TensorFlow with R
R document that details the process of creating and deploying the model.
A folder that contains a Keras Neural Network model developed, and created in the R Notebook. The resulting files are Tensorflow, not R. The main file is a Protocol Buffer
The model was also deployed to RStudio Connect and can be found here: http://colorado.rstudio.com/rsc/connect/#/apps/2230/access
Source code for the Shiny application that uses the deployed Keras model in RStudio Connect as its source. It uses shinymaterial
to make it more mobile-friendly.
Companion presentation deck in PDF format: https://github.com/sol-eng/tensorflow-w-r/blob/master/presentation/tensorflow-w-R.pdf
An example of how to automate the workflow using the drake
package is in the workflow folder.