Revisiting the Inverted Indices for Billion-Scale Approximate...
neighbor. [2] also describes themulti-sequenceprocedure that produces the sequence of regions that are the closest to the particular query. For dataset compression, [2] also uses Product Quantization with codeb