Configuration-free imbalanced learning

The long-term goal of the project is to facilitate automated design of machine learning pipelines in the case of imbalanced distribution of target classes.

Project status

Currently, only binary classification setting is implemented.

Benchmark experiments are available for Auto-gluon and Imba.

Prerequisites

1. Python interpreter 3.10.
2. Installation of requirements.

Usage example

To run a benchmark just type in the terminal:

./experiment.sh

By default, benchmark for Imba will be run. To change to Auto-gluon, add the ag argument.

Stdout is in a file. To change to console output, add the c argument.