Approximate Nearest Neighbors Search in Multidimensional Space
Neural Networks Seminar · Moscow State University · May 2023
Many media-processing applications require fast similarity search across very large datasets. This problem can be separated into two related parts: finding a feature representation that captures the meaning of an item, and building an index that retrieves similar items efficiently.
This talk discusses the second part: approximate nearest-neighbour search in multidimensional space. It introduces the main approaches known in 2023 and practical tools for working with the problem, with examples based on FAISS.