Introduction to Machine Learning

Short interactive course to introduce basics of Machine Learning in R with Torch.

Features

• Presentation of the basics of Machine Learning which:
  • Explains main concepts of Machine Learning (e.g. gradient descent, overfitting)
  • Presents Machine Learning frameworks in R (Tensorflow vs. Torch)
  • Gives examples of uses of Machine Learning in different fields
• Provide an interactive Rmarkdown to create a simple neural network with Torch (also available in pdf and html) which go over the following steps:
  • Cleaning the data
  • Building a deep neural network
  • Training the neural network
  • Evaluating performances of the neural network
• Give references to go further (see below)

Requirements

The Torch framework needs to be installed to run introduction-machine-learning.Rmd. The installation of the different depencies are detailed on the first cells of that file. Otherwise, you can run with R the following lines to install the Torch framework:

install.packages("torch")
install.packages("torchvision")

Then to install the dataset run:

remotes::install_github("mlverse/torchdatasets")

Lastly, to check that Torch is installed correctly, do:

library(torch)
torch_tensor(1)

Further reading

• Torch framework for R (with the example that inspired the Rmarkdown)
• PyTorch documentation that can complete the incomplete R documentation (if you know a bit about Python, it can be easily transposed in R)
• Introductory course to Machine Learning by M. Pichler and F. Hartig