Fast Search of Massive High-Dimensional Vectors Similarity

摘要

It is common to perform similarity lookups of massive high-dimensional vectors in the fields of computer vision, natural language processing, big data and artificial intelligence. However, in actual business scenarios, it is not easy to calculate the results accurately, efficiently and quickly due to its large database. We use approximate nearest neighbor search based on product quantization and optimize the computer system to achieve fast lookups on the premise of making full use of memory.