DBATU · Third Year B.Tech Artificial Intelligence & Allied · NEP 2020
Vector Databases
26AFAIPE507 · Elective I · Semester V · 2 credits · Continuous assessment 20 + mid-semester 20 + end-semester 60
One of the Elective I options.
Unit 1: Fundamentals of Vector Databases and Embeddings
5 hoursTraditional versus vector databases; structured and unstructured data; vector representations, feature vectors and embeddings; similarity search and the Euclidean, cosine and Manhattan metrics; uses in recommendation, semantic search, image retrieval and NLP.
Examined in the 60-mark end-semester paper, and in the mid-semester exam if it is taught by then
- Traditional versus vector databases
- Feature vectors and embeddings
- Similarity search and distance metrics
- Applications of vector databases
Unit 2: Vector Embeddings and Data Preparation
5 hoursEmbedding models for text, sentences, words and images; feature extraction and preprocessing; dimensionality reduction; metadata management and vector storage; generating embeddings with pre-trained and transformer-based models.
Examined in the 60-mark end-semester paper, and in the mid-semester exam if it is taught by then
- Text, word and image embeddings
- Preprocessing data for vector databases
- Dimensionality reduction for embeddings
- Generating embeddings with pre-trained models
Unit 3: Vector Indexing and Similarity Search
4 hoursExact versus approximate nearest-neighbour search; brute force, HNSW, inverted file indexes and product quantisation; recall, precision and query processing; filtering and hybrid search; performance and scalability.
Examined in the 60-mark end-semester paper, and in the mid-semester exam if it is taught by then
- Exact and approximate nearest-neighbour search
- HNSW indexes
- Inverted file indexes and product quantisation
- Filtering and hybrid search
Unit 4: Vector Database Systems and Applications
5 hoursVector database architecture, ingestion, storage and query pipelines; popular vector databases; similarity-search workflows and metadata filtering; real-time search, recommendation, semantic search, multimedia retrieval, fraud detection and knowledge retrieval.
Examined in the 60-mark end-semester paper, and in the mid-semester exam if it is taught by then
- Vector database architecture
- Popular vector databases
- Semantic search and recommendation systems
Unit 5: Retrieval-Augmented Generation and Trends
5 hoursRAG and knowledge retrieval with LLMs; building a simple RAG pipeline with chunking, embedding retrieval and context generation; evaluating retrieval; challenges, security, privacy and scaling of vector search; future trends.
Examined in the 60-mark end-semester paper, and in the mid-semester exam if it is taught by then
- Retrieval-augmented generation
- Building a RAG pipeline
- Document chunking strategies
- Evaluating retrieval systems
Unit contents are summarised from the official DBATU syllabus . AiCanCode.org is not affiliated with Dr. Babasaheb Ambedkar Technological University (DBATU).