This code contains an implementation of the method described in the Annals of Statistics paper "Density-Sensitive Semisupervised Inference" by Martin Azizyan, Aarti Singh, Larry Wasserman (2013).

Executing main.R will reproduce the experimental results presented in the paper (errplot.pdf). The code can take more than 10 minutes to run. Raw experimental results have been stored in ress.300.dat, which can be used to avoid re-running the experiment (see comments in main.R).

dbd-regression.R contains the implementation of the method, as well as some helper functions. See comments inside dbd-regression.R and main.R for details.
