Weighted k-Means coresets for active learning

This repository contains my geometric algorithms (Compsci 634, Spring 2022) capstone project. This project proposes a batch-mode active learning method using weighted k-means coresets.

Installation

Install the following dependencies:

$ pip install numpy pandas matplotlib modAL 

In addition, this uses pushkarnimkar/zalanborsos-coresets, a fork of zalanborsos/coresets with minor modifications for coreset implementation. To add it to the PYTHONPATH:

$ export PYTHONPATH="$PYTHONPATH:/path/to/pushkarnimkar/zalanborsos-coresets"

Repository Organization

• CS634GeomALProp.pdf: project report
• datagen.py: generator for synthetic dataset
• stub.py: executable file - entry point

Running the script

Execute the script using python stub.py with no arguments