A Genome Compiler for High Performance Genetic Programming

Created by W.Langdon from gp-bibliography.bib Revision:1.8120

@InProceedings{fukunaga:1998:gchpGP,

• author = "Alex Fukunaga and Andre Stechert and Darren Mutz",
• title = "A Genome Compiler for High Performance Genetic Programming",
• booktitle = "Genetic Programming 1998: Proceedings of the Third Annual Conference",
• year = "1998",
• editor = "John R. Koza and Wolfgang Banzhaf and Kumar Chellapilla and Kalyanmoy Deb and Marco Dorigo and David B. Fogel and Max H. Garzon and David E. Goldberg and Hitoshi Iba and Rick Riolo",
• pages = "86--94",
• address = "University of Wisconsin, Madison, Wisconsin, USA",
• publisher_address = "San Francisco, CA, USA",
• month = "22-25 " # jul,
• publisher = "Morgan Kaufmann",
• keywords = "genetic algorithms, genetic programming, SPARC machine language",
• ISBN = "1-55860-548-7",
• URL = "http://metahack.org/gp98-compiler.pdf",
• URL = "http://fukunaga.bol.ucla.edu/gp98-compiler.pdf",
• URL = "http://citeseer.ist.psu.edu/fukunaga98genome.html",
• size = "9 pages",
• abstract = "Genetic Programming is very computationally expensive. For most applications, the vast majority of time is spent evaluating candidate solutions, so it is desirable to make individual evaluation as efficient as possible. We describe a genome compiler which compiles s-expressions to machine code, resulting in significant speedup of individual evaluations over standard GP systems. Based on performance results with symbolic regression, we show that the execution of the genome compiler system is comparable to the fastest alternative GP systems. We also demonstrate the utility of compilation on a real-world problem, lossless image compression. A somewhat surprising result is that in our test domains, the overhead of compilation is negligible.",
• notes = "'Our compiler directly generates the machine executable code that corresponds to this assembly-level code.' 'eliminates the the overhead of invoking an external compiler.' 'The genome compiler approach is similar to \cite{Nordin:1995:tcp}'

  target = x**9. 30 generations

  RISC 296 MHz UltraSparc 2 https://en.wikipedia.org/wiki/SPARC

  'reaching a maximum speedup of around 50 when the number of test cases is 1000.' 'The genome compiler, due to its use of machine code, achieves about an order of magnitude speedup over HiGP' \cite{stoffel:1996:hpsbGP}. 'The genome compiler performs roughly 50-60 times faster than 'Tackett and Carmi' SGPC \cite{sgpc_readme} running on the same machine, which is comparable to the execution speeds for CGPS reported by Nordin and Banzhaf \cite{Nordin:1995:tcp}.'

  Lossless Image Compression future work 'compiler optimizations which use editing operations \cite{koza:book} or standard compiler optimization techniques to collapse instructions together, remove redundant operations, reorder operations, etc.,'

  GP-98

  Thu, 25 Jun 1998 10:31:36 PDT We've recently developed a gp system based on lil-gp which evolves s-expressions and compiles it to machine code (specifically, Sparc machine code) to speed up evaluation. In our system, we've found that the overhead of compilation is negligible, since the vast majority of the time spent in execution in an s-expression interpreter (in our case, the lil-gp interpreter) is consumed by the recursive traversal of the tree.

  A full description, comparisons with previous GP-compiler systems and some experimental results with symbolic regression and image compression are described",

}

Genetic Programming entries for Alex S Fukunaga Andre Stechert Darren Mutz

Citations