FITC 2005 Presentations - Information Visualization

Information Visualization

Branden Hall - [www.waxpraxis.org]

FITC 2005

-> Machine Creation to design layouts

-> Genetic algorithms

Apply "evolution" to code
You can "breed" for the qualities you want
Create an initial random population of organisms

-> Represent code in tree form

Simple data structure (Array)
Handle crossover and mutation
Use split and splice etc. to form crossovers

-> Computer knows "pretty" from "ugly"

How do we teach it?
Symmetry
Balance
Self-similarity (Fractal)
Computers are quite good at figuring this out

-> Neural Networks

Individual nodes do not make sense but the "whole" does
Perseptron based back propogation network
Old neural network

-> Nurture NET

Flash first to generate output
Feeds output into Python neuralNet

-> L-Systems

Used to represent how some natural organisms grow
String replacement
"ABAB"
Set up replacement rules and run for each generation
He mentions a MIDI L-System music generator from the 80s

-> Zen of Code Optimization (Book Link)

The author wrote the Quake engine
Flash is so removed from the CPU you have to be "smarter" to optimize speed (ah look at that finally Flash programmers are "smarter" :)

-> Look up L-System rendered on his Blog

-> Deptofnotation.com

-> Notes will be posted here