How a Neural Network Learned Its Own Fraud Rules: A Neuro-Symbolic AI Experiment

Most neuro-symbolic systems inject rules written by humans. But what if a neural network could discover those rules itself? In this experiment, I extend a hybrid neural network with a differentiable rule-learning module that automatically extracts IF-THEN fraud rules during training. On the Kaggle C