Skip to content

AndyWangSFU/Active_Learning_in_R

Repository files navigation

Active Learning in R: A Quick Introduction for Beginners

In this project, we aim to implement some baseline active learning strategies in R, experiment on a famous dataset Iris, highlight some insights and suggest future directions for active learning in the Statistics domain.

Layout of Sources (with short descriptions)

Active_Learning_in_R
.
├── Active_Learning_in_R_Rcode.Rmd            (Source code)
├── iris.data                                 (data)
├── Active_Learning_in_R_Rcode_Simulation.pdf (Source code simulation using RMarkdown)
├── Active_Learning_in_R.pdf                  (A seemingly formal report)
├── Active_Learning_in_R_PPTslides.pptx       (A short talk and user guide)
├── Active_Learning_in_R_Presentation.mp4     (A short talk and user guide)
├── README.md

Public Post

I also published this project in Medium if you are interested to look at here

Experimental Settings

Active Learning Querying Strategies

We implement two basic strategies "Uncertainty Sampling" and "Random Sampling".

Uncertainty Sampling

Select the point with least confidence. One criteria is the point nearest to the current decision boundary.

Random Sampling

Query the point randomly from the unlabeled pool.

Model

The model we currently use is logistic regression, which is a classifier for binary classification problems.

About

An active learning demo in R

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published