sslMincut: Mincut

In SSL: Semi-Supervised Learning

Description

sslMincut implements the Mincut algorithm for maxflow graph partition in the k-nearest neighbor graph.

Usage

sslMincut(xl, yl, xu, simil.type = "correlation", k = 10)

Arguments

xl	a n * p matrix or data.frame of labeled data
yl	a n * 1 binary labels(1 or -1).
xu	a m * p matrix or data.frame of unlabeled data.
simil.type	character string; this parameter controls the type of similarity measurement.(see simil or pr_DB).
k	an integer parameter controls a k-nearest neighbor graph.

Details

sslMincut creates a k-nearest neighbor graph and finds a maxflow from the first postive observation to the first negative one based on MPLA algorithm. This maxflow partitions the graph into postive labels and negative ones.

Value

a m * 1 integer vector representing the predicted labels of unlabeled data.

Author(s)

Junxiang Wang

References

Blum, A., & Chawla, S. (2001). Learning from labeled and unlabeled data using graph mincuts. Proc. 18th International Conf. on Machine Learning.

See Also

pr_DB simil

Examples

data(iris)
xl<-iris[c(1:20,51:70),-5]
xu<-iris[c(21:50,71:100),-5]
yl<-rep(c(1,-1),each=20)
yu<-sslMincut(xl,yl,xu)

SSL documentation built on May 29, 2017, 7:14 p.m.