nystroem_method_test.cpp File Reference

#include <mlpack/core.hpp>
#include "catch.hpp"
#include <mlpack/methods/nystroem_method/ordered_selection.hpp>
#include <mlpack/methods/nystroem_method/random_selection.hpp>
#include <mlpack/methods/nystroem_method/kmeans_selection.hpp>
#include <mlpack/methods/nystroem_method/nystroem_method.hpp>

Include dependency graph for nystroem_method_test.cpp:

Functions
	TEST_CASE ("FullRankTest", "[NystroemMethodTest]")
	Make sure that if the rank is the same and we do a full-rank approximation, the result is virtually identical (a little bit of tolerance for floating point error).
	TEST_CASE ("Rank10Test", "[NystroemMethodTest]")
	Can we accurately represent a rank-10 matrix?
	TEST_CASE ("GermanTest", "[NystroemMethodTest]")
	Can we reproduce the results in Zhang, Tsang, and Kwok (2008)? They provide the following test points (approximately) in their experiments in Section 4.1, for the german dataset: More...

Detailed Description

Author

Ryan Curtin

Test the NystroemMethod class and ensure that the reconstructed kernel matrix errors are comparable with those in the literature.

mlpack is free software; you may redistribute it and/or modify it under the terms of the 3-clause BSD license. You should have received a copy of the 3-clause BSD license along with mlpack. If not, see http://www.opensource.org/licenses/BSD-3-Clause for more information.

Function Documentation

TEST_CASE()

TEST_CASE	(	"GermanTest"	,
		""	[NystroemMethodTest]
	)

Can we reproduce the results in Zhang, Tsang, and Kwok (2008)? They provide the following test points (approximately) in their experiments in Section 4.1, for the german dataset:

rank = 0.02n; approximation error: ~27 rank = 0.04n; approximation error: ~15 rank = 0.06n; approximation error: ~10 rank = 0.08n; approximation error: ~7 rank = 0.10n; approximation error: ~3