Looking for a simple way to do #deeplearning in #Rstats? Check out cito, available on #CRAN via https://cran.r-project.org/web/packages/cito/index.html and presented in this recent Ecography paper https://nsojournals.onlinelibrary.wiley.com/doi/full/10.1111/ecog.07143
Cito is an interface to torch for R and supports building and training
Cito is an interface to torch for R and supports building and training
1 / 2
Comments
library(qeML)
data(svcensus)
z <- dnn(wageinc ~ ., svcensus)
# "Bad training..."
data(lsa)
z <- dnn(lsat ~ .,data=lsa[,c(1,4:8)])
# "Bad training..."
vignette('B-Training_neural_networks')
# vignette ‘B-Training_neural_networks’ not found
vignette(package='cito')
# no vignettes found
# try classification example instead of regression
z <- dnn(gender ~ ., svcensus)
# Error in torch_...
# maybe it wants dummy variable for Y?
sex <- as.numeric(svcensus$gender == 'male')
svcensus$sex <- sex
svcensus$gender <- NULL
z <- dnn(sex ~ ., svcensus)
# "Bad training..."
The current CRAN version supports only fully connected DNNs, but our development version on GitHub also includes CNNs and multi-modal architectures.