Vector Similarity Search and Faiss Course

Learn vector similarity search fundamentals and master Faiss implementation through comprehensive hands-on practice. Explore traditional similarity search methods including Jaccard similarity, w-shingling, and Levenshtein distance before advancing to vector-based approaches using TF-IDF, BM25, and SBERT. Discover how to choose appropriate indexes for similarity search applications and implement Locality Sensitive Hashing (LSH) with shingling and MinHashing techniques. Master LSH random projection methods and utilize IndexLSH for fast similarity search in Faiss. Understand Product Quantization for vector compression and implement IVFPQ indexes for efficient large-scale search. Explore Hierarchical Navigable Small World (HNSW) algorithms and their implementation in Faiss for high-performance vector search. Learn evaluation measures for search and recommender systems, implement metadata filtering techniques, and discover the latest filtering technologies. Master composite indexes using the Faiss Index Factory and identify the best indexes for different similarity search scenarios in production environments.

Syllabus

3 Traditional Methods for Similarity Search (Jaccard, w-shingling, Levenshtein)
3 Vector-based Methods for Similarity Search (TF-IDF, BM25, SBERT)
Faiss - Introduction to Similarity Search
Choosing Indexes for Similarity Search (Faiss in Python)
Locality Sensitive Hashing (LSH) for Search with Shingling + MinHashing (Python)
How LSH Random Projection works in search (+Python)
IndexLSH for Fast Similarity Search in Faiss
Product Quantization for Vector Similarity Search (+ Python)
Faiss - Vector Compression with PQ and IVFPQ (in Python)
HNSW for Vector Search Explained and Implemented with Faiss (Python)
Evaluation Measures for Search and Recommender Systems
Metadata Filtering for Vector Search + Latest Filter Tech
Composite Indexes and the Faiss Index Factory
Best Indexes for Similarity Search in Faiss