convergence {work in progress}

Repetition is an integral characteristic of most machines. The ability to repeat tasks extremely fast with small variations is central to almost any machine driven problem solving. Such iterative procedures are used to synthesize images, to construct paths through the landscape of all possible images that arrive at or approach some desired outcome. Most of these processes when they try to reach a solution, they start by first creating erratic changes and variations and over time slowing down and making micro adjustments. This is a convergent process that also reveals what features of an object are most significant and most sensitive to changes in relation to that being recognizable.

In this series we study the sequence of adjustments that an AI needs to do to increase the recognizability of a relatively constrained set of shapes.