/self-reinforcing-bias

Experiments on ML systems creating self-reinforcing bias

Primary LanguageGoGNU General Public License v3.0GPL-3.0

self-reinforcing-bias

Experiments on ML systems creating self-reinforcing bias

Phase 1: Logistic Regression

This phase will be the easiest, and will most-obviously work. More sophisticated ML will come later.

The idea is to show that training a model without explicitly recording or looking for negative training examples will produce bias in a wide variety of situations.