Compositional pattern producing networks: A novel abstraction of development

  title={Compositional pattern producing networks: A novel abstraction of development},
  author={Kenneth O. Stanley},
  journal={Genetic Programming and Evolvable Machines},
  • Kenneth O. Stanley
  • Published 1 June 2007
  • Computer Science
  • Genetic Programming and Evolvable Machines
Natural DNA can encode complexity on an enormous scale. Researchers are attempting to achieve the same representational efficiency in computers by implementing developmental encodings, i.e. encodings that map the genotype to the phenotype through a process of growth from a small starting point to a mature form. A major challenge in in this effort is to find the right level of abstraction of biological development to capture its essential properties without introducing unnecessary inefficiencies… 

A Compiler for CPPNs: Transforming Phenotypic Descriptions Into Genotypic Representations

  • S. Risi
  • Computer Science
    AAAI Fall Symposia
  • 2013
The concept of a CPPN-Compiler, which allows the user to directly compile a high-level description of the desired starting structure into the CPPNs itself, is introduced, which opens up a new research direction in GDS, in which specialized CPPn-Compilers for different domains could help to overcome the black box of evolutionary optimization.

CA-NEAT: Evolved Compositional Pattern Producing Networks for Cellular Automata Morphogenesis and Replication

This paper proposes a new principle of morphogenesis based on compositional pattern producing networks (CPPNs), an abstraction of development that has been able to produce complex structural motifs without local interactions to provide a valuable mapping for morphogenetic systems, beyond CA systems, where development through local interactions is desired.

Evolutionary Development based on Genetic Regulatory Models for Behavior-Finding

Supporting the latest biological hypothesis, the findings show that the sole variation in the regulation of gene expression is indeed a sufficient condition for the evolutionary emergence of novelty and diversity.

Evolutionary development of tensegrity structures

Artificial evolutionary development

The developmental process can be seen as a genotype-to-phenotype mapping, and it is possible to describe development detached from the optimization method as a black-box that has a vector of object variables as input, and returns a spatial distribution of cells ready for evaluation.

On the Performance of Indirect Encoding Across the Continuum of Regularity

This paper presents the first comprehensive study showing that phenotypic regularity enables an indirect encoding to outperform direct encoding controls as problem regularity increases, and suggests a path forward that combines indirect encodings with a separate process of refinement.

Genotype–phenotype mapping and the end of the ‘genes as blueprint’ metaphor

  • M. Pigliucci
  • Biology
    Philosophical Transactions of the Royal Society B: Biological Sciences
  • 2010
The concept of ‘developmental encoding’ (as opposed to the classical one of genetic encoding) provides a promising computational–theoretical underpinning to coherently integrate ideas on evolvability, modularity and robustness and foster a fruitful framing of the G→P mapping problem.

Evolving Novel Cellular Automaton Seeds Using Compositional Pattern Producing Networks (CPPN)

These results stand as the first baseline for further exploration into encoding cellular automata using CPPN, a variation of artificial neural networks known to evolve organic patterns when used to process visual data.

Does Aligning Phenotypic and Genotypic Modularity Improve the Evolution of Neural Networks?

Results suggest encouraging modularity in both the genotype and phenotype as an important step towards solving large-scale multi-modal problems, but also indicate that more research is required before the authors can evolve structurally organized networks to solve tasks that require multiple, different neural modules.

Evolving three-dimensional objects with a generative encoding inspired by developmental biology

This paper describes how 3D shapes can be evolved and then transferred to reality via 3D printing technology and shows that evolving objects with encodings based on concepts from biological development can be a powerful way to evolve complex, interesting objects.



Exploiting Morphological Conventions for Genetic Reuse

It is believed that that duplication of genetic modules is not the primary evolutionary factor leading to reuse of phenotypic structures, and natural evolution first establishes a morphological convention, such as bilateral symmetry, and then exploits that convention as a framework for repetition of phenotypesic features.

A Taxonomy for Artificial Embryogeny

This taxonomy provides a unified context for long-term research in AE, so that implementation decisions can be compared and contrasted along known dimensions in the design space of embryogenic systems, and allows predicting how the settings of various AE parameters affect the capacity to efficiently evolve complex phenotypes.

Multi-cellular Development: Is There Scalability and Robustness to Gain?

Analysis of scalability and robustness to phenotypic faults of two developmental systems inspired by Cellular Automata and Artificial Embryogeny shows that, while for direct encoding scalability is limited by the size of the search space, developmental systems performance appears to be related to the amount of regularity that they can extract from the phenotype.

Repeated structure and dissociation of genotypic and phenotypic complexity in artificial ontogeny

It is demonstrated that evolved genetic regulatory networks in AO give rise to hierarchical, repeated phenotypic structures, and the claim that artificial ontogeny is a useful design tool for the evolutionary design of virtual agents and real-world robots is supported.

A connectionist model of development.

Creating High-Level Components with a Generative Representation for Body-Brain Evolution

Applying GENRE to the task of evolving robots for locomotion and comparing it against a non-generative (direct) representation shows that the generative representation system rapidly produces robots with significantly greater fitness.

Growing neural networks

Simulation results shed some light on why modular architectures are likely to emerge, why similar successions of stages tend to appear in both evolution and development, and why a developmental age is preserved evolutionarily although the mature state may appear to be more efficient from the point of view of fitness.

Evolving a neurocontroller through a process of embryogeny

The New AI hypothesizes that intelligent behaviour must be understood within the framework provided by the agent’s physical interactions with the environment: subjective sensations and bodily interactions, and proposes a bottom-up exploration, which starts from the lowest adaptive mechanisms to reach the topmost cognitive abilities.

A Developmental Model for the Evolution of Artificial Neural Networks

This work presents a model of decentralized growth and development for artificial neural networks (ANNs), inspired by developmental biology and the physiology of nervous systems, and demonstrates the power of the artificial chemistry by analyzing engineered genomes that lead to the growth of simple networks with behaviors known from physiology.

Cell division and migration in a ‘genotype’ for neural networks

This paper presents a model of neural development which includes cell division and cell migration in addition to axonal growth and branching, and finds that mutations that affect different phases of development have very different evolutionary consequences.