Evolving fractal gene regulatory networks for graceful degradation of software

• Authors: Peter J. Bentley
• Venue: Self-star Properties in Complex Information Systems
• Year: 2005

Fractal proteins are an evolvable method of mapping genotype to phenotype through a developmental process, where genes are expressed into proteins comprised of subsets of the Mandelbrot Set. The resulting network of gene and protein interactions can be designed by evolution to produce specific patterns that in turn can be used to solve problems. In this paper, adaptive developmental programs, capable of developing different solutions in response to different signals from an environment, are investigated. The evolvability of solutions and the capability of these solutions to survive damage is assessed. Evolution is used to create a fractal gene regulatory network (GRN) thatcalculates the squareroot of the input (its environment). This is compared with a GP-evolved squareroot function and a human-designed squareroot function. The programs are damaged by corrupting their compiled executable code, and the ability for each of them to survive such damage is assessed. Experiments demonstrate that only the evolutionary developmental code shows gracefuldegradation after damage. This provides evidence that software based on gene, protein and cellular computation is far more robust than traditional methods. Like a multicellular organism, with its genes evolved and developed, it shows graceful degradation. Should it be damaged, it is designed to continue to work.