Survey Recent Advances: Examine some of the latest breakthroughs and trends in deep learning, setting the stage for further study and research in advanced topics.

Pre-requisite

1. Make sure you are fundamental understanding in Linear Algebra and Probability. Below are reference material to revise the basics. Review Content: Linear Algebra and Probability

2. Python Programming

3. Foundational course in AI or ML

Loss functions: Mean Square Error, Negative Log Likelihood, Cross Entropy

Optimization techniques: Gradient Descent, Stochastic Gradient Descent, Constrained convex optimization (Lagrangians)

Supervised Techniques: Linear Regression, Logistic Regression, Perceptron Classification, Decision Trees, Random Forests, SVMs, Naive Bayes, k-NN

Unsupervised Techniques: K-means Clustering, Gaussian Mixture Modeling (GMMs), Expectation-Maximization (EM), Principal Component Analysis (PCA)

4. Chapters 1-5 of book Deep Learning by Ian Goodfellow

Course Content

• Introduction and Motivation

• Multi-layered Perceptrons, Neural Networks, Backpropagation

• Loss Functions, Regularization: L1-L2 Norms

• Optimization (Stochastic Gradient Descent, RMSProp, Adam, Adagrad)

• Dropouts, Batch and Layer Normalization

• Convolutional Networks (CNNs), Residual Networks

• Recurrent Neural Networks (RNNs), LSTMs

• Attention: Bahdanau Attention, Transformers

• Word2Vec Embeddings

• Understanding Language Modeling, BERT

• Improving Language Understanding by Generative Pre-Training (GPT 1)

• Language Models as Unsupervised Multitask Learners (GPT 2)

• Language Models are Few-Shot Learners (GPT 3)

• Other LLMs: LLaMa, Minstral, Gemini, Deepseek

• Vision Language Models: VIT, CLIP

Assignments

• Python is the default programming languages for the course.

• Submit via Moodle or GitHub—- as specified in each assignment.

• Honor Code: Any cases of copying will be awarded a zero on the assignment. More severe penalties may follow.

• Late submissions will incur penalties, as announced with assignment.

References and Resources

Book: Deep Learning by Ian Goodfellow

• Chapter 6:   Multi-layered Perceptrons, Backpropagation

• Chapter 7:   Regularization - L1/L2, Other Techniques

• Chapter 8:   Optimization Techniques. Normalization

• Chapter 9:   Deep Learning for Vision - Basic Models (CNNs)

• Chapter 10:  Deep Learning for NLP - Basic Models (RNNs, LSTM)

Papers:

• Batch Normalization

• AlexNet

• Residual Networks

• LSTMs

• Bahdanau Attention

• Transformers

• Word2Vec 

• BERT

• GPT-1 GPT-2 GPT-3

Evaluation Policy (tentative)

• Continuous Assessment (CWS): 30%

  • • Announced & Surprise Quizzes

    • Assignments

• Mid-Term Exam (MTE): 30%

• End-Term Exam (ETE): 40%