NECSI Resources

 International Conference on Complex Systems (ICCS2007)

Wavelike Design of Social Agents Simulated as System of Interacting Net of Neural Networks

Darius Plikynas
Vilnius Management Academy

     Full text: PDF
     Last modified: September 26, 2007

Abstract
Instead of following particles like interaction scheme between agents in the social domain, we have abandoned mechanistic and proposed conceptually wavelike approach. In fact, we assumed that the major modern capital accumulation is in the form of knowledge (information and environmental resources), which, being dimensionless according to information theory, have some similarities with energy. Therefore, we constructed universal rules to represent agentís various wealth and behavioral patterns in the form of wave packets composing an agentís unique energy spectrum. In practical terms, proposed model rests on the net of simple backpropagation NN, which are designed for simulation of agentsí energy spectra, where each NN receives, transforms and transmits the spectra bands, following initial training on a priori settled rules. Hence, our simulated economic system rests on the economic model, which is based on the set of horizontal and vertical production rules used appropriately for NN training. The simulation model is also linked with some practical aspects targeted for empirical validation and model realization.

Appendices
      Fully functional framework for collaborative research, data&algorithm exchange and shared computing based on the proposed approach is going to be available on-line soon.







Maintained by NECSI Webmaster Copyright © 2000-2007 New England Complex Systems Institute. All rights reserved.