Neutrality, Robustness, and Evolvability in Genetic Programming

Redundant mapping from genotype to phenotype is common in evolutionary algorithms, especially in genetic programming (GP). Such a redundancy leads to neutrality, a situation where mutations to a genotype may not alter its phenotypic outcome. The effects of neutrality can be better understood by quantitatively analyzing its two observed properties, robustness and evolvability. In this chapter, we summarize our previous work on this topic in examining a compact Linear GP algorithm. Due to the choice of this particular system we can characterize its entire genotype, phenotype, and fitness networks, and quantitatively measure robustness and evolvability at the genotypic, phenotypic, and fitness levels. We then investigate the relationship between robustness and evolvability at those different levels. Technically, we use an ensemble of random walkers and hill climbers to study how robustness and evolvability are related to the structure of genotypic, phenotypic, and fitness networks and influence the evolutionary search process.

[1]  M. Collins.  Finding needles in haystacks is harder with neutrality , 2006, Genetic Programming and Evolvable Machines.

[2]  Meredith V. Trotter,et al.  Robustness and evolvability. , 2010, Trends in genetics : TIG.

[3]  Riccardo Poli,et al.  An empirical investigation of how and why neutrality affects evolutionary search , 2006, GECCO '06.

[4]  P. Schuster,et al.  Generic properties of combinatory maps: neutral networks of RNA secondary structures. , 1997, Bulletin of mathematical biology.

[5]  Wolfgang Banzhaf,et al.  Evolution on Neutral Networks in Genetic Programming , 2006 .

[6]  A. Wagner,et al.  Innovation and robustness in complex regulatory gene networks , 2007, Proceedings of the National Academy of Sciences.

[7]  P. Schuster,et al.  From sequences to shapes and back: a case study in RNA secondary structures , 1994, Proceedings of the Royal Society of London. Series B: Biological Sciences.

[8]  Terence Soule,et al.  Resilient Individuals Improve Evolutionary Search , 2006, Artificial Life.

[9]  A. Wagner.  Robustness, evolvability, and neutrality , 2005, FEBS letters.

[10]  P. Turner,et al.  Robustness promotes evolvability of thermotolerance in an RNA virus , 2008, BMC Evolutionary Biology.

[11]  Ting Hu,et al.  Neutrality and variability: two sides of evolvability in linear genetic programming , 2009, GECCO.

[12]  Risto Miikkulainen,et al.  Acquiring evolvability through adaptive representations , 2007, GECCO '07.

[13]  G. Wagner,et al.  Mutational robustness can facilitate adaptation , 2010, Nature.

[14]  W. Banzhaf,et al.  Evolvability and Speed of Evolutionary Algorithms in Light of Recent Developments in Biology , 2010, Journal of Artificial Evolution and Applications.

[15]  Franz Rothlauf,et al.  Redundant Representations in Evolutionary Computation , 2003, Evolutionary Computation.

[16]  Ting Hu,et al.  Evolutionary dynamics on multiple scales: a quantitative analysis of the interplay between genotype, phenotype, and fitness in linear genetic programming , 2012, Genetic Programming and Evolvable Machines.

[17]  Andreas Wagner,et al.  Protein robustness promotes evolutionary innovations on large evolutionary time-scales , 2008, Proceedings of the Royal Society B: Biological Sciences.

[18]  Wolfgang Banzhaf,et al.  Genotype-Phenotype-Mapping and Neutral Variation - A Case Study in Genetic Programming , 1994, PPSN.

[19]  Phil Husbands,et al.  Neutral Networks and Evolvability with Complex Genotype-Phenotype Mapping , 2001, ECAL.

[20]  Scott A. Rifkin,et al.  Genetic Properties Influencing the Evolvability of Gene Expression , 2007, Science.

[21]  Andreas Wagner,et al.  Genotype networks, innovation, and robustness in sulfur metabolism , 2010, BMC Systems Biology.

[22]  Julian Francis Miller,et al.  Through the Interaction of Neutral and Adaptive Mutations, Evolutionary Search Finds a Way , 2006, Artificial Life.

[23]  Ting Hu,et al.  The Effects of Recombination on Phenotypic Exploration and Robustness in Evolution , 2014, Artificial Life.

[24]  Malcolm I. Heywood,et al.  A Linear Genetic Programming Approach to Intrusion Detection , 2003, GECCO.

[25]  Eric L. Miller,et al.  The Ascent of the Abundant: How Mutational Networks Constrain Evolution , 2008, PLoS Comput. Biol..

[26]  Markus Brameier,et al.  On linear genetic programming , 2005 .

[27]  A. Wagner.  Robustness and evolvability: a paradox resolved , 2008, Proceedings of the Royal Society B: Biological Sciences.

[28]  P. Schuster,et al.  IR-98-039 / April Continuity in Evolution : On the Nature of Transitions , 1998 .

[29]  Ting Hu,et al.  Robustness, Evolvability, and Accessibility in Linear Genetic Programming , 2011, EuroGP.

[30]  Aytac Guven,et al.  Linear genetic programming for time-series modelling of daily flow rate , 2009 .

[31]  Wolfgang Banzhaf,et al.  A comparison of linear genetic programming and neural networks in medical data mining , 2001, IEEE Trans. Evol. Comput..

[32]  J. Miller,et al.  Guidelines: From artificial evolution to computational evolution: a research agenda , 2006, Nature Reviews Genetics.