lshkit::HyperPlaneLsh Class Reference

Random hyperplane based LSH for cosine similarity. More...

#include <lsh.h>

List of all members.

Public Types
typedef const float *	Domain
Public Member Functions
template<typename RNG>
void	reset (const Parameter &param, RNG &rng)
template<typename RNG>
	HyperPlaneLsh (const Parameter &param, RNG &rng)
unsigned	getRange () const
unsigned	operator() (Domain obj) const
unsigned	operator() (Domain obj, float *delta) const
template<class Archive>
void	serialize (Archive &ar, const unsigned int version)
Classes
struct	Parameter

---

Detailed Description

Random hyperplane based LSH for cosine similarity.

Random hyperplane based LSH can be used to approximate cosine similarity. This LSH is defined on the D-dimensional vector space. For a vector X, the hash value is defined as

where <a1,...,aD> is a random vector sampled from the unit hypersphere.

The domain of the LSH is (const float *), and the parameter is defined as

      struct Parameter {
          unsigned dim;
      };

The range of this LSH is 0.

For more information on stable distribution based LSH, see the following reference.

Charikar, M. S. 2002. Similarity estimation techniques from rounding algorithms. In Proceedings of the Thiry-Fourth Annual ACM Symposium on theory of Computing (Montreal, Quebec, Canada, May 19 - 21, 2002). STOC '02. ACM, New York, NY, 380-388. DOI= http://doi.acm.org/10.1145/509907.509965

---

The documentation for this class was generated from the following file:

• include/lshkit/lsh.h

---