Go to the source code of this file.
Namespaces | |
namespace | lshkit |
Classes | |
class | lshkit::StableDistLsh< DIST > |
Stable distribution based LSH. More... | |
struct | lshkit::StableDistLsh< DIST >::Parameter |
class | lshkit::HyperPlaneLsh |
Random hyperplane based LSH for cosine similarity. More... | |
struct | lshkit::HyperPlaneLsh::Parameter |
class | lshkit::ThresholdingLsh |
Random hyperplane based LSH for L1 distance. More... | |
struct | lshkit::ThresholdingLsh::Parameter |
Typedefs | |
typedef StableDistLsh< Cauchy > | lshkit::CauchyLsh |
LSH for L1 distance. | |
typedef StableDistLsh< Gaussian > | lshkit::GaussianLsh |
LSH for L2 distance. |
Be aware that the LSH classes defined here are only atomic hash functions. You need to compose them with the templates found in composite.h in order to reproduce the well-known published LSHes. For example, the Stable distribution based LSH in M. Datar's paper (and also Q. Lv's Multi-Probe LSH paper) is really RepeatHash<StableDistLsh>.