test_mask_siNMF.R
In nnTensor: Non-Negative Tensor Decomposition

X <- toyModel("siNMF_Hard")
M1 <- list()
M2 <- list()
length(M1) <- length(X)
length(M2) <- length(X)
for(i in seq(length(X))){
	M1[[i]] <- kFoldMaskTensor(X[[i]], seed=12345)[[i]]
	M2[[i]] <- kFoldMaskTensor(X[[i]], seed=54321)[[i]]
}

out1 <- siNMF(X, M=M1, J=3, algorithm="Frobenius", num.iter=2)
out2 <- siNMF(X, M=M1, J=3, algorithm="KL", num.iter=2)
out3 <- siNMF(X, M=M1, J=3, algorithm="IS", num.iter=2)
out4 <- siNMF(X, M=M1, J=3, algorithm="PLTF", p=1, num.iter=2)

expect_equivalent(length(out1), 6)
expect_equivalent(length(out2), 6)
expect_equivalent(length(out3), 6)
expect_equivalent(length(out4), 6)



out5 <- siNMF(X, M=M2, J=3, algorithm="Frobenius", num.iter=2)
out6 <- siNMF(X, M=M2, J=3, algorithm="KL", num.iter=2)
out7 <- siNMF(X, M=M2, J=3, algorithm="IS", num.iter=2)
out8 <- siNMF(X, M=M2, J=3, algorithm="PLTF", p=1, num.iter=2)

expect_true(rev(out1$TestRecError)[1] != rev(out5$TestRecError)[1])
expect_true(rev(out2$TestRecError)[1] != rev(out6$TestRecError)[1])
expect_true(rev(out3$TestRecError)[1] != rev(out7$TestRecError)[1])
expect_true(rev(out4$TestRecError)[1] != rev(out8$TestRecError)[1])



out9 <- siNMF(X, M=M1, J=4, algorithm="Frobenius", num.iter=2)
out10 <- siNMF(X, M=M1, J=4, algorithm="KL", num.iter=2)
out11 <- siNMF(X, M=M1, J=4, algorithm="IS", num.iter=2)
out12 <- siNMF(X, M=M1, J=4, algorithm="PLTF", p=1, num.iter=2)

expect_true(rev(out1$TestRecError)[1] != rev(out9$TestRecError)[1])
expect_true(rev(out2$TestRecError)[1] != rev(out10$TestRecError)[1])
expect_true(rev(out3$TestRecError)[1] != rev(out11$TestRecError)[1])
expect_true(rev(out4$TestRecError)[1] != rev(out12$TestRecError)[1])