- author = "John Koza",
- title = "Automated Design Using {Darwinian} Evolution and Genetic Programming",
- howpublished = "Stanford University, EE380: Computer Systems Colloquium",
- year = "2009",
- month = "18 " # feb,
- keywords = "genetic algorithms, genetic programming",
-
video_url = "
https://www.youtube.com/watch?v=xIoytwJWJP8",
- size = "75 minutes",
- abstract = "John Koza describes an automated: What You Want Is What You Get, process for designing complex structures based on the principles of natural selection, sexual recombination, and developmental biology.",
-
notes = "Routine Human Competitive.
Darwin on the origin of species \cite{darwin:book}
Parallel computing
Alan Turing 1948 \cite{turing48:_intel}
Cross validation is required
Sherman's march through 1970-80s AI and ML problems.
31 minutes: Developmental GP \cite{koza:gp3} 39 minutes GP as program constructing trees => NetGraph => NetList => SPICE => output => Fitness. 43 minutes Patent generation, invention machines. Novelty driven evolution: fitness = do job and avoid existing patent. \cite{koza:gp4}. Quantum computing algorithms, games, Humies. Cellular automata. Controllers, NASA antenna. AI ratio. An automated way of routinely automatically designing complex objects.
54 minutes you do not have to be smart to use GP
56 minutes GP starts small. Simulation v. lab work, eg Laser, FPGA. 59 Langdon memory. Evolving ADFs. Reusable parameterised subroutines. Parameterised solutions. Parallel computing: Farm, island model. Moore's Law 1:100000 speedup, 100x speed up in ten years time. Use GP for black art problem domains.
67mins questions. Syntax preserving crossover. Antennas. HPC. Seti at home (informal cloud) gamification of GP. Intel micro payments. Memristor.
Koza videos \cite{koza:video} \cite{koza:video2} \cite{koza:video3} \cite{koza:video4} \cite{koza:2009:gpt}",
