Can a Monkey with a Computer Create Art.pptVIP

Can a Monkey with a Computer Create Art? J. C. Sprott Department of Physics University of Wisconsin - Madison Presented to the Society for Chaos Theory in Psychology Life Sciences in Madison, Wisconsin on August 4, 2001 Outline How this project came about Properties of strange attractors Search techniques Aesthetic evaluation The computer art critic Samples Example (2-D Quadratic Iterated Map) xn+1 = a1 + a2xn + a3xn2 + a4xnyn + a5yn + a6yn2 yn+1 = a7 + a8xn + a9xn2 + a10xnyn + a11yn + a12yn2 Solutions Are Seldom Chaotic Probability of chaotic solutions Types of Attractors Strange Attractors Limit set as t ? ? Set of measure zero Basin of attraction Fractal structure non-integer dimension self-similarity infinite detail Chaotic dynamics sensitivity to initial conditions topological transitivity dense periodic orbits Aesthetic appeal Stretching and Folding Fractals Geometrical objects generally with non-integer dimension Self-similarity (contains infinite copies of itself) Structure on all scales (detail persists when zoomed arbitrarily) Natural Fractals Human Evaluations Aesthetic Evaluation A Simple 4-D Example xn+1 = a1xn + a2xn2 + a3yn + a4yn2 + a5zn + a6zn2 + a7cn + a8cn2 (horizontal) yn+1 = xn (vertical) zn+1 = yn (depth) cn+1 = zn (color) “Infinite” Variety 8 adjustable coefficients Like settings on combination lock 26 values of each coefficient 8-character name: KKGEOLMM Compact coding! DOS filename 268 = 2 x 1011 different codes ~0.01% are visually interesting Would take 1 year to see interesting ones at a rate of 1 per second Symmetric Icons Selection Criteria Must be bounded (|x| 100) Must be chaotic (positive LE) 1.2 fractal dimension 1.9 More than 10% of pixels on Less than 50% of pixels on Artificial Neural Networks Computer Art Critique Network trained on 100 “good” images and 100 “bad” images Inputs are first 8000 bytes of gif file Network has 16 neurons A single output (can be + or -) Gives ~85% accuracy on training set (200 cas