RTMNU · Third Year B.Tech Information Technology · NEP 2020
Generative AI
BCSE6T24.3 · Elective III · Semester VI · 3 credits · College 30 + university 70
One of the Elective III options.
Unit 1: Generative AI and Autoencoders
7 hoursAn introduction to generative AI; autoencoders — representational power, layer size and depth; undercomplete, denoising, contractive and variational autoencoders; case study: autoencoders for dimension reduction.
Examined in the 70-mark end-semester paper
- Evolution of generative AI
- Undercomplete autoencoders
- Sparse and denoising autoencoders
- Contractive autoencoders
- Variational autoencoders
- Applications of autoencoders
Unit 2: GANs, Autoregressive Models and Transformers
8 hoursGenerative adversarial networks — structure and training, deep convolutional GANs; autoregressive models with finite memory and long-range memory through RNNs and CNNs; transformers — encoders, decoders and scaling laws; case study: a GAN-aided intrusion-detection system.
Examined in the 70-mark end-semester paper
- Generative adversarial networks
- Deep convolutional GANs
- Autoregressive models: RNN- and CNN-based
- Transformer architecture basics
Unit 3: Structured Probabilistic Models
7 hoursIssues with unstructured models; directed and undirected graphs to describe models; the partition function, separation and d-separation, converting between graphs and sampling from graphical models; case study: the restricted Boltzmann machine.
Examined in the 70-mark end-semester paper
- Structured probabilistic models: directed and undirected graphs
- Boltzmann machine and Boltzmann learning
Unit 4: Generative AI Applications
7 hoursGenerative AI in art and creativity, image and video generation, text generation, music composition, healthcare and finance; real-world use cases and the challenges of deploying generative models.
Examined in the 70-mark end-semester paper
- Applications of generative AI
- Text, image, audio and video generation
Unit 5: Large Language Models and Foundation Models
7 hoursFoundation models and the architecture of modern LLMs (GPT, BERT); pre-training versus fine-tuning and transfer learning; prompt engineering — prompt design, zero-, one- and few-shot prompting and chain-of-thought prompting.
Examined in the 70-mark end-semester paper
- Foundation models and large language models
- Training, fine-tuning and inference
- Transfer learning and domain adaptation
- Prompt design principles
- Zero-shot, one-shot and few-shot prompting
- Chain-of-thought and role-based prompting
Unit contents are summarised from the official RTMNU syllabus . AiCanCode.org is not affiliated with Rashtrasant Tukadoji Maharaj Nagpur University (RTMNU).