This paper studies the simple dynamic binary neural network characterized by the signum activation function, ternary weighting parameters and integer threshold parameters. The network can be regarded as a digital version of the recurrent neural network and can output a variety of binary periodic orbits. The network dynamics can be simplified into a return map, from a set of lattice points, to itself. In order to store a desired periodic orbit, we present two learning algorithms based on the correlation learning and the genetic algorithm. The algorithms are applied to three examples: a periodic orbit corresponding to the switching signal of the dc-ac inverter and artificial periodic orbit. Using the return map, we have investigated the storage of the periodic orbits and stability of the stored periodic orbits.
Ryota KOUZUKI
Hosei University
Toshimichi SAITO
Hosei University
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Ryota KOUZUKI, Toshimichi SAITO, "Learning of Simple Dynamic Binary Neural Networks" in IEICE TRANSACTIONS on Fundamentals,
vol. E96-A, no. 8, pp. 1775-1782, August 2013, doi: 10.1587/transfun.E96.A.1775.
Abstract: This paper studies the simple dynamic binary neural network characterized by the signum activation function, ternary weighting parameters and integer threshold parameters. The network can be regarded as a digital version of the recurrent neural network and can output a variety of binary periodic orbits. The network dynamics can be simplified into a return map, from a set of lattice points, to itself. In order to store a desired periodic orbit, we present two learning algorithms based on the correlation learning and the genetic algorithm. The algorithms are applied to three examples: a periodic orbit corresponding to the switching signal of the dc-ac inverter and artificial periodic orbit. Using the return map, we have investigated the storage of the periodic orbits and stability of the stored periodic orbits.
URL: https://global.ieice.org/en_transactions/fundamentals/10.1587/transfun.E96.A.1775/_p
Copy
@ARTICLE{e96-a_8_1775,
author={Ryota KOUZUKI, Toshimichi SAITO, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={Learning of Simple Dynamic Binary Neural Networks},
year={2013},
volume={E96-A},
number={8},
pages={1775-1782},
abstract={This paper studies the simple dynamic binary neural network characterized by the signum activation function, ternary weighting parameters and integer threshold parameters. The network can be regarded as a digital version of the recurrent neural network and can output a variety of binary periodic orbits. The network dynamics can be simplified into a return map, from a set of lattice points, to itself. In order to store a desired periodic orbit, we present two learning algorithms based on the correlation learning and the genetic algorithm. The algorithms are applied to three examples: a periodic orbit corresponding to the switching signal of the dc-ac inverter and artificial periodic orbit. Using the return map, we have investigated the storage of the periodic orbits and stability of the stored periodic orbits.},
keywords={},
doi={10.1587/transfun.E96.A.1775},
ISSN={1745-1337},
month={August},}
Copy
TY - JOUR
TI - Learning of Simple Dynamic Binary Neural Networks
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 1775
EP - 1782
AU - Ryota KOUZUKI
AU - Toshimichi SAITO
PY - 2013
DO - 10.1587/transfun.E96.A.1775
JO - IEICE TRANSACTIONS on Fundamentals
SN - 1745-1337
VL - E96-A
IS - 8
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - August 2013
AB - This paper studies the simple dynamic binary neural network characterized by the signum activation function, ternary weighting parameters and integer threshold parameters. The network can be regarded as a digital version of the recurrent neural network and can output a variety of binary periodic orbits. The network dynamics can be simplified into a return map, from a set of lattice points, to itself. In order to store a desired periodic orbit, we present two learning algorithms based on the correlation learning and the genetic algorithm. The algorithms are applied to three examples: a periodic orbit corresponding to the switching signal of the dc-ac inverter and artificial periodic orbit. Using the return map, we have investigated the storage of the periodic orbits and stability of the stored periodic orbits.
ER -