Jiang Mingyan

A fast hybrid algorithm of global optimization for feedforward neural networks

This paper presents the hybrid algorithm of global optimization of dynamic learning rate for multilayer feedforward neural networks (MLFNN). The effect of inexact line search on conjugacy was studied and a generalized conjugate gradient method based on this effect was proposed and shown to have global convergence for error backpropagation of MLFNN. The descent property and global convergence was given for the improved hybrid algorithm of the conjugate gradient algorithm, the results of the proposed algorithm show a considerable improvement over the Fletcher-Reeves algorithm and the conventional backpropagation (BP) algorithm, it overcomes the drawback of conventional BP and Polak-Ribieve conjugate gradient algorithm that maybe plunge into local minima.

A multi-grade mean morphologic edge detection

A method of edge detection based on multi-grade mean mathematical morphology is put forward. The properties of the edge detector are analyzed, and other morphologic methods are compared with it. The method results in good edge detection and at the same time raises the de-noising capacity.

Some methods of image enlarging with interpolation

In this paper some methods are given for the gray images to be 2/spl times/ enlarged with interpolation. After comparison, the methods of wavelet and fractal are more suitable to the processing of image enlarging.

Diabetes expert system

This medical expert system is applied to the diagnosis, treatment, and teaching of diabetes. It adopts a forward, backward and forward-backward chaining inference mechanism and an uncertainty handling method which can quickly and efficiently, based on the patients symptoms, judge the possibility of illness, its severity, and its potential complications. Based upon this assessment the system gives prescriptions for treatment and makes useful suggestions. The system can also be used in teaching practice.

The new method of reducing the NMR image's ringing artifacts by wavelet transform

It is difficult to reduce an NMR images ringing artifacts and obtain better images. This paper uses the method of dyadic wavelets and self-adaptive threshold to process the images that have the ringing artifacts. The test results prove that the method is effective and it can process the NMR images ringing artifacts. Finally the paper discusses using the fuzzy method.

Analysis of the effects of quantization in high-order function neural network

A statistical quantization model is used to analyze of the effects of quantization in digital implementation of high-order function neural network. From the theory we analyse the performance degradation and fault tolerance of the neural network caused by the number of quantization bits and by changing the order. We try to predict the error in the high-order function neural network (HOFNN) given the properties of the network and the number of quantization bits. Experimental results show the error rate is inversely proportional to quantized bits M for HRFNN. The recognition performance of the backpropagation (BP) network and the HRFNN are almost the same for different quantization bits. The networks performance degradation gets worse when the number of bits is lower than 4-bit quantization. The networks performance degradation gets worse when the number of bits is lower than 4-bit quantization.

The upper limit of the stability of delay-type cellular neural networks

This paper gives an improved sufficient condition of ensuring the stability of delayed cellular neural networks (DCNN) by means of the analysis of a correspondent Lyapunov function. This result improved the limit given by Civalleri and gives an optional upper limit of the stability of DCNN.