Logo

Josef

An Artificial Neural Network That Learns How to Draw a Path

Featured on PitchWall

Published At

Josef is feedforward neural network based on synapticjs and distant relative of Istanbul's famous robot poet Deniz Yilmaz. The network tries to predict next action of the drawing grammar. Works like an Lindenmayer interpreter running by a neural network which is a non-deterministic way to evaluate that kind of self-rewriting systems. The network has 35 inputs consist the last 5 operations of the grammar and 7 output variables that represents a single action character.

Submitted By

Fatih Erikli

Fatih Erikli

creator of arguman.org project, amateur generative artist.