Haruya Matsumoto

A synaptic modification algorithm in considaeration of the generation of rhythmic oscillation in a ring neural network

In consideration of the generation of bursts of nerve impulses (that is, rhythmic oscillation in impulse density) in the ring neural network, a synaptic modification algorithm is newly proposed. Rhythmic oscillation generally occurs in the regular ring network with feedback inhibition and in fact such signals can be observed in the real nervous system. Since, however, various additional connections can cause a disturbance which easily extinguishes the rhythmic oscillation in the network, some function for maintaining the rhythmic oscillation is to be expected to exist in the synapses if such signals play an important part in the nervous system. Our preliminary investigation into the rhythmic oscillation in the regular ring network has led to the selection of the parameters, that is, the average membrane potential (AMP) and the average impulse density (AID) in the synaptic modification algorithm, where the decrease of synaptic strength is supposed to be essential. This synaptic modification algorithm using AMP and AID enables both the rhythmic oscillation and the non-oscillatory state to be dealt with in the algorithm without distinction. Simulation demonstrates cases in which the algorithm catches and holds the rhythmic oscillation in the disturbed ring network where the rhythmic oscillation was previously extinguished.

Ring neural network qua a generator of rhythmic oscillation with period control mechanism

For the ring neural network to function as a generator of rhythmic oscillation, mechanisms are required by which rhythmic oscillation is generated and maintained and then its period controlled. This paper demonstrates by simulation that those mechanisms can be actualized by employing a synaptic modification algorithm and by applying inputs from the outside to excitatory and inhibitory cells. When the constants in the synaptic modification algorithm are fixed, it is possible to select two modes, that is, the modification mode and the non-modification mode, using the excitatory input level to excitatory cells alone. This property solves the problem of the re-modification caused by the dispersion of AIDs (average impulse densities) with the application of the excitatory synchronous input to inhibitory cells.

Acoustic diagnosis for compressor with hybrid neural network

Describes an acoustic diagnosis technique for a compressor using a hybrid neural network (HNN). The HNN is composed of two neural networks: an acoustic feature extraction network, and a fault discrimination network. The acoustic feature extraction network uses an auto-associative neural network (ANN) whose target patterns are the same as the input patterns. The five-layered neural network is composed of two three-layered neural networks to compress the input information and to restore the compressed information. The authors examine the architecture of the ANN for acoustic diagnosis, the proper form of the activation function in the output layer and the proper number of hidden layers. The fault discrimination network uses a multilayered neural network whose input patterns are the output values of the hidden layer in the ANN. The authors examine the possibility of discriminating between eight types of compressor faults with high accuracy by using an HNN.<<ETX>>

A Hardware Syntactic Analysis Processor

An innovative algorithm for syntactic analysis could be the first step toward placing grammar on a chip.

Hybrid neural networks for acoustic diagnosis

Describes the acoustic diagnosis technique for a compressor using a hybrid neural network (HNN). The HNN is composed of two neural networks, an acoustic feature extraction network using a backpropagation network (BPN) and a fault discrimination network using a Gaussian potential function network (GPFN). The BPN is composed of five layers and the number of the middle hidden units is smaller than the others. The target patterns for the output layer are the same as the input patterns. After the learning of the network, the middle hidden layer acquires the compressed input information. The input patterns of the GPFN are the output values of the middle hidden layer in the BPN. The task of the HNN is to discriminate four conditions of the valve under various experimental conditions. As a result, 93.6% discrimination accuracy is obtained in this experiment. This suggests that the proposed model is effective for the acoustic diagnosis.

Basic dynamical properties of cross-coupled Hopfield nets

The authors clarify basic dynamical properties of cross-coupled Hopfield nets (CCHNs) using the simple CCHN in which two Hopfield nets (HNs) are connected to each other via two-layered feedforward neural networks. In the case that each HN composed of two units has two point attractors and each equilibrium state of one HN has a one-to-one mapping relation to that of the other, the authors investigate which of the equilibrium points the state parameters converge onto. They study the nature of the energy plane in one HN for the variation of two state-parameters in the other HN. The comparison with the original HN shows that energy planes in both HNs are dynamically varied by interactions of the network states. After sufficient learning, the CCHNs output converges onto a desired stable state so as to be satisfied with a given relation between two HNs. In the simple CCHN employed, the connection weights of internetworks can be regarded as those of the original HN itself. Therefore, the simple CCHN gives a method for determining connection weights in HNs efficiently.<<ETX>>