Active-Learning-Homology

This is a project for the paper entitled "Finding the Homology of Decision Boundaries with Active Learning" in NeurIPS2020. Link to full paper

Running environment and required packages

Python 3.6.8

matlab.engine

Ripser

How to Run

Empirical results: ./Run.sh

Numerical Simulation: Python Simulate.py

Outputs

Empirical results:

1. Figures with queried examples highlighted
2. Persistent diagrams
3. Betti number
4. Bottleneck distance

Numerical Simulation:

1. Comparison between sample complexity from passive learning and query complexity from active learning