Rejection Sampling for Weighted Jaccard Similarity Revisited

Efficiently computing the weighted Jaccard similarity has become an active research topic in machine learning and theory. For sparse data, standard technique is based on consistent weighed sampling (CWS). dense however, methods rejection (RS) can be much more efficient. Nevertheless, existing RS are still slow for practical purposes. In this paper, we propose to improve by a strategy, which call efficient (ERS), ``early stopping + densification''. We analyze statistical property of ERS provide experimental results compare with other algorithms hashing Jaccard. The demonstrate that significantly improves estimating relatively data.