Polysemous codes

TL;DR

This paper introduces polysemous codes for approximate nearest neighbor search that combine the accuracy of product quantization with the efficiency of binary code comparison, achieving state-of-the-art results on large-scale datasets.

Contribution

The paper proposes polysemous codes, a novel coding scheme that enables fast and accurate approximate nearest neighbor search by combining Hamming and asymmetric distance comparisons.

Findings

Achieved state-of-the-art query times below 0.3 ms on BIGANN dataset.

Enabled approximate k-NN graph computation for 100 million images in under 8 hours.

Demonstrated compatibility with coarse feature space partitioning methods.

Abstract

This paper considers the problem of approximate nearest neighbor search in the compressed domain. We introduce polysemous codes, which offer both the distance estimation quality of product quantization and the efficient comparison of binary codes with Hamming distance. Their design is inspired by algorithms introduced in the 90's to construct channel-optimized vector quantizers. At search time, this dual interpretation accelerates the search. Most of the indexed vectors are filtered out with Hamming distance, letting only a fraction of the vectors to be ranked with an asymmetric distance estimator. The method is complementary with a coarse partitioning of the feature space such as the inverted multi-index. This is shown by our experiments performed on several public benchmarks such as the BIGANN dataset comprising one billion vectors, for which we report state-of-the-art results for query times below 0.3\,millisecond per core. Last but not least, our approach allows the approximate computation of the k-NN graph associated with the Yahoo Flickr Creative Commons 100M, described by CNN image descriptors, in less than 8 hours on a single machine.