Amoeba-based neurocomputing with chaotic dynamics
Communications of the ACM - ACM's plan to go online first
Amoeba-based nonequilibrium neurocomputer utilizing fluctuations and instability
UC'07 Proceedings of the 6th international conference on Unconventional Computation
Tug-of-war model for multi-armed bandit problem
UC'10 Proceedings of the 9th international conference on Unconventional computation
Hi-index | 0.00 |
An amoeboid organism, Physarum , exhibits rich spatiotemporal oscillatory behavior and various computational capabilities. Previously, the authors created a recurrent neurocomputer incorporating the amoeba as a computing substrate to solve optimization problems. In this paper, considering the amoeba to be a network of oscillators coupled such that they compete for constant amounts of resources, we present a model of the amoeba-based neurocomputer. The model generates a number of oscillation modes and produces not only simple behavior to stabilize a single mode but also complex behavior to spontaneously switch among different modes, which reproduces well the experimentally observed behavior of the amoeba. To explore the significance of the complex behavior, we set a test problem used to compare computational performances of the oscillation modes. The problem is a kind of optimization problem of how to allocate a limited amount of resource to oscillators such that conflicts among them can be minimized. We show that the complex behavior enables to attain a wider variety of solutions to the problem and produces better performances compared with the simple behavior.