The plain-English neural network (PENN) is a binary classifier expressed as prose, eliminating the need for external calculators.
In PENN, the weights in traditional deep neural networks are spelled out as sentences, improving explainability.
The capabilities of the plain-English neural net are limited by the context-window length, similar to the limitation of neurons in deep neural networks.
PENN allows for each weight to be self-documented, providing reasons for predictions and enhancing understanding.