具有自适应类警戒参数的模糊ARTMAP神经网络

提出了一种具有自适应类警戒参数的模糊ＡＲＴＭＡＰ神经网络,为不同的模糊ＡＲＴ的类族设置了不同的警戒测试参数,并在学习过程中进行适应调整。还提出了新的非交叠超方形以及非交叠的Ｎｅｓｔｅｄ超主形的建立与扩展学习规则。     

Fingerprint minutiae filtering using ARTMAP

Fingerprints are widely used for unique personal identification based on minutiae matching. Minutiae are the terminations and bifurcations of ridges in a fingerprint image. Generally fingerprint images are of low quality due to the presence of noise and contrast deficiency resulting in discontinuity in ridges producing false minutiae points. It is worth noting that there is a fundamental difference between a neural network (NN) approach for minutiae location and minutiae filtering. In this paper, the spurious minutiae points and the bug pixels introduced during the thinning process are eliminated based on the neighborhood pixel information. A new minutiae filtering algorithm using a NN is introduced to improve the accuracy of the extraction algorithm proposed in the literature. Each minutia, as detected by the algorithm, is classified through ARTMAP NN whose output indicates whether it is a termination, a bifurcation or a false minutia. Experimental results show that the efficiency of minutiae classification has significantly improved using the proposed filtering algorithm.     

基于模糊自适应共振理论映射算法的单样本三维人脸识别

针对传统的三维人脸识别分类算法大多需要多个样本进行训练,而在单训练样本的前提下识别性能会严重降低的问题,提出了基于模糊自适应共振理论映射(Fuzzy ARTMAP)的算法对三维人脸数据库进行分类识别。首先对三维人脸深度图像进行局部二值模式(LBP)统一模式算子的特征提取,再对LBP特征进行Log-Gabor小波变换,提取图像的频域特征向量作为训练的输入向量,最后将单样本训练向量集送入Fuzzy ARTMAP分类器进行训练识别。该算法在FRGC v2.0三维人脸数据库中的识别率可达到87.15%,分类器的训练时间为24.88s,单张待识别人脸样本与单张已注册的人脸匹配时间为0.0015s,一张新的人脸样本在数据库完成一次搜索匹配则需要1.08s。实验结果表明,所提方法在测试中的性能优于概率神经网络(PNN)和极限学习机神经网络(ELM),既能保证较高的识别率,又能拥有较短的训练时间,且时间增幅稳定,可控性强。     

Deformation invariant visual object recognition: Experiments with a self-organising neural architecture

A self-organising neural network architecture for grey-scale visual object rcognition is presented. The network is composed of three processing layers with an architecture designed to give deformation tolerance. The processing layers involve feature extraction, sub-pattern detection and classification. Training is generally performed on-line in an unsupervised manner, classes being created when objects are presented that cannot be classified. The results given show the effect of the two discrimination parameters when the network is applied to two very different sets of images, namely hand written numerals and hand gestures images. The sensitivity of the network to the parameters that govern the size of detectable patterns and the areas over which they are detected is also tested. The robustness of the network to the order of image presentation is also demonstrated. The results show that parameter choice is not critical and heuristically chosen parameters provide near optimum performance.     

L-p Fuzzy ARTMAP neural network architecture

In this paper, an L-p based Fuzzy ARTMAP neural network is presented. The category choice of this network is based on the L-p norm. Geometrical properties of this architecture are presented. Comparisons between this category choice and the category choice of the Fuzzy ARTMAP are illustrated. And simulation results on the databases taken from the UCI repository are performed. It will be shown that using the L-p norm is geometrically more attractive. It will operate directly on the input patterns without the need for doing any preprocessing. It should be noted that the Fuzzy ARTMAP architecture requires two preprocessing steps: normalization and complement coding. Simulation results on different databases show the good generalization performance of the L-p Fuzzy ARTMAP compared to the performance of Fuzzy ARTMAP.     

Probabilistic ensemble simplified fuzzy ARTMAP for sonar target differentiation

This study investigates the processing of sonar signals with ensemble neural networks for robust recognition of simple objects such as plane, corner and trapezium surface. The ensemble neural networks can differentiate the target objects with high accuracy. The simplified fuzzy ARTMAP (SFAM) and probabilistic ensemble simplified fuzzy ARTMAP (PESFAM) are compared in terms of classification accuracy. The PESFAM implements an accurate and effective probabilistic plurality voting method to combine outputs from multiple SFAM classifiers. Five benchmark data sets have been used to evaluate the applicability of the proposed ensemble SFAM network. The PESFAM achieves good accuracy based on the twofold cross-validation results. In addition, the effectiveness of the proposed ensemble SFAM is delineated in sonar target differentiation. The experiments demonstrate the potential of PESFAM classifiers in offering an optimal solution to the data-ordering problem of SFAM implementation and also as an intelligent classification tool in mobile robot application.     

A Fast Simplified Fuzzy ARTMAP Network

We present an algorithmic variant of the simplified fuzzy ARTMAP (SFAM) network, whose structure resembles those of feed-forward networks. Its difference with Kasuba's model is discussed, and their performances are compared on two benchmarks. We show that our algorithm is much faster than Kasuba's algorithm, and by increasing the number of training samples, the difference in speed grows enormously.The performances of the SFAM and the MLP (multilayer perceptron) are compared on three problems: the two benchmarks, and the Farsi optical character recognition (OCR) problem. For training the MLP two different variants of the backpropagation algorithm are used: the BPLRF algorithm (backpropagation with plummeting learning rate factor) for the benchmarks, and the BST algorithm (backpropagation with selective training) for the Farsi OCR problem.The results obtained on all of the three case studies with the MLP and the SFAM, embedded in their customized systems, show that the SFAM's convergence in fast-training mode, is faster than that of MLP, and online operation of the MLP is faster than that of the SFAM. On the benchmark problems the MLP has much better recognition rate than the SFAM. On the Farsi OCR problem, the recognition error of the SFAM is higher than that of the MLP on ill-engineered datasets, but equal on well-engineered ones. The flexible configuration of the SFAM, i.e. its capability to increase the size of the network in order to learn new patterns, as well as its simple parameter adjustment, remain unchallenged by the MLP.     

Optimising memory usage in n-tuple neural networks

The use of n-tuple or weightless neural networks as pattern recognition devices is well known (Aleksander and Stonham, 1979). They have some significant advantages over the more common and biologically plausible networks, such as multi-layer perceptrons; for example, n-tuple networks have been used for a variety of tasks, the most popular being real-time pattern recognition, and they can be implemented easily in hardware as they use standard random access memories.

In operation, a series of images of an object are shown to the network, each being processed suitably and effectively stored in a memory called a discriminator. Then, when another image is shown to the system, it is processed in a similar manner and the system reports whether it recognises the image; is the image sufficiently similar to one already taught?

If the system is to be able to recognise and discriminate between m-objects, then it must contain m-discriminators. This can require a great deal of memory.

This paper describes various ways in which memory requirements can be reduced, including a novel method for multiple discriminator n-tuple networks used for pattern recognition. By using this method, the memory normally required to handle m-objects can be used to recognise and discriminate between 2^m — 2 objects.     

Performance comparison of neural network models: backpropagation vs. fuzzy artmap

Neural networks have been increasingly applied to many problems in civil engineering. Even though there are currently many different types of neural network models, Backpropagation is the most popular neural network model. It is also known that Fuzzy ARTMAP, which is a combination of fuzzy logic and Adaptive Resonance Theory (ART), is superior to any other neural network models in terms of computing cost and predictive accuracy. In this research, two neural network paradigms, Backpropagation and Fuzzy ARTMAP have been studied to compare their performance in terms of computing cost and predictive accuracy through the experiment with real world image data of traffic scenes, as well as biological and theoretical aspects. In addition, three enhanced Backpropagation models, Backpropagation with Momentum, Quickprop, BPMP (Backpropagation with Momentum and Prime-offset) have been considered to compare the network performance of each model.     

规则抽取的Fuzzy ARTMAP方法

着重阐述了如何使用有教师监督的自组织神经网络－模糊自适应共振映射网络（Ｆｕｚｚｙ ＡＲＴＭＡＰ）从例子中抽取知识规则。叙述了规则抽取中的两个细节：网络修剪，即删除那些对网络抽取规则贡献不大的节点及其相连的权值；权值的量化，以使系统最终能释译成一套可使用的规则。本文对Ｆｕｚｚｙ ＡＲＴＭＡＰ网络作了改进和简化，并用于医学上心电图（ＥＣＧ）信号中室性早搏（ＰＶＣ）诊断规则的自动获取，取得了比较满意的结     

An investigation of the modified direction feature for cursive character recognition

This paper describes and analyses the performance of a novel feature extraction technique for the recognition of segmented/cursive characters that may be used in the context of a segmentation-based handwritten word recognition system. The modified direction feature (MDF) extraction technique builds upon the direction feature (DF) technique proposed previously that extracts direction information from the structure of character contours. This principal was extended so that the direction information is integrated with a technique for detecting transitions between background and foreground pixels in the character image.In order to improve on the DF extraction technique, a number of modifications were undertaken. With a view to describe the character contour more effectively, a re-design of the direction number determination technique was performed. Also, an additional global feature was introduced to improve the recognition accuracy for those characters that were most frequently confused with patterns of similar appearance. MDF was tested using a neural network-based classifier and compared to the DF and transition feature (TF) extraction techniques. MDF outperformed both DF and TF techniques using a benchmark dataset and compared favourably with the top results in the literature. A recognition accuracy of above 89% is reported on characters from the CEDAR dataset.     

Multi-class pattern classification using neural networks

Multi-class pattern classification has many applications including text document classification, speech recognition, object recognition, etc. Multi-class pattern classification using neural networks is not a trivial extension from two-class neural networks. This paper presents a comprehensive and competitive study in multi-class neural learning with focuses on issues including neural network architecture, encoding schemes, training methodology and training time complexity. Our study includes multi-class pattern classification using either a system of multiple neural networks or a single neural network, and modeling pattern classes using one-against-all, one-against-one, one-against-higher-order, and P-against-Q. We also discuss implementations of these approaches and analyze training time complexity associated with each approach. We evaluate six different neural network system architectures for multi-class pattern classification along the dimensions of imbalanced data, large number of pattern classes, large vs. small training data through experiments conducted on well-known benchmark data.     

Fuzzy ARTMAP classification of invariant features derived using angle of rotation from a neural network

Conventional regular moment functions have been proposed as pattern sensitive features in image classification and recognition applications. But conventional regular moments are only invariant to translation, rotation and equal scaling. It is shown that the conventional regular moment invariants remain no longer invariant when the image is scaled unequally in the x- and y-axis directions. We address this problem by presenting a technique to make the regular moment functions invariant to unequal scaling. However, the technique produces a set of features that are only invariant to translation, unequal/equal scaling and reflection. They are not invariant to rotation. To make them invariant to rotation, moments are calculated with respect to the principal axis of the image. To perform this, the exact angle of rotation must be known. But the method of using the second-order moments to determine this angle will also be inclusive of an undesired tilt angle. Therefore, in order to correctly determine the amount of rotation, the tilt angle which differs for different scaling factors in the x- and y-axis directions for the particular image must be obtained. In order to solve this problem, a neural network using the back-propagation learning algorithm is trained to estimate the tilt angle of the image and from this the amount of rotation for the image can be determined. Next, the new moments are derived and a Fuzzy ARTMAP network is used to classify these images into their respective classes. Sets of experiments involving images rotated and scaled unequally in the x- and y-axis directions are carried out to demonstrate the validity of the proposed technique.     

Application of fuzzy ARTMAP and fuzzy <Emphasis Type="Italic">c</Emphasis>-means clustering to pattern classification with incomplete data

In this paper, a hybrid neural network that is capable of incremental learning and classification of patterns with incomplete data is proposed. Fuzzy ARTMAP (FAM) is employed as the constituting network for pattern classification while fuzzy c-means (FCM) clustering is used as the underlying algorithm for processing training as well as test samples with missing features. To handle an incomplete training set, FAM is first trained using complete samples only. Missing features of the training samples are estimated and replaced using two FCM-based strategies. Then, network training is conducted using all the complete and estimated samples. To handle an incomplete test set, a non-substitution FCM-based strategy is employed so that a predicted output can be produced rapidly. The performance of the proposed hybrid network is evaluated using a benchmark problem, and its practical applicability is demonstrated using a medical diagnosis task. The results are compared, analysed and quantified statistically with the bootstrap method. Implications of the proposed network for pattern classification tasks with incomplete data are discussed.     

Analysis and application of a self-organising image recognition neural network

This paper investigates the effect of various feature extraction methods on the recognition ability of a self-organising neural network called Paradise when applied to the problems of the classification of face images and hand written character recognition. The feature extraction methods investigated are, oriented Gaussian filters, Gabor filters and oriented Laplacian of Gaussian (LG) filters. The recognition results for the two applications are shown to compare favourably with other techniques designed specifically for the two tasks.     

神经网络VC维计算研究

1 引言神经网络技术已经在很多领域得到了成功的应用.但由于神经网络并不具有一个统一的理论框架,其经验性成分相当高,这对其进一步发展造成了极大的阻碍。如果能为神经网络应用提供一些指导性的分析方法,不仅将促进该领域的理论研究,还可以在应用     

Spectral recognition using a modified Eckhorn neural network model

Studies of the visual cortex of the cat highlight the role of temporal processing using synchronous oscillations for object identification. In this paper, the original neural network model of Eckhorn has been modified according to the proposal of Johnson and others and used for spectral recognition. The method developed turns out to be a much simpler, faster and elegant way of spectral recognition than reported elsewhere.     

A novel fault diagnosis method of bearing based on improved fuzzy ARTMAP and modified distance discriminant technique

This paper presents a novel intelligent diagnosis method based on multiple domain features, modified distance discrimination technique and improved fuzzy ARTMAP (IFAM). The method consists of three steps. To begin with, time-domain, frequency-domain and wavelet grey moments are extracted from the raw vibration signals to demonstrate the fault-related information. Then through the modified distance discrimination technique some salient features are selected from the original feature set. Finally, the optimal feature set is input into the IFAM incorporated with similarity based on the Yu’s norm in the classification phase to identify the different fault categories. The proposed method is applied to the fault diagnosis of rolling element bearing, and the test results show that the IFAM identify the fault categories of rolling element bearing more accurately and has a better diagnosis performance compared to the FAM. Furthermore, by the application of the bootstrap method to the diagnosis results it can testify that the IFAM has more capacity of reliability and robustness.     

The prediction of protein secondary structure with a Cascade Correlation Learning Architecture of neural networks

A Cascade Correlation Learning Architecture (CCLA) of neural networks is tested on the task of predicting the secondary structure of proteins. The results are compared with those obtained with Neural Networks (NN) trained with the back-propagation algorithm (BPNN) and generated with genetic algorithms. CCLA proceeds towards the global minimum of the error function more efficiently than BPNN. However, only a slight improvement in the average efficiency value is noticeable (61.82% as compared with 61.61% obtained with BPNN). The values of the three correlation coefficients for the discriminated secondary structures are also rather similar (C_t8,C ,C and C_coil are 0.36, 0.29 and 0.36 with CCLA, and 0.36, 0.31 and 0.35 with BPNN). This indicates that the efficiency of the prediction does not depend upon the training algorithm, and confirms our previous observation that when single sequences are used as input code to the network system, different NN architectures can perform similarly.     

基于改进的RBF神经网络在线辨识算法及其应用

针对径向基函数(RBF)神经网络用于非线性系统辨识时存在的问题，对径向基函数网络的拓扑结构作了改进，并给出了改进的径向基函数(MRBF)神经网络的中心选取方法和权值在线调整算法，最后用改进的径向基函数网络对一个典型工业对象(CSTR)进行了应用研究，结果表明方法有效。