NEWS

---

nnTensor 1.3.0 (2024-05-13)

• sym option in kFoldMaskTensor was added

nnTensor 1.2.0 (2023-06-22)

• kFoldMaskTensor was modified to avoid NA
• Vignette for kFoldMaskTensor was added
• NMF(), siNMF(), jNMF(), NTF(), and NTD() were modified to allow missing values (NA) in the input matrix/tensor

nnTensor 1.1.14

• Accepted by JOSS

nnTensor 1.1.13 (2023-03-17)

• Vignettes were modified

nnTensor 1.1.12 (2023-02-15)

• Help messages for NTF(), NTD(), and recTensor() were updated to explain higher order tensor cases
• Vignettes were added

nnTensor 1.1.11

• Debug in NTF() and NTD(), (when the init="ALS")

nnTensor 1.1.10 (2022-11-15)

• Inf/-Inf/NaN/NA is replaced by machine epsilon in NTF()

nnTensor 1.1.9

• Some bugs were fixed
• orthU orthV, and root options were added in NMTF()
• ALS-type NMF was added in NMF()

nnTensor 1.1.8 (2022-07-14)

• NMTF() for nonnegative matrix tri-factorization was added
• pseudocount 1e-10 for NMF(), siNMF(), jNMF(), NTF(), NTD() was replaced by the machine epsilon

nnTensor 1.1.7 (2022-06-17)

• Exponent term depending on Beta parameter was added in NMF, siNMF, jNMF, NTF, and NTD

nnTensor 1.1.6 (2022-05-02)

• Bug with NTF() with rank=1 was fixed
• KFoldMaskTensor to generate mask tensors for K-fold cross validation was added

nnTensor 1.1.5 (2021-09-03)

• method, sign, and option were added in plotTensor3D.
• plotTensor2D was added

nnTensor 1.1.4 (2021-08-17)

• Projected NMF ("Projected"), Nonnegative Hebbian Rule (NHR) ("NHR"), Ding-Ti-Peg-Park (DTPP) algorithm ("DTPP"), (Column vector-wise) Orthogonal NMF ("Orthogonal"), and (Column vector-wise) Orthogonality-regularized NMF ("OrthReg") were added in NMF.
• Algorithm "NMF" is introduced in each iteration of NTD.

nnTensor 1.1.2 (2021-07-06)

• pseudocount option to avoid zero-division was added in NMF, jNMF, siNMF, NTF, and NTD

nnTensor 1.1.0 (2021-06-24)

• Refactored some codes (e.g. check argument parts)
• L1/L2-regularization terms were introduced in NMF, jNMF, siNMF, NTF, and NTD

nnTensor 1.0.7 (2021-05-18)

• Resubmission to CRAN due to temporary inaccessibility of rTensor

nnTensor 1.0.5 (2020-06-04)

• Mask matrix/tensor were implemented for NMF, jNMF, siNMF, NTF, and NTD. These options are for missing values estimation (Imputation).
• Initial matrix/tensor were implemented for NMF, jNMF, siNMF, NTF, and NTD.
• fix* options were implemented for transfer learning.
• Gabriel-style cross-validation was implemented as GabrielNMF.

nnTensor 1.0.4 (2020-04-20)

• Rank estimation methods and plot functions were implemented for NMF

nnTensor 1.0.3

• Higher-order (>=4) tensor decomposition were implemented for NTF and NTD VERSION 1.0.1 - 1.0.2
• Some bugs were fixed

nnTensor 1.0.0 (2019-07-30)

• Simutaneous NMF (siNMF) and joint NMF (jNMF) were added

nnTensor 0.99.5 (2019-07-28)

• modes parameter was added in NTF (NTD1 and NTD2 model)
• Convergence was judged by relative error

nnTensor 0.99.4 (2018-09-16)

• set.seeds argument was added in toyModel

nnTensor 0.99.3

• Date fields in DESCRIPTION

nnTensor 0.99.2

• Reconstuction Error and relative change of the error for NMF, NTF, and NTD
• toyModel to return integar random tensor

nnTensor 0.99.1 (2018-06-04)

• Revised
• DESCRIPTION was fixed
• No intermediate files are saved, unless viz=FALSE (Default) was changed as viz=TRUE VERSION 0.99.0
• Package released