Department of Computer Science & Big Data Analytics
This master archive consolidates the comprehensive curriculum, lecture notes, coding tutorials, student projects showcases, and end-semester examinations across academic iterations of CS411 for PMRF teaching work. Commencing with foundational mathematical optimization and PyTorch programming, the course progresses toward practical implementation of cutting-edge deep learning architectures, including Variational Autoencoders (VAEs), Convolutional Neural Networks (CNNs), Vision Transformers, and Large Language Models (LLMs).
Academic Year Archives
Course EditionsGenerative Vision, Transformers & LLMs
Comprehensive archives covering PyTorch pipelines, Softmax & Cross-Entropy derivations, VAEs, Deep NLP, Vision Transformers, LLM Training & Tuning, LLM workflows, and specialized guest research seminars on Diffusion Frameworks (GENIE) and Domain Generalization (HiDISC).
Advanced Sampling, VAEs & Image/Video Processing
Extensive coverage of sampling techniques (Rejection, Gibbs, Inverse Transform, Langevin, MALA, Metropolis-Hastings), parameter calculations, VAE theory via EM, and computer vision segmentation methods. Features guest talks on Open-Vocabulary Segmentation (Saikat Dutta), Adaptive Task-Arithmetic (Vaibhav Rathore), and Multimodal Learning (Aniket Thomas).
Neural Network Foundations, NLP & Segmentation Pipelines
The foundational online course offering covering PyTorch autograd, classification pipelines, U-Net semantic segmentation (Skin Lesion), VAE theory, and Transformers/Attention. Includes specialized lectures on NLP, Word2Vec, and Neural Machine Translation by guest speaker Seshadri Mazumder.
Project Showcase
ACVDL Student Projects & Applications Gallery
Explore end-to-end computer vision applications, empirical evaluations, and custom deep learning pipelines engineered and deployed by graduate and undergraduate students.
Special Interest Groups (SIG)
Competitive Coding & Problem Solving with Python (CCP-SIG)
Organized to maintain accountability and consistency in algorithmic mastery. The group meets biweekly to analyze optimal time complexities and implement solutions in rapid-prototyping Python. Focuses heavily on LeetCode Data Structures & Algorithms (Arrays, Two-Pointers, Subarrays, Linked Lists) before transitioning to competitive programming challenges on Codeforces.