Diverse Yet Efficient Retrieval using Locality Sensitive Hashing

Abstract

Typical retrieval systems have three requirements: a) Accurate retrieval i.e., the method should have high precision, b) Diverse retrieval, i.e., the obtained set of points should be diverse, c) Retrieval time should be small. However, most of the existing methods address only one or two of the above mentioned requirements. In this work, we present a method based on randomized locality sensitive hashing which tries to address all of the above requirements simultaneously. While earlier hashing approaches considered approximate retrieval to be acceptable only for the sake of efficiency, we argue that one can further exploit approximate retrieval to provide impressive trade-offs between accuracy and diversity. We extend our method to the problem of multi-label prediction, where the goal is to output a diverse and accurate set of labels for a given document in real-time. Moreover, we introduce a new notion to simultaneously evaluate a method's performance for both the precision and diversity measures. Finally, we present empirical results on several different retrieval tasks and show that our method retrieves diverse and accurate images/labels while ensuring 100x-speed-up over the existing diverse retrieval approaches.

Unsupervised Rank-Preserving Hashing for Large-Scale Image Retrieval

“…Many approaches have been proposed to learn to produce binary codes [6,26,12,32,33,11,29,40] in past decades. In this section, we are going to review the rank-preserving [36,35,41,8], and reconstruction-related hashing methods [31,23,5,15,16,10,20,19,8,9], which are the most relevant to our proposed method.…”

Blackthorn: Large-Scale Interactive Multimodal Learning

“…LSH is a well-known scalar-based indexing technique [4]; it has been successfully applied to many contexts (e.g., see [30], [37]). The NV-tree is another scalar-based method, which has been shown to significantly outperform LSH [24]; the NV-tree only requires one eighth of of the processing time of LSH, while only storing about 6 bytes per vector.…”