This repo regards a project "An Interactive Visualization Tool for Understanding Active Learning", which was accepted by the HCAI@NeurIPS2021 workshop on Human Centered AI. Please access Active_Learning_Visualization_Demo.ipynb to run the whole demo.

We proposed a visualization tool to help users understand how active learning algorithms work as more data points are labeled and used to train the model. The visualization tool allows users to select specific data points and see how they affect the trained model. Given the role human labelers play in training machine learning models, having such visualization tools is important for the active learning literature.