An improved pulse coupled neural network for image processing

To develop new image processing applications for pulse coupled neural network (PCNN), this paper proposes an improved PCNN model by redesigning the linking input, activity strength, linking weight, pulse threshold and pixel update rule. Two typical image processing examples based on such a model, namely fingerprint orientation field estimation and noise removal, are presented for explaining how to use the PCNN and determine parameters in image processing. Experiments show that the improved model is quite useful, and the PCNN-based approaches achieve better image processing results than the traditional ones. This work was supported by National Science Foundation of China under Grant 60471055 and Specialized Research Fund for the Doctoral Program of Higher Education under Grant 20040614017.     

基函数神经网络权值直接确定的图像复原

给出了基函数神经网络图像复原的模型,该神经网络模型是由三层构成的前向神经网络,以一组正交基为隐层神经元的激励函数。为了避免反复迭代权值修正的冗长BP训练过程,提出了一种权值直接确定的算法。实验结果表明,该种权值直接确定算法不仅能一步确定权值而获得更快的运算速度,而且能达到更高的精度。     

A novel activation function for multilayer feed-forward neural networks

Traditional activation functions such as hyperbolic tangent and logistic sigmoid have seen frequent use historically in artificial neural networks. However, nowadays, in practice, they have fallen out of favor, undoubtedly due to the gap in performance observed in recognition and classification tasks when compared to their well-known counterparts such as rectified linear or maxout. In this paper, we introduce a simple, new type of activation function for multilayer feed-forward architectures. Unlike other approaches where new activation functions have been designed by discarding many of the mainstays of traditional activation function design, our proposed function relies on them and therefore shares most of the properties found in traditional activation functions. Nevertheless, our activation function differs from traditional activation functions on two major points: its asymptote and global extremum. Defining a function which enjoys the property of having a global maximum and minimum, turned out to be critical during our design-process since we believe it is one of the main reasons behind the gap observed in performance between traditional activation functions and their recently introduced counterparts. We evaluate the effectiveness of the proposed activation function on four commonly used datasets, namely, MNIST, CIFAR-10, CIFAR-100, and the Pang and Lee’s movie review. Experimental results demonstrate that the proposed function can effectively be applied across various datasets where our accuracy, given the same network topology, is competitive with the state-of-the-art. In particular, the proposed activation function outperforms the state-of-the-art methods on the MNIST dataset.     

A novel deep learning neural network for fast-food image classification and prediction using modified loss function

Multimedia Tools and Applications - Accurate food image classification is often critical to accurately monitor the dietary assessment to reduce the risk of different heart-related diseases,...     

A novel radial basis function neural network for discriminant analysis

A novel radial basis function neural network for discriminant analysis is presented in this paper. In contrast to many other researches, this work focuses on the exploitation of the weight structure of radial basis function neural networks using the Bayesian method. It is expected that the performance of a radial basis function neural network with a well-explored weight structure can be improved. As the weight structure of a radial basis function neural network is commonly unknown, the Bayesian method is, therefore, used in this paper to study this a priori structure. Two weight structures are investigated in this study, i.e., a single-Gaussian structure and a two-Gaussian structure. An expectation-maximization learning algorithm is used to estimate the weights. The simulation results showed that the proposed radial basis function neural network with a weight structure of two Gaussians outperformed the other algorithms.     

Raster cellular neural network simulator for image processing applications with numerical integration algorithms

In this paper, a universal simulator for cellular neural network (CNN) is presented. This simulator is capable of performing Raster simulation for any size of input image, and thus is a powerful tool for researchers investigating potential applications of CNN. This paper reports the latency properties of CNNs along with popular numerical integration algorithms; results and comparisons are also presented.     

The facial expression recognition technology under image processing and neural network

The Journal of Supercomputing - A facial expression recognition (FER) algorithm is built on the advanced convolutional neural network (CNN) to improve the current FER algorithms’ recognition...     

Analysis of pupil fluctuations after a light stimulus by image processing and neural network

In this paper the analysis of pupil fluctuations after a light stimulus is considered; it is useful for non-invasive diagnosis of many different diseases. When a light stimulus is presented to a subject, the pupil response is not instantaneous because of the action of the sphincter muscle. A sequence of images will be caught by a pupillometer and each image of the sequence will be binarized; for each segmented image, a useful parameter will be considered, the major diameter that is the length (in pixels) of the major axis of the ellipse that has the same second moments of the pupil. The aim is the identification of the response time after a light stimulus, from the sequence of the major diameters. The considered signal is degraded because of the presence of the measurement noise, the natural fluctuation of the pupil (usually called “pupil noise”), and the general health state of the subject. To enhance the significant part of this noisy signal a neural network is suitable trained. From the clean signal the identification of the response time of the pupil will be easier and a simple method will be proposed.     

Artificial neural network models for lot-sizing problem: a case study

Lot-sizing is one of the most difficult problems in production planning. The main purpose of this study is to propose a new lot-sizing based on artificial neural network (ANN), which may lead to a better performance than commonly used lot-sizing heuristics (SM, EOQ, PPB, LUC, and LTC). The data obtained are the results of years 2004 thru 2009 for 186 different types of stock items from the 2nd Air Supply and Maintenance Centre Command, a state-funded factory in Kayseri, Turkey. Factual data were applied under the coverage of the study, and the system from which the data have been obtained is still in live and active status. In the study, the purchasing costs, holding costs, and set-up costs were taken into consideration. These data were obtained from the administration data system of the enterprise. The solutions of this lot-sizing heuristics were found by WinQSB software accordingly. The ANN was constituted by using the NeuroSolutions software. The criterion of deviation from the optimum solution and the criterion of percentage of times obtaining the optimum order pattern were taken into account for the comparison purposes. The performance values of 400 ANNs were compared to lot-sizing methods. MS Excel and Visual Basic Macro were utilized for all calculations applied after this stage. The results showed that the proposed ANN-based method outperformed all lot-sizing methods taken into account in this study.     

A one-layer recurrent neural network with a discontinuous hard-limiting activation function for quadratic programming.

In this paper, a one-layer recurrent neural network with a discontinuous hard-limiting activation function is proposed for quadratic programming. This neural network is capable of solving a large class of quadratic programming problems. The state variables of the neural network are proven to be globally stable and the output variables are proven to be convergent to optimal solutions as long as the objective function is strictly convex on a set defined by the equality constraints. In addition, a sequential quadratic programming approach based on the proposed recurrent neural network is developed for general nonlinear programming. Simulation results on numerical examples and support vector machine (SVM) learning show the effectiveness and performance of the neural network.     

An efficient radial basis function neural network for hyperspectral remote sensing image classification

A very simple radial basis function neural network (RBFNN) is investigated for hyperspectral remote sensing image classification. Its training can be analytically solved with a closed-form equation, and no parameter needs to be manually tuned. Its computational cost is much lower than the popular support vector machine (SVM). Surprisingly, such an RBFNN can achieve the performance that is similar to or even better than the SVM. By incorporating a simple spatial averaging filter or a Gaussian lowpass filter with negligible additional computational cost, classification accuracy can be further improved. Considering the large matrix inversion operation in the RBFNN when the number of training samples being very large, we also propose a parallel processing method to reduce computing time in matrix inversion.     

Quantum convolutional neural network for image classification

Pattern Analysis and Applications - In this paper we propose two scale-inspired local feature extraction methods based on Quantum Convolutional Neural Network (QCNN) in the Tensorflow quantum...     

A neural network accelerator for image analysis

At the heart of this image analysis system are two NET32K analog neural network chips, computing over 100 billion multiply-accumulates per second. The system simultaneously scans sixty-four 16×16-pixel templates over bi-level images, producing feature maps that mark matches between the image and the templates. When we code simple, generic shapes into the templates, the feature maps allow us to make quick, robust analyses of complex, noisy images     

Kernel-blending connection approximated by a neural network for image classification

Computational Visual Media - This paper proposes a kernel-blending connection approximated by a neural network (KBNN) for image classification. A kernel mapping connection structure, guaranteed by...     

用于信息处理的延时混沌神经网络

为了提高混沌神经网络用于信息处理的能力,采用一种参数调节控制方法,通过对一种延时对称全局耦合混沌神经网络的黏合参数的控制研究了网络的动态联想记忆,使被控网络在仅有部分神经元进入周期态的情况下达到输出稳定,并且稳定输出序列只包含与输入模式相关的存储模式及其相反模式。仿真实验说明网络具有良好的容错能力和很高的回忆正确率,适合应用于信息处理和模式识别。     

Predictions of bridge scour: Application of a feed-forward neural network with an adaptive activation function

In this study, a new procedure to determine the optimum activation function for a neural network is proposed. Unlike previous methods of optimising activation functions, the proposed approach regards selection of the most suitable activation function as a discrete optimisation problem, which involves generating various combinations of function then evaluating their performance as activation functions in a neural network, returning the function or combination of functions which yields best result as the optimum. The efficacy of the proposed optimisation method is compared with conventional approaches using the data generated from several synthetic functions. Numerical results indicate that the network produced using the proposed method achieves a better accuracy with a smaller network size, compared to other approaches.Bridge scour problem is used to further demonstrate the performance of the proposed algorithm. Based on the training and validation results, a better estimation of both equilibrium and time dependent scour depth is produced by the neural network developed using the proposed optimisation method, compared to networks with a priori chosen activation functions. Furthermore, the performance of the proposed model is compared with predictions of empirical methods, with the former making more accurate predictions.     

基于径向基函数的多帧图像超分辨重建算法

神经网络具有强大的非线性学习能力,基于神经网络的多帧超分辨重建方法获得了初步研究,但这些方法一般只能应用于帧间具有标准位移的控制成像情形,难以推广应用到其他实际情况。为了将神经网络强大的学习能力应用到非控制成像多帧超分辨重建中,以获得更好的超分辨效果,提出了一种利用径向基函数(RBF)神经网络进行解模糊的算法,并将其与多帧非均匀插值结合起来,形成了一种新的两步超分辨算法。仿真实验结果表明,该算法的结构相似度为0.55～0.7。该算法不但扩展了RBF神经网络的应用范围,还获得了更好的超分辨性能。     

Confidence interval prediction for neural network models

To derive an estimate of a neural network's accuracy as an empirical modeling tool, a method to quantify the confidence intervals of a neural network model of a physical system is desired. In general, a model of a physical system has error associated with its predictions due to the dependence of the physical system's output on uncontrollable or unobservable quantities. A confidence interval can be computed for a neural network model with the assumption of normally distributed error for the neural network. The proposed method accounts for the accuracy of the data with which the neural network model is trained.     

一类非线性系统的多模型神经网络解耦控制器

针对多变量非线性离散时间系统设计多模型神经网络解耦控制器，在每个平衡点处用一神经网络离线辨识非线性系统的线性部分，利用另一神经网络在线辨识非线性部分，将非线性部分视为可测干扰并采用前馈的方法予以消除，所有平衡点处得到的系统模型汇集起来构成多模型集，在每一采样时刻基于切换指标选出最优模型作为当前模型，并据此设计解耦控制器实现控制，仿真结果表明系统在多个平衡点处仍然可以得到较好的控制效果。     

Observer-participant models of neural processing

A model is proposed in which the neuron serves as an information channel. Channel distortion occurs through the channel since the mapping from input Boolean codes to output codes are many-to-one in that neuron outputs consist of just two distinguished states. Within the described model, the neuron performs a decision-making function. Decisions are made regarding the validity of a question passively posed by the neuron. This question becomes defined through learning hence learning is viewed as the process of determining an appropriate question based on supplied input ensembles. An application of the Shannon information measures of entropy and mutual information taken together in the context of the proposed model lead to the Hopfield neuron model with conditionalized Hebbian learning rules. Neural decisions are shown to be based on a sigmoidal transfer characteristic or, in the limit as computational temperature tends to zero, a maximum likelihood decision rule. The described work is contrasted with the information-theoretic approach of Linsker.