README.md

In woobe/deepr: deepr: Steamlining Deep Learning Procedures in R (Prototype)

deepr

An R package to streamline the training, fine-tuning and predicting processes for deep learning. It aims to further simplify the functions in packages such as 'h2o' and 'deepnet'.

Installation

library(devtools)
install_github("woobe/deepr")

Example 1

Extracting hidden features from original predictors (x) and creating new predictors (x_new).

1.1 Load Library

library(deepr)

1.2 Generate random numbers as original predictors (x)

x <- matrix(rnorm(1000000), nrow = 5000)
dim(x)

[1] 5000  200

1.3 Train a RBM model with 100 hidden features

model_rbm <- train_rbm(x, n_features = 100)

=====================================================================
[deepr]: Training a Restricted Boltzmann Machine
=====================================================================
[deepr]: Removing variables with near zero variance ...
[deepr]: Normalising x to values between 0 and 1 ...
[deepr]: Training a RBM Layer of 10 Hidden Features ...
[deepr]: Returning the RBM Model ...
[deepr]: All Done! Total Duration: 474 sec.
=====================================================================

1.4 Transform original x into new x

x_new <- transform_x(model_rbm, x)

=====================================================================
[deepr]: Transforming Predictors Using a Trained RBM
=====================================================================
[deepr]: Pre-processing predictors (x) based on the RBM object ...
[deepr]: Converting original predictors (x) into new ones (x_new) ...
[deepr]: Returning new predictors (x_new) ...
[deepr]: All Done! Total Duration: 1 sec.
=====================================================================

dim(x_new)

[1] 5000   100

1.5 Install or upgrade H2O's R package quickly

## Automatically install the latest version
install_h2o()

## Install a specific version
install_h2o(h2o_ver = "1500")

## Overwrite the current R package 
install_h2o(force = TRUE)

woobe/deepr documentation built on May 4, 2019, 9:47 a.m.