How it works: Input → Weights → Σ → Step function → Output No hidden layers. No backprop (in classic form). Just raw logic. Yet, it can linearly separate classes. Mind = blown. #NeuralNetworks #PerceptronMath @PerceptronNTWK