Object-oriented image analysis for very-low-bitrate video-coding systems using the CNN universal machine

The CNN universal machine (CNNUM) is applied to object-oriented video compression and proves its universality for future applications in the field of very-low-bitrate coding. This proposal joins recent work of Venetianer and Roska in unfolding the enormous computational abilities of the CNNUM for a wide class of video compression techniques. Here a novel image analysis technique is considered and realized in the form of analogic CNN algorithms. The specific features of the scheme, among them the extensive use of dynamic (finite running time) CNN cloning templates, are outlined and discussed through different computer simulations. When implemented on the CNNUM, its performances outdo those of equivalent digital systems and qualify the CNNUM as a serious competitor for future video coding hardware. © 1997 John Wiley & Sons, Ltd.     

Pixel‐level snakes on the CNNUM: algorithm design,on‐chip implementation and applications

In this paper, a new algorithm for the cellular active contour technique called pixel‐level snakes is proposed. The motivation is twofold: on the one hand, a higher efficiency and flexibility in the contour evolution towards the boundaries of interest are pursued. On the other hand, a higher performance and suitability for its hardware implementation onto a cellular neural network (CNN) chip‐set architecture are also required. Based on the analysis of previous schemes the contour evolution is improved and a new approach to manage the topological transformations is incorporated. Furthermore, new capabilities in the contour guiding are introduced by the incorporation of inflating/deflating terms based on the balloon forces for the parametric active contours. The entire algorithm has been implemented on a CNN universal machine (CNNUM) chip set architecture for which the results of the time performance measurements are also given. To illustrate the validity and efficiency of the new scheme several examples are discussed including real applications from medical imaging. Copyright © 2005 John Wiley & Sons, Ltd.     

基于组合神经网络模型的电力变压器故障诊断方法

对故障空间的划分以及组合神经网络的构造方式 ,是利用组合神经网络进行变压器故障识别的关键。在讨论变压器故障空间划分方法及其存在问题的基础上 ,针对已积累的故障变压器的大量溶解气体数据 ,考察了各类故障的气体特征及聚类分析结果 ,并在此基础上构造了组合神经网络分层结构模型 ,实现了对变压器故障由粗到细的逐级划分 ,以提高诊断的准确性 ,为制定维修策略提供了依据。最后 ,结果显示了该模型的有效性     

Compact CMOS implementation of a low‐power,current‐mode programmable cellular neural network

We report on the design and characterization of a full‐analog programmable current‐mode cellular neural network (CNN) in CMOS technology. In the proposed CNN, a novel cell‐core topology, which allows for an easy programming of both feedback and control templates over a wide range of values, including all those required for many signal processing tasks, is employed. The CMOS implementation of this network features both low‐power consumption and small‐area occupation, making it suitable for the realization of large cell‐grid sizes. Device level and Monte Carlo simulations of the network proved that the proposed CNN can be successfully adopted for several applications in both grey‐scale and binary image processing tasks. Results from the characterization of a preliminary CNN test‐chip (8×1 array), intended as a simple demonstrator of the proposed circuit technique, are also reported and discussed. Copyright © 2001 John Wiley & Sons, Ltd.     

Robustness of cellular neural networks in image deblurring and texture segmentation

Template parameters of cellular neural networks (CNNs) should be robust enough to random variability of VLSI tolerances and noise. Using the CNN for image processing, one of the main problems is the robustness of a given task in a real VLSI chip. It will be shown that very different tasks such as 2D or 3D deconvolution and texture segmentation can be solved in a real VLSI CNN environment without significant loss of efficiency and accuracy under low precision (about 6–8 bits) and random variability of the VLSI parameters. The CNN turns out to be very robust against template noise, image noise, imperfect estimation of templates and parameter accuracy. The parameters of a template are tuned using genetic learning. These optimized parameters depend on the precision of the architecture. It was found that about 6–8 bits of precision is enough for a complicated multilayer deconvolution, while only 4 bits of precision is enough for difficult texture segmentation in the presence of noise and parameter variances. The tolerance sensitivity of template parameters is considered for VLSI implementation. Theory and examples are demonstrated by many results using real-life microscopic images and natural textures.     

Cellular neural networks for real-time DNA microarray analysis

A new methodology for real-time processing of DNA chip images is proposed. The idea developed here is to use the cellular neural network (CNN) array to analyze the DNA microarray. A CNN is an analog dynamic processor array that reflects this property: the processing elements interact directly within a finite local neighborhood. Due to its architecture, a two-dimensional CNN array is widely used to solve image processing and pattern recognition problems; moreover, the parallelism characteristic of this structure allows one to perform the most computationally expensive image analysis tasks three orders of magnitude faster than a classical CPU-based computer. This approach, thanks to the supercomputing capabilities of the CNN architecture, makes the whole DNA chip methodology fully parallel and also makes the processing phase, until now very time consuming, a real-time step. We discuss the results of testing an algorithm based on the CNN universal machine (CNN-UM) that has been designed to classify the image data. The algorithm is implemented in an analogic (analog and logic) microprocessor.     

Analogic CNN algorithms for 3D interpolation-approximation and object rotation using controlled switched templates

Analogic CNN algorithms are presented for various interpolation and approximation tasks in 3D. They are designed on the basis of mechanical analogies. Symmetric space-variant operations are implemented by the CNN algorithms; with switched templates, a key example is object rotation. Direction and speed coding are shown in detail.     

DESIGN AND LEARNING WITH CELLULAR NEURAL NETWORKS

The template coefficients (weights) of a CNN which will give a desired performance, can either be found by design or by learning. ’By design‘ means that the desired function to be performed can be translated into a set of local dynamic rules, while ’by learning‘ is based exclusively on pairs of input and corresponding output signals, the relationship of which may be far too complicated for the explicit formulation of local rules. An overview of design and learning methods applicable to CNNs, which sometimes are not clearly distinguishable, will be given from an engineering point of view.     

Spatial logic algorithms using basic morphological analogic CNN operations

The CNN implementation of basic scale-, rotation- and translation-invariant morphological functions and novel types of analogic CNN algorithms using spatial logic are introduced for object recognition (detection) purposes. The power of the technique is illustrated on bank-note recognition tasks.     

基于阶段聚焦损失和并行增广策略的 遥感图像场景分类

随着深度学习的不断普及,卷积神经网络已成为遥感图像场景分类的主要手段,然而当前的研究主要集中于多网络主干的信息融合以及注意力机制等领域,在提高分类精度的同时也带来更高的计算复杂度。针对上述问题,分别从改进卷积神经网络损失函数和设计新的样本训练策略两个角度出发,在不增加计算复杂度的前提下,提升卷积神经网络的分类性能。首先,在对传统交叉熵和Focal loss损失函数进行分析的基础上,提出一种阶段聚焦损失函数,该损失函数可以在训练阶段对卷积网络进行有侧重的性能挖掘。其次,设计了一种并行样本训练策略,将采用Gridmask算法增广后的样本图像和原始样本图像,分为两路输入卷积网络进行并行训练,进一步提升卷积网络的分类性能。实验结果表明,所提出的算法分别在AID和NWPU-RESISC45两个大规模数据库上取得了96.72%和93.95%的检测精度,可以显著提升遥感图像场景分类的性能。     

A new approach for edge detection of noisy image based on CNN

A new approach for edge detection of noisy image by cellular neural network (CNN) is proposed in this paper. In order to get the reasonable template, the statistical characteristics of image are utilized, and Gibbs image model is employed to describe the stochastic dependence of an edge pixel on its neighbourhood. Based on stochastic edge image models, edge detection of noisy image is equivalent to seeking a minimum of a cost function. If the template of CNN is designed carefully, the energy function can be mapped properly to the cost function of stochastic edge image model, then CNN can be used for seeking the minimum of cost function. Genetic algorithm is efficient in the field of optimization, and we also utilized this algorithm to get the correct form of template. The results of computer simulation confirm that the new approach is very effective. Furthermore, this result also confirms that we can design template for many different questions based on statistical image model, and the area of application of CNN will be widened. Copyright © 2003 John Wiley & Sons, Ltd.     

The analogic cellular neural network as a bionic eye

The conception of the CNN universal machine has led quite naturally to the invention of the analogic CNN bionic eye (henceforth referred to simply as the bionic eye). the basic idea is to combine the elementary functions, the building blocks, of the retina and other 2 1/2 D sensory organs, algorithmically, in a stored programme of a CNN universal machine, through the use of artificial analogic programmes. the term bionic is defined in a rigorous way: it is a nonlinear, dynamic, spatiotemporal biological model implemented in a stored programme electronic (optoelectronic) device; this device is in our case the analogic CNN universal machine (or chip). The aim of this paper is to report on this new invention, particularly to electronic and computer engineers, in a tutorial way. We begin by summarizing (1) the biological aspects of the range of retinal function (the retinal universe), (2) the CNN paradigm and the CNN universal machine architecture and (3) the general principles of retinal modelling in CNN. Next we describe new CNN circuit and template design innovations that can be used to implement physiological functions in the retina and other sensory organs using the CNN universal machine. Finally we show how to combine given retinal functional elements implemented in the CNN universal machine with analogic algorithms to form the bionic retina. the resulting system can be used not only for simulating biological retinal function but also for generating functions that go far beyond biological capabilities. Several bionic retina functions, different topographic modalities and analogic CNN algorithms can then be combined to form the analogic CNN bionic eye. the qualitative aspects of the models, especially the range of dynamics and accuracy considerations in VLSI optoelectronic implementations, are outlined. Finally, application areas of the bionic eye and possibilities of constructing innovative devices based on this invention (such as the bionic eyeglass or the visual mouse) are described.     

基于融合LDA与双层CNN的文本分类研究

针对基于主题的文本分类任务存在的主题特征表征能力不足、数据高维导致的特征维度高等问题，本文对输入的特征表示与卷积神经网络结构（CNN）做出了改进。在特征表示时提出了使用LDA模型计算逆主题空间频率从而得到文本的主题向量矩阵，降低了噪声主题的特征表达，增强了关键主题的权重；分别将文本的主题向量矩阵与词向量矩阵作为CNN模型的输入。提出了双层CNN网络结构，在每层CNN的池化层后增加一层多通道池化层，以融合每层CNN的池化结果，降低特征维度的同时获取更多的局部显著特征；最后使用Attention机制对融合的特征进行加权后输入到全连接层进行分类。由实验结果可知，改进的模型在文本分类任务上的准确率、召回率均在98%以上,F1值较基准实验提高了近6%。     

基于CNN(1D)-LSTM模型的电站锅炉SCR入口NOx浓度预测

为了解决电站锅炉操作人员依赖经验调节锅炉运行参数降低SCR入口NOx浓度，提高脱硝效果的问题，提出一种SCR入口处NOx浓度预测方法。该方法建立了基于卷积神经网络和长短期记忆神经网络的CNN(1D)-LSTM模型，通过提取锅炉在时序上的特征参数，可预测5 min后SCR入口处NOx浓度。电厂运行人员可将该模型的预测结果作为SCR入口处NOx浓度的重要参考，更加有效地调节锅炉参数进行脱硝优化。结果表明，预测3 min后SCR入口处NOx浓度LSTM模型优于CNN(1D) LSTM；预测5 min后的SCR入口浓度CNN(1D)-LSTM模型相比于LSTM模型预测精度有很大的所提高，在测试集上Emape为7.05%，取得了期望的效果。     

基于改进卷积神经网络的风电轴承故障诊断策略

针对风电机组滚动轴承故障特征微弱、提取困难、诊断效率低下等问题,提出一种基于改进卷积神经网络(Convolution Neural Network,CNN)的故障诊断算法。改进CNN模型结构,在全连接层前增加新的卷积层,挖掘信号的深层特征以提高模型的泛化能力。对卷积层数据进行批归一化处理,采用带有动量的随机梯度下降训练算法来加速训练速度。详细介绍了改进CNN的工作原理,给出了采用改进CNN进行故障诊断的流程。最后利用凯斯西储大学滚动轴承数据库的数据进行验证。证明该方法不需要预先提取信号的故障特征,可直接实现对轴承的故障特征提取以及故障识别,诊断率高。     

Stability analysis of generalized cellular neural networks

A rather general class of neural networks, called generalized cellular neural networks (CNNs), is introduced. the new model covers most of the known neural network architectures, including cellular neural networks, Hopfield networks and multilayer perceptrons. Several sets of conditions ensuring the input-output stability and global asymptotic stability of generalized CNNs have been obtained. the conditions for the stability of individual cells are checked in the frequency domain, while the stability of the overall network is analysed in terms of the stability of individual cells and the connectivity characteristics. the results on the global asymptotic stability are useful for the design of a generalized CNN such that the orbit of each state converges to a globally asymptotically stable equilibrium point which depends only on the input and not on the initial state. Such a network defines an algebraic map from the space of external inputs to the space of steady state values of the outputs and hence can accomplish cognitive and computational tasks.     

CNN-based difference-controlled adaptive non-linear image filters

In this paper, we develop a common cellular neural network framework for various adaptive non-linear filters based on robust statistic and geometry-driven diffusion paradigms. The base models of both approaches are defined as difference-controlled non-linear CNN templates, while the self-adjusting property is ensured by simple analogic (analog and logic) CNN algorithms. Two adaptive strategies are shown for the order statistic class. When applied to the images distorted by impulse noise both give more visually pleasing results with lower-frequency weighted mean square error than the median base model. Generalizing a variational approach we derive the constrained anisotropic diffusion, where the output of the geometry-driven diffusion model is forced to stay close to a pre-defined morphological constraint. We propose a coarse-grid CNN approach that is capable of calculating an acceptable noise-level estimate (proportional to the variance of the Gaussian noise) and controlling the fine-grid anisotropic diffusion models. A combined geometrical–statistical approach has also been developed for filtering both the impulse and additive Gaussian noise while preserving the image structure. We briefly discuss how these methods can be embedded into a more complex algorithm performing edge detection and image segmentation. The design strategies are analysed primarily from VLSI implementation point of view; therefore all non-linear cell interactions of the CNN architecture are reduced to two fundamental non-linearities, to a sigmoid type and a radial basis function. The proposed non-linear characteristics can be approximated with simple piecewise-linear functions of the voltage difference of neighbouring cells. The simplification makes it possible to convert all space-invariant non-linear templates of this study to a standard instruction set of the CNN Universal Machine, where each instruction is coded by at most a dozen analog numbers. Examples and simulation results are given throughout the text using various intensity images. © 1998 John Wiley & Sons, Ltd.     

Gear Transmission Fault Classification using Deep Neural Networks and Classifier Level Sensor Fusion

Gear transmissions are widely used in industrial drive systems. Fault diagnosis of gear transmissions is important for maintaining the system performance, reducing the maintenance cost, and providing a safe working environment. This paper presents a novel fault diagnosis approach for gear transmissions based on convolutional neural networks (CNNs) and decision-level sensor fusion. In the proposed approach, a CNN is first utilized to classify the faults of a gear transmission based on the acquired signals from each of the sensors. Raw sensory data is sent directly into the CNN models without manual feature extraction. Then, classifier level sensor fusion is carried out to achieve improved classification accuracy by fusing the classification results from the CNN models. Experimental study is conducted, which shows the superior performance of the developed method in the classification of different gear transmission conditions in an automated industrial machine. The presented approach also achieves end-to-end learning that can be applied to the fault classification of a gear transmission under various operating conditions and with signals from different types of sensors.     

A denoising-classification neural network for power transformer protection

Artificial intelligence (AI) can potentially improve the reliability of transformer protection by fusing multiple features. However, owing to the data scarcity of inrush current and internal fault, the existing methods face the problem of poor generalizability. In this paper, a denoising-classification neural network (DCNN) is proposed, one which integrates a convolutional auto-encoder (CAE) and a convolutional neural network (CNN), and is used to develop a reliable transformer protection scheme by identifying the exciting voltage-differential current curve (VICur). In the DCNN, CAE shares its encoder part with the CNN, where the CNN combines the encoder and a classifier. Based on the interaction of the CAE reconstruction process and the CNN classification process, the CAE regards the saturated features of the VICur as noise and removes them accurately. Consequently, it guides CNN to focus on the unsaturated features of the VICur. The unsaturated part of the VICur approximates an ellipse, and this significantly differentiates between a healthy and faulty transformer. Therefore, the unsaturated features extracted by the CNN help to decrease the data ergodicity requirement of AI and improve the generalizability. Finally, a CNN which is trained well by the DCNN is used to develop a protection scheme. PSCAD simulations and dynamic model experiments verify its superior performance.     

基于卷积神经网络的暂态电压稳定评估及风险量化

提出了一种暂态电压稳定性评估及其风险量化方法.首先,探讨卷积神经网络(CNN)与暂态电压稳定评估的关联性和匹配性,建立了基于CNN的暂态电压稳定评估模型.其次,在可信度框架下引入四元评估结构,可有效解决CNN在稳定边界识别上对时域仿真依赖的难题.然后,通过评估结果获取暂态电压稳定裕度,并将其与可信度相结合来构建风险函数,从而实现对暂态电压稳定的风险量化分级.实际电网算例分析结果验证了该方法的有效性.