A Novel Hybrid Spiking Neuron: Response Analysis and Learning Potential.
ABSTRACT In this paper, we propose a hybrid spiking neuron which can exhibit various bifurcation phenomena and response characteristics
of inter spike intervals. Using a discrete/continuous-states hybrid map, we can clarify typical bifurcation mechanisms and
can analyze the response characteristics. In addition, we propose a learning algorithm of the hybrid spiking neuron and show
that the neuron can approximate given response characteristics of inter spike intervals.
- SourceAvailable from: hosei.ac.jp[show abstract] [hide abstract]
ABSTRACT: A digital spiking neuron is a wired system of shift registers and can generate various spike-trains by adjusting the wiring pattern. In this paper we analyze the basic relations between the wiring pattern and characteristics of the spike-train. Based on the relations, we present a learning algorithm which utilizes successive changes of the wiring pattern. It is shown that the neuron can reproduce spike-trains of another neuron which has an unknown wiring pattern. It is also shown that the neuron can approximate various spike-trains of a chaotic analog spiking neuron.Neural Networks 01/2008; 21(2-3):140-9. · 1.93 Impact Factor
Conference Proceeding: A hardware-oriented learning algorithm for a digital spiking neuron[show abstract] [hide abstract]
ABSTRACT: The digital spiking neuron is a wired system of shift registers and behaves like a simplified neuron model. By adjusting the wirings among the registers, the neuron can generate various spike-trains. In this paper some basic relations between the wiring pattern and spike-train characteristics are analyzed. Based on the analysis results, a hardware-oriented learning algorithm is proposed. The learning algorithm and the digital neuron are implemented by a hardware description language (HDL). It is shown that the learning algorithm enables the digital neuron to approximate various spike-trains generated by an analog spiking neuron model. In addition, some basic experimental measurements are provided by using a field programmable gate array (FPGA).Neural Networks, 2008. IJCNN 2008. (IEEE World Congress on Computational Intelligence). IEEE International Joint Conference on; 07/2008
- [show abstract] [hide abstract]
ABSTRACT: We present a novel digital spiking neuron that can generate various periodic spike trains depending on initial states. The spike trains can be symbolized by digital codes based on spike-position modulation. It is theoretically clarified that different types of codes (e.g., binary and Gray number codes) can be realized by adjusting system parameters. We also present a multiplex communication system. A transmitter neuron can map multiple digital codes into a single spike train. A pulse-coupled network of the neurons accepts the spike train and can retrieve the digital codes based on a synchronization phenomenon. Typical phenomena are confirmed by HDL simulationCircuits and Systems II: Express Briefs, IEEE Transactions on 09/2006; · 1.33 Impact Factor