Page 243 - 49A Field Guide to Genetic Programming

INDEX 229

guidelines, applications, 111 Java, 64
T7, 64
halting probability, 100 SIMD and GPU, 92
halting program, Markov chain model, 100 introns
Hamming distance, 71 troubleshooting, 136
hardware implementations of GP, 93 useful with mutation, 64
headless chicken crossover, 16 invention, evolution of, 119
heuristic, 126 inverse kinematics, 115
hierarchical structure, 47 inverse problem, 112
higher-order type systems, 53 iterated functions system (IFS), 128
hill climbing, 46
hits, 76 jet engine optimisation, 125
hoist mutation, 43, 106 journals, 147
homologous crossover, 44
linear GP, 64 kinematics, 120
theory, 98
Huffman encoding, 129 L-system, 74, 76
human competitive tree adjoining grammar, 58
artificial intelligence, 117–121, 141 Lagrange
awards (Humies), 119 distribution of the second kind, 104
human readable programs, 51 initialisation, 41
hyper-heuristic, evolving, 126 large populations, 137
learning, machine (ML), 1
image lexicographic
compression, 128 parsimony bloat control, 77, 80
processing, 121, 122 preference, 77, 80
watermark, 122 libraries, dynamic, 47, 48
implementations of GP, 147–148 Lil-GP, 148
FPGA, 93 limits, size and depth, 104
GPU, 90–92 Lindenmayer grammar, see L-system
TinyGP, 151–162 linear GP, 61–64
inactive code, 102 Cartesian GP, 67
incremental fitness, 58 crossover, 64
indirect representation, 58 homologous, 64
industrial modelling, multi-objective GP, instruction format, 62, 63
79 interpreted, 62, 64
information content of population, 139 introns, 64
information theory, 46 removal, 64
infrared images, 121 Java, 64
infrared spectra, 125 mutation, 64
initialisation speed, 62
effects on bloat, 40 T7, 64
full method, 12–13 linear representation, 61–64
grammar-based GP, 57 Cartesian GP, 67
grow method, 12–14 linearised tree-based GP
tree size bias, 13 prefix, 62
Lagrange, 41 reverse polish, 92
ramped half-and-half, 13 logic network, evolution, 66
ramped uniform, 40 lossless compression, 129
type system, 56 lossy compression, 128
uniform, 40
integer overflow, 22 machine code GP, 62–64
intelligence, artificial (AI), 1 Intel x86, 63
interpreted GP, 24 SPARC, 62, 63
linear, 62, 64 Z80, 62, 63
100 1000 2
Avg Size Avg Fitness sin(x)
Best Fitness GP (gen=99)
90
1.5
80
1
70
100
0.5
60
Generation 99 Average Size 50 Fitness 0
40 -0.5
10
30
(see Sec. B.4) 20 -1
-1.5
10
1 -2
0 20 40 60 80 100 0 20 40 60 80 100 0 1 2 3 4 5 6
Generations Generations x