“Hashing Algorithms for Large-Scale Learning”

In “Hashing Algorithms for Large-Scale Learning” Li, Shrivastava, Moore, and Konig (2011) modify Minwise hashing by storing only the $b$ least significant bits.  They use the $b$ bits from $k$ hashing functions as features for training a support vector machine and logistic regression classifiers.  They compare their results against Count-Min, Vowpal Wabbit, and Random hashes on a large spam database.  Their algorithm compares favorably against the others.