dom_adapt

Code for the NIPS 2018 paper "Domain Adaptation by Using Causal Inference to Predict Invariant Conditional Distributions"

Installation instructions

The code expects a Clingo solver in the folder ASP (point 1) and a few R packages (point 2):

1. Download the new clingo

• From http://sourceforge.net/projects/potassco/files/clingo and rename it as "clingo"
• To see whether your clingo installation is working, you can try running:

  ./ASP/clingo
  

2. Install the R packages(*)

• 2.1 Install Bioconductor packages

  • With R version 3.5 or greater, use BiocManager:

  if (!requireNamespace("BiocManager", quietly = TRUE))
      install.packages("BiocManager")
  BiocManager::install(version = "3.12")
  
  BiocManager::install(c('graph','RBGL','gmp','RcppArmadillo'))
  

  • With R version lesser than 3.5, install:

  source('http://bioconductor.org/biocLite.R')
  biocLite(c('graph','RBGL','gmp','RcppArmadillo'))
  

• 2.2 Install the remaining packages:

  install.packages(c('deal','combinat','hash','bnlearn','foreach','doMC','caTools','expm'))
  install.packages(c('pcalg'))
  

  Note that:

  • Installing 'pcalg' may require you to install also a few other packages (e.g. robustbase, ggm).
  • For R version 4.0.4 'pcalg' dependencies are managed automatically.

3. Start R

• Navigate to the R/ directory and run:

  source('load.R')
  loud()
  

(*) If you want to keep your global environment unchanged, please, consider using the renv package (https://rstudio.github.io/renv/).