How a Generative Encoding Fares As Problem-Regularity Decreases

2007

We propose expression simplification and tree compression as aids in understanding the evolution of regular structure in Genetic Programming individuals. We apply the analysis to two previously-published algorithms, which aimed to promote regular and repeated structure. One relies on subtree duplication operators, the other uses repeated evaluation during a developmental process. Both successfully generated solutions to difficult problems, their success being ascribed to promotion of regular structure. Our analysis modifies this ascription: the evolution of regular structure is more complex than anticipated, and the success of the techniques may have arisen from a combination of promotion of regularity, and other, so far unidentified, effects.

2007 IEEE Congress on Evolutionary Computation, 2007

Compression algorithms generate a predictive model of data, using the model to reduce the number of bits required to transmit the data (in effect, transmitting only the differences from the model). As a consequence, the degree of compression achieved provides an estimate of the level of regularity in the data. Previous work has investigated the use of these estimates to understand the replication of building blocks within Genetic Programming (GP) individuals, and hence to understand how different GP algorithms promote the evolution of repeated common structure within individuals. Here, we extend this work to the population level, and use it to understand the extent of similarity between sub-structures within individuals in GP populations.

Genetic Programming and Evolvable Machines, 2001

Biological representations underly biological evolution. Moreover, they are also a product of evolution and consequently well adapted for their purpose. The argument presented in this paper is that the representations of biology are also suitable for representing artificial executable systems in genetic programming and, furthermore, that biomimetic representations could improve both the adaptability and evolvability of GP. To this end a biomimetic approach to GP, enzyme genetic programming, is introduced and its behaviour is analysed when applied to the domain of combinational circuit design.

arXiv (Cornell University), 2021

The purpose of this paper is to abstractly describe the notion of a generative mechanism that implements a code and to provide a number of examples including the DNA-RNA machinery that implements the genetic code, Chomsky's Principles & Parameters model of a child acquiring a specific grammar given 'chunks' of linguistic experience (which play the role of the received code), and embryonic development where positional information in the developing embryo plays the role of the received code. A generative mechanism is distinguished from a selectionist mechanism that has heretofore played an important role in biological modeling (e.g., Darwinian evolution and the immune system).

… to the Search for Life on …, 2005

Lecture Notes in Computer Science, 2005

Recently, all the human genes were identified. But understanding the functions coded in the genes is of course a much harder problem. We are used to view DNA as some sort of a computer code, but there are striking differences. For example, by using entropy, it has been shown that the DNA code is much closer to random code than written text, which in turn is less ordered than ordinary computer code. Instead of saying that the DNA is badly written, using common programming standards, we might say that it is written in a different style − an evolutionary style. In this paper the coding style of creatures from the artificial life platform Avida has been studied. Avida creatures that have evolved under different size merit methods and mutation rates have been analysed using the notion of stylistic measures. The analysis has shown that the evolutionary coding style depends on the environment in which the code evolved, and that the choice of size merit method and mutation probabilities affect different stylistic properties of the genome. A better understanding of Avida's coding style, might eventually lead to insights of evolutionary codes in general.

2005 IEEE Congress on Evolutionary Computation

This paper considers the development of redundant representations for evolutionary computation. Two new families of redundant binary representations are proposed in the context of a simple mutationselection evolutionary model. The first is a family of linear encodings in which the connectivity of the search space may be designed directly via a decoding matrix. The second is a family of representations exhibiting various degrees of neutrality, and is constructed using mathematical tools from error-control coding theory. The study of these representations provides additional insight into the properties of redundant encodings, such as synonymity, locality, and connectivity, and into their interrelationships.

The self-referential model for the formation of the genetic code proposes that protein synthesis was initiated by proto-tRNA dimers. Proto-tRNAs in the dimers recognize each other through anticodon pairing. The proteins produced recognize the producing dimers through binding, forming (proto)ribonucleoprotein (RNP) aggregates. Their functions were stimulated and specificities evolved through cycling. Such cycles would be among the first in the construction of living networks, and examples of processes that might be relevant for modeling cognitive networks. The protein synthesis process is considered a main drive for the living system 0 s specific attributes of anabolic and evolutionary semi-autonomy. Structures of the anticodon dimer networks are presented. Biological data point to the encoding having been installed on the modules of dimers formed by nonself-complementary triplets. Aminoacyl-tRNA adhesion interactions integrated the dimer networks into RNP networks. Specific questions are proposed for simulation and modeling that should help in designing experimental procedures aiming at testing the model and the development of synthetic genetic codes. Keywords Genetic code Á Origins Á Evolution Á tRNA dimers Á Aminoacyl-tRNA synthetases Á Networks Á Self-reference Á Self-organization Á Simulation Á Experimental Á Pushing dynamics Á Anabolic drive Abbreviations Amino acids Are designated by three-letter abbreviations; groups of amino acids may be indicated by one-letter abbreviations Anticodon The code triplet of transfer RNA, the default notation of code triplets aRS Synthetase, aminoacyl-tRNA synthetase Codon The code triplet of messenger RNA iMet

Progress in Biological Chirality, 2004

A model for construction of the genetic code is presented, based on successive recruitment of pairs of palindromic (bases at the 5´ and 3´ positions are the same in each strand) anticodons. The anticode matrix contains eight pairs of boxes, each box defined by the principal dinucleotide (pDiN; central 3´ base). Triplets may be depicted as •pDiN (•, the 5´ bases). In split boxes, the first occupier amino acid developed four-fold degeneracy then conceded 5´ Y to new attributions, keeping the 5' R. This rule is consistent with the criteria of aminoacyl-tRNA synthetases class 2 (aRS2) preceding aRS1, late entry of specific punctuation, and of amino acids precursor to biosynthesis routes, or originated independently from other amino acids belonging in the code, preceding those derived from precursors. The succession of anticodon pairs is from the homogeneous (Ho) pDiN sector (•RR : •YY quadrants) to the mixed (Mx) sector (•RY : •YR) and from thermodynamically more to less stable pairs. Synthetases formed six couples of the same class in the eight pairs of boxes. In the four-stage scheme, the initial and final ones are composed, respectively, of aRS2-only and aRS1-only boxes, both aRS classes occurring in the pairs of the intermediate stages. Various other correlations indicated the physiologic relevance of the model. Stage 1 is composed by the Ho sector pairs •CC : •GG and •GA : •CU containing, respectively, Gly / Pro and Ser / Ser. It is indicated that early catalysts may not have been of the proteic synthetase type, due to absence of hydropathy correlation in the attributions. Gly may have preceded Pro in the •GG box, due to Pro being derived from Glu, which enters in Stage 2. The three amino acids in this stage contribute strongly to protein N-end stabilization and to RNA-binding motifs, so that a Progress in Biological Chirality 2 stabilized nucleoprotein system could be formed. Stage 2 completes the Ho sector, entering •UC : •AG and •UU : •AA. The central U boxes are initially occupied, respectively, by Asp and Asn. Asp conceded 5' Y to Glu and this with its •AG pair, occupied by Leu, formed an aRS1 couple. Asn conceded 5' Y to Lys, and this plus its •AA pair, occupied by Phe, formed the couple of atypical synthetases: some organisms use LysRS1 instead of class 2; PheRS is class 2 but aminoacylating in the aRS1 fashion, on the 2´ hydroxyl of the terminal adenosine of tRNA. Stage 3 initiates the Mx sector with the pairs •GC : •CG, •UG : •AC and •GU : •CA, containing the last aRS2 attributions together with the abundant entry of aRS1 and of amino acids characteristic of DNA binding motifs. The first and third pairs form couples of synthetases of different classes, AlaRS2 / ArgRS1 and ThrRS2 / CysRS1, TrpRS1. In the second pair, His enters the •UG box and concedes 5´ Y to Gln; this plus its •AC pair, occupied by Val, formed an aRS1 couple. From Thr onward, all amino acids were derived from biosynthesis families and the average amino acid set became more hydrophobic. In Stage 4, the tips of the Mx sector •AU : •UA are occupied by Ile, Met / Tyr and the punctuation signs fMet and X UA . tRNAs corresponding to all stop codons were fished by the initiation system, this with slipped pDiN (fMet CAU instead of Met CAU), generating conflicts solved with deletion of the X tRNAs. The aRS1 hexacodonic expansions were developed, LeuRS and ArgRS receiving, respectively, YAA and YCU. The succession of amino acids indicates that protein secondary structures developed increasing organization: the set characteristic of coils and turns was completed early, then that of helices, lastly the set of strands. In the model of recruitment of tRNA pairs, one of the members of a pair functions as codon to the other. In this way, Stage 1 was entirely self-referential and only coding for protein synthesis, not having been formed for decoding mRNAs. In Stage 2, the linear ordering of codon strings (primitive mRNAs, the first genes) was established, indicating the rise of genetic information: protein-stabilizing amino acids constituted the N-ends, destabilizers added as C-ends. It can be said that genes were defined simultaneously with the definition of products, at the moment of the binding of RNAs and proteins, in the process of formation of the coding system.

Proceedings of the …, 2009