Learn R Programming

sMTL (version 0.1.0)

Sparse Multi-Task Learning

Description

Implements L0-constrained Multi-Task Learning and domain generalization algorithms. The algorithms are coded in Julia allowing for fast implementations of the coordinate descent and local combinatorial search algorithms. For more details, see a preprint of the paper: Loewinger et al., (2022) .

Copy Link

Version

Install

install.packages('sMTL')

Monthly Downloads

145

Version

0.1.0

License

MIT + file LICENSE

Issues

Pull Requests

Stars

Forks

Maintainer

Gabriel Loewinger

Last Published

February 6th, 2023

Functions in sMTL (0.1.0)

tuneZscale

tuneZscale: scale lambda_z depending on magnitude. For internal package use only.
sparseL0Tn_iht

sparseCV_L0: cross-validation functions. For internal package use only.
multiTaskRmse_MT

multiTaskRmse: calculate average (across tasks) RMSE for multi-label prediction problems
seReturn

seReturn: find smallest rho within 1 se of smallest cv error. For internal package use.
method_nm

methods names: give name for printing. Internal package use only.
reName_cv

reName_cv: rename output from CV. For internal package use only.
rhoScale

rhoScale: scale lambda_z depending on magnitude. For internal package use only.
grid.gen

grid.gen: generate grid for cross-validation function. For internal package use only.
predict

predict: predict on smtl model object
maxEigen

maxEigen: maximum eigenvalue wrapper for Julia TSVD package. internal package use only
cv.smtl

cv.smtl: cross-validation function
multiTaskRmse

multiTaskRmse: RMSE for multi-task problems (averaged across tasks)
sparseCV

sparseCV: cross-validation functions. For internal package use only.
smtl_setup

smtl_setup: setup Julia path and/or install Julia or Julia packages using functions based on external package JuliaCall::julia_setup().
smtl

smtl: make model-fitting function
sparseCV_MT

sparseCV_MT: internal cross-validation functions. For internal package use only.