We discuss the concept of extradimensional bypass as it was developed by the late theoretical biologist Michael Conrad. An evolving system that optimizes its performance by gradient ascent (hill climbing) can avoid being trapped in local maxima by increasing the effective dimensionality of its search space. Many local maxima may become saddle points in the higher dimensional space, such that gradient ascent can continue unimpeded. Extradimensional bypass as a concept has parallels in theories of open-ended learning and functional emergence, where new structural, functional, and informational primitives can increase the effective dimensionality of material systems.
Download full-text PDF |
Source |
|---|---|
| http://dx.doi.org/10.1016/s0303-2647(01)00174-5 | DOI Listing |
Publication Analysis
Top Keywords
Similar Publications
Studies on Manfred Eigen's model for the self-organization of information processing.
Eur Biophys J
May 2018
Leibniz Institute for Baltic Research, 18119, Rostock, Germany.
In 1971, Manfred Eigen extended the principles of Darwinian evolution to chemical processes, from catalytic networks to the emergence of information processing at the molecular level, leading to the emergence of life. In this paper, we investigate some very general characteristics of this scenario, such as the valuation process of phenotypic traits in a high-dimensional fitness landscape, the effect of spatial compartmentation on the valuation, and the self-organized transition from structural to symbolic genetic information of replicating chain molecules. In the first part, we perform an analysis of typical dynamical properties of continuous dynamical models of evolutionary processes.
View Article and Find Full Text PDFExtradimensional bypass.
Biosystems
January 2002
Eaton Peabody Laboratory of Auditory Physiology, Department of Otology and Laryngology, Harvard Medical School, Massachusetts Eye and Ear Infirmary, 243 Charles Street, Boston, MA 02114, USA.
We discuss the concept of extradimensional bypass as it was developed by the late theoretical biologist Michael Conrad. An evolving system that optimizes its performance by gradient ascent (hill climbing) can avoid being trapped in local maxima by increasing the effective dimensionality of its search space. Many local maxima may become saddle points in the higher dimensional space, such that gradient ascent can continue unimpeded.
View Article and Find Full Text PDFWant AI Summaries of new PubMed Abstracts delivered to your In-box?
Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!