An evolving ontogenetic cellular system for better adaptiveness
Introduction
In this paper, we present an original cellular system named Phuon. The main motivation behind this project is to go beyond classical cellular systems, such as cellular automata. CAs are powerful while remaining conceptually very simple, but they often lack adaptability,2 and are highly prone to synchronization and general failure. The idea here is to add ontogeny to cellularity, growth and development being means of adaptation and thus robustness.
In the first section, we detail the motivations behind the conception of Phuon. We also overview the general principles of the model and relate them to some inspiring works. In 3 A detailed description of the system: the developmental part, 4 The evolutionary engine, we describe the implementation of the system. This is not a technical description as such, though many technical details will be given. This section is rather an explanation of the inner workings of the system, where we discuss the questions surrounding such a system, including synchronization and growth. Finally, Section 5 presents the results obtained to this day with the system. More precisely, we will explain two tasks on which successful solvers were found, and demonstrate their robustness and adaptability qualities (or lack thereof).
The results are still preliminary but in our view they validate some of the hypotheses formulated in Section 2 and open the paths for many future research. The concluding Section 6, will be a discussion of the problems encountered, the limits and the capabilities of the system.
Section snippets
Motivations
Cellular interactions are at the foundation of the complex phenomena of life. Kennedy and Eberhart even argued in their recent book (Kennedy and Eberhart, 2001) that human intelligence and intelligence in general derives from “social” interactions, i.e., from the group. While not going this far, as we have shown (Capcarrere et al., 1996), even the simplest models of interactions, such as Cellular Automata, are capable of generating complex emergent behavior from simplicity. However the
A detailed description of the system: the developmental part
Phuon can be viewed as a two-layer cellular system. A passive environmental layer, and an active cellular layer. We will first present the former briefly and then describe the latter at length. In the second part, we will first explain the cell globally, then its language and its growth, to finally conclude with the question of synchronization.
The evolutionary engine
As exposed earlier, there are two motivations in using evolutionary computation techniques. The first one is practical. It is always very complex, and most often mathematically intractable to devise algorithms for complex systems. The second one is that though the aim here is to develop problem solvers, the questions related to the study of self-organization are also of interest. Evolutionary techniques often propose unexpected means of self-organization. Given the choice that a cell would be
Results
We present here the results obtained with Phuon on two simple tasks: Food foraging and Controlled growth. These results are still preliminary, but they provide a good opportunity to validate some of the hypotheses made about adaptiveness and fault-tolerance in cellular systems.
Concluding remarks
The results presented in this paper are interesting on many respects. However they remain preliminary and do not yet fulfill one of our original motivations: problem solving. We will now recapitulate what we can learn from these, and draw from there some paths for future research.
The food foraging example seems to demonstrate that such a developmental system can provide and even favor good qualities of adaptation: adaptation to a changing environment but also adaptation to faulty functioning.
References (20)
- Axelrod, R., 1997. Advancing the art of simulation in the social sciences. In: Conte, R., Hegselmann, Terna, P. (Eds.),...
- Bünzli, D.C., Capcarrere, M.S., 2001. Fault-tolerant structures: Towards robust self-replication in a probabilistic...
- Capcarrere, M.S., 2002. Cellular Automata and Other Cellular Systems: Design and Evolution. Ph.D. Thesis No. 2541,...
- Capcarrere, M.S., Sipper, M., Tomassini, M., December 1996. Two-state, r = 1 cellular automaton that classifies...
- Cramer, N.J., 1985. A representation of the adaptive generation of simple sequential program. In: Grefenstette, J.J....
- De Garis, H., 1999. Artificial embryology and cellular differentaition. In: Bentley, P. (Ed.), Evolutionary Design by...
- Fleischer, K., 1995. A Multiple Mechanism Developmental Model for Defining Self-Organing Structures. Ph.D. Thesis....
- et al.
Emergence of multicellular organisms with dynamic differentiation and spatial pattern
Artif. Life J.
(1998) - et al.
The ecology of echo
Artif. Life J.
(1997) - Kennedy, J., Eberhart, R.C., 2001. Swarm Intelligence. Morgan Kaufmann Publishers (secondary author: Y....
Cited by (2)
Self-repair ability of evolved self-assembling systems in cellular automata
2014, Genetic Programming and Evolvable MachinesA developmental approach to evolving scalable hierarchies for multi-agent swarms
2010, Proceedings of the 12th Annual Genetic and Evolutionary Computation Conference, GECCO '10 - Companion Publication
- 1
Present address: Computing Laboratory, University of Kent, Canterbury CT2 7NF, UK.