Pulsed Neural Networks Research Articles

パルスニューラルネットワークにおけるネットワーク拡張型強化学習アルゴリズム

In this paper, we propose a new hierarchical pulse neural network and its reinforcement learning algorithm with network extension. The proposed pulse neural network has three layers, and all of the neurons are pulse neurons. This network learns relations between input pulse sequences and the desired outputs by updating connection weights and by adding neurons dynamically. We carried out a computer simulation to confirm the performance of the proposed algorithm.

Compound binomial processes in neural integration

Explores some of the properties of stochastic digital signal processing in which the input signals are represented as sequences of Bernoulli events. The event statistics of the resulting stochastic process may be governed by compound binomial processes, depending upon how the individual input variables to a neural network are stochastically multiplexed. Similar doubly stochastic statistics can also result from datasets which are Bernoulli mixtures, depending upon the temporal persistence of the mixture components at the input terminals to the network. The principal contribution of these results is in determining the required integration period of the stochastic signals for a given precision in pulsed digital neural networks.

Stochastic resonance in a pulse neural network with a propagational time delay

Stochastic resonance in a coupled FitzHugh–Nagumo equation with a propagational time delay is investigated. With an appropriate set of parameter values, i.e. the frequency of the periodic input, the propagational time delay, and the coupling strength, a deterministic firing induced by additive noise is observed, and its dependence on the number of neurons is examined. It is also found that a network composed of two assemblies shows a competitive behavior under control of the noise intensity.

Associative memory retrieval induced by fluctuations in a pulsed neural network.

An associative memory retrieval in a pulsed neural network composed of the FitzHugh-Nagumo neurons is investigated. The memory is represented in the spatio-temporal firing pattern of the neurons, and the memory retrieval is accomplished using the fluctuations in the system. The storage capacity of the network is investigated numerically. It is demonstrated that this pulsed neural network is capable of an alternate retrieval of two patterns.

Separation and purification of useful proteins using hydrogel ultrafiltration

The hydrogel process is a different form of ultrafiltration and has been used to separate biological molecules. In this study, the gel pore size was predicted by pulse NMR technique and neural network using a data base obtained from gel filtration chromatography and diffusion experiment. Recombinant alkaline phosphatase expressed in insect cells was concentrated 1.5 times by hydrogel ultrafiltration by swelling at 20°C and collapsing at 35°C at 53–65% separation efficiency and 78–83% enzyme recovery. Wild and recombinantAutographa californica nuclear polyhedrosis viruses (AcNPV) were also concentrated 1.4 and 1.6 times of the feed solution at 48.5 and 60.0% separation efficiency, respectively. Hydrogel ultrafiltration appears to be an attractive alternative for the concentration of AcNPV and recombinant proteins from insect cells.

パルスニューラルネットワークにおける相関コーディングとディレイ適応

パルスニューラルネットワークにおける相関コーディングとディレイ適応

Prediction of hydrogel pore size by pulse NMR and neural networks

Temperature-sensitive hydrogels swell and collapse reversibly in aqueous solution. The hydrogel process is a different form of ultrafiltration and has been used to separate biological molecules. In this study, the gel pore size was predicted by pulse NMR technique and neural network using a data base obtained from gel filtration chromatography and diffusion experiment.

A pulsed neural network capable of universal approximation

The authors describe a pulsed network version of the cerebellar model articulation controller (CMAC), popularized by Albus (1981). The network produces output pulses whose times of occurrence are a function of input pulse intervals. Within limits imposed by causality conditions, this function can approximate any bounded measurable function on a compact domain. Simulation results demonstrate the viability of training the network with a least mean square algorithm.