Synthetic organisms and self-designing systems

This paper examines the need for complex, adaptive solutions to certain types of complex problems typified by the Strategic Defense System and NASA's Space Station and Mars Rover. Since natural systems have evolved with capabilities of intelligent behavior in complex, dynamic situations, it is proposed that biological principles be identified and abstracted for application to certain problems now facing industry, defense, and space exploration. Two classes of artificial neural networks are presented — a nonadaptive network used as a genetically determined “retina,” and a frequency-coded network used as an adaptive “brain.” The role of a specific environment coupled with a system of artificial neural networks having simulated sensors and effectors is seen as an ecosystem. Evolution of synthetic organisms within this ecosystem provides a powerful optimization methodology for creating intelligent systems able to function successfully in any desired environment. A complex software system involving a simulation of an environment and a program designed to cope with that environment are presented. Reliance on adaptive systems, as found in nature, is only part of the proposed answer, though an essential one. The second part of the proposed method makes use of an additional biological metaphor—that of natural selection—to solve the dynamic optimization problems every intelligent system eventually faces. A third area of concern in developing an adaptive, intelligent system is that of real-time computing. It is recognized that many of the problems now being explored in this area have their parallels in biological organisms, and many of the performance issues facing artificial neural networks may find resolution in the methodology of real-time computing.     

Morphology neural networks: An introduction with applications

The area of artificial neural networks has recently seen an explosion of theoretical and practical results. In this paper, we present an artificial neural network that is algebraically distinct from the classical artificial neural networks, and several applications which are different from the typical ones. In fact, this new class of networks, calledmorphology neural networks, is a special case of a general theory of artificial neural nets, which includes the classical neural nets. The main difference between a classical neural net and a morphology neural net lies in the way each node algebraically combines the numerical information. Each node in a classical neural net combines information by multiplying output values and corresponding weights and summing, while in a morphology neural net, the combining operation consists of adding values and corresponding weights, and taking the maximum value. We lay a theoretical foundation for morphology neural nets, describe their roots, and give several applications in image processing. In addition, theoretical results on the convergence issues for two networks are presented.This research was supported in part by National Science Foundation, Contract No. ECS-9010403.     

用于入侵的自适应遗传算法训练人工神经网络

给出了一种能和网络结构一一对应的、合适的染色体编码方法.用物种入侵的遗传算法训练人工神经网络,在入侵过程中,遗传算法自适应地调整交叉算子和变异算子.提出了一种根据平均适应度值确定入侵物种规模的方法,并详细描述了算法步骤,最后通过实验证明了本文算法的有效性和优越性.     

基于神经网络的语音识别与自动检验研究

语音识别系统在语音识别中自我判定识别结果，并从错误中自动获取经验改正错误实现知识的自我完善具有重要意义。采用人工神经网络可以有效学习与更新知识，人工神经网络与语音识别结果自动检验方法结合实现一种新的有效学习与更新系统。在该系统中采用基于LEA判别法的梯度牛顿有效结合神经网络快速学习方法。该系统实现在语音识别实践中能够自学习并提高识别率，具有一定的智能。文中给出系统原理图和实验结果。     

A Unified Artificial Neural Network Architecture for Active Power Filters

In this paper, an efficient and reliable neural active power filter (APF) to estimate and compensate for harmonic distortions from an AC line is proposed. The proposed filter is completely based on Adaline neural networks which are organized in different independent blocks. We introduce a neural method based on Adalines for the online extraction of the voltage components to recover a balanced and equilibrated voltage system, and three different methods for harmonic filtering. These three methods efficiently separate the fundamental harmonic from the distortion harmonics of the measured currents. According to either the Instantaneous Power Theory or to the Fourier series analysis of the currents, each of these methods are based on a specific decomposition. The original decomposition of the currents or of the powers then allows defining the architecture and the inputs of Adaline neural networks. Different learning schemes are then used to control the inverter to inject elaborated reference currents in the power system. Results obtained by simulation and their real-time validation in experiments are presented to compare the compensation methods. By their learning capabilities, artificial neural networks are able to take into account time-varying parameters, and thus appreciably improve the performance of traditional compensating methods. The effectiveness of the algorithms is demonstrated in their application to harmonics compensation in power systems     

平坦瑞利环境下多用户检测的量子神经网络方法

提出采用量子神经网络(QNN)方法在平坦瑞利环境下进行多用户检测的方法。量子神经网络是量子计算与人工神经网络(ANN)相结合的产物,由于利用量子并行计算和量子纠缠等特性从而克服了传统人工神经网络的固有缺点。研究结果表明:该算法具有较强的鲁棒性;能有效地抑制噪声干扰,克服远近效应,在平坦瑞利衰减下具有较好地误码性能。     

Separating core and noncore knowledge: an application of neural network rule extraction to a cross-national study of brand image perception

Recent advances in algorithms that extract rules from artificial neural networks make it feasible to use neural networks as a tool for acquiring knowledge hidden in the data. Findings are reported from the use of such algorithms to separate core and noncore knowledge in a cross-national study of automobile brand image perception. Respondents from five Western European countries have been asked to associate individual and corporate brand associations for a number of well-known automobile brands. Knowledge, expressed as concise and accurate rules that distinguish between the respondents' perceptions of German and Japanese brands, is extracted from trained neural networks. This paper explains how both core knowledge, which captures the perceptions shared by the respondents in all countries, and country-specific noncore knowledge can be acquired and differentiated by a proposed two-step approach to train and extract rules from a multi-neural network system. The experimental results show that, in addition to providing a better understanding of the differences and similarities in the brand image perceptions of consumers in various countries, the proposed approach also yields better predictive accuracy than a decision tree method.     

Modelling of LDMOS transistor using artificial neural networks: DC and RF performances

This article presents an accurate method based on artificial neural networks (ANNs) for DC and RF modelling of laterally diffused metal oxide semiconductor (LDMOS) transistors, under various temperature conditions. In LDMOS transistors, temperature is an effective factor, so the proposed models include this parameter. Two neural networks‐based procedures have been proposed for LDMOS transistor modelling, first for DC and second for RF modelling. In each case, two kinds of neural networks have been used, multilayer perceptron and radial basis function neural networks. Two models are compared to each other in terms of accuracy, and for both of them, an excellent agreement between modelled and measured data is obtained. The ANN model is developed and trained with the help of data obtained by simulation of a Si‐LDMOS transistor using ADS software.     

Phased arrays in communication system based on Taguchi‐neural networks

Phased antenna array design is one of the most important electromagnetic optimization problems. This research combined the Taguchi method and artificial intelligence methods, used them as the prediction tool in designing parameters for the communication system, and then constructed a set of the optimal parameter analysis flow and steps. In this paper, we present an application of artificial neural networks in the electromagnetic domain. We particularly look at the multilayer perceptron network, which has been the most used of artificial neural networks architectures both in the electromagnetic domain and in the Taguchi optimization technique and describes the Taguchi method to optimize the excitations elements of the linear array to produce a radiation pattern with minimum side lobe level and null placement control. This paper investigates how the implementation of the signal processing in hardware affects the performance of the adaptive array antenna. The investigation is confined to uplink or receive antenna array only. Results of a prototype of antenna array with feeding values designed using the proposed techniques are also presented. Copyright © 2013 John Wiley & Sons, Ltd.     

New neural method for bias dependent noise modelling of microwave transistors

A new method for accurate determination of noise parameters of microwave transistors for various bias conditions is proposed in this paper. The proposed model consists of a transistor empirical noise model (modification of Pospieszalski’s noise model) and two artificial neural networks. With the aim to avoid extraction of the empirical model parameters for each bias point, an artificial neural network is used to introduce bias-dependence of the equivalent circuit parameters. Accuracy of such bias-dependent model is further improved by using an additional neural network aimed to correct the noise parameters’ values. The proposed modeling approach is exemplified by modelling of a MESFET device in packaged form. The noise parameters obtained by the simulation agree well with the measured data.     

A Proportional-Integrator Type Adaptive Critic Design-Based Neurocontroller for a Static Compensator in a Multimachine Power System

A novel nonlinear optimal controller for a static compensator (STATCOM) connected to a power system, using artificial neural networks, is presented in this paper. The action dependent heuristic dynamic programming, a member of the adaptive critic designs family is used for the design of the STATCOM neurocontroller. This neurocontroller provides optimal control based on reinforcement learning and approximate dynamic programming. Using a proportional-integrator approach, the proposed neurocontroller is capable of dealing with actual rather than deviation signals. Simulation results are provided to show that the proposed controller outperforms a conventional PI controller for a STATCOM in a small and large multimachine power system during large-scale faults, as well as small disturbances     

Neural network based adaptable control method for linearization of high power amplifiers

The proposed adaptable control method for linearization of high power amplifiers is powered by the neural networks technique that supports analogue polynomial type of predistorters, which are widely utilized in commercial power amplifiers for wireless communication purposes. This paper presents an algorithm to determine the coefficients by use of artificial neural networks and its generalization feature that helps to map the power amplifier response with optimal coefficients of the polynomial, which in a proper way pre-distorts an input signal of the amplifier. The concept of the predistortion has been introduced. Furthermore, the overall step-by-step initialization and functionality of the control process has also been described. The method has been tested successfully in a real power amplifier equipped with an analogue predistorting circuits. Presented measurement results imply that this approach is robust and well suited for such category of the power amplifier design, in which the artificial neural networks play substantial role.     

Application of wavelet analysis and artificial neural networks in solving the problem concerning shape recognition of noisy pulsed signals

The problem concerning recognition of single pulses under the action of interferences is discussed by the example of classification of neuron action potentials. Joint applications of wavelets and artificial neural networks in solving the the given problem are analyzed. The recognition method, which is based on wavelet neural networks and ensures adjustment of the synapses of a supplementary (??wavelet??) layer, has been proposed. It is demonstrated that experimental data can efficiently be analyzed via the proposed method.     

Synthesis of artificial neural networks by transconductors only

Hardware implementation of artificial neural networks has been attracting great attention recently. In this work, the analog VLSI implementation of artificial neural networks by using only transconductors is presented. The signal flow graph approach is used in synthesis. The neural flow graph is defined. Synthesis of various neural network configurations by means of neural flow graph is described. The approach presented in this work is technology independent. This approach can be applied to new neural network topologies to be proposed or used with transconductors designed in future technologies.     

人工神经网络的多维空间几何分析及其理论

本文系统地讨论了作为用于分析人工神经网络的一种方法即多维空间几何方法,并对多维空间几何学进行系统的研究,推导了必要的定理,为人工神经网络的分析提供了新的手段.     

Towards a high quality Arabic speech synthesis system based on neural networks and residual excited vocal tract model

Text-to-speech conversion has traditionally been performed either by concatenating short samples of speech or by using rule-based systems to convert a phonetic representation of speech into an acoustic representation, which is then converted into speech. This paper describes a text-to-speech synthesis system for modern standard Arabic based on artificial neural networks and residual excited LPC coder. The networks offer a storage-efficient means of synthesis without the need for explicit rule enumeration. These neural networks require large prosodically labeled continuous speech databases in their training stage. As such databases are not available for the Arabic language, we have developed one for this purpose. Thus, we discuss various stages undertaken for this development process. In addition to interpolation capabilities of neural networks, a linear interpolation of the coder parameters is performed to create smooth transitions at segment boundaries. A residual-excited all pole vocal tract model and a prosodic-information synthesizer based on neural networks are also described in this paper.     

Artificial Neural Network Based Vertical Handoff Algorithm for Reducing Handoff Latency

One of the most challenging topics for next generation wireless networks is vertical handoff concept since several wireless technologies are assumed to cooperate. Plenty of parameters related to user preferences, application requirements, and network conditions, such as; data rate, service cost, network latency, speed of mobile, battery level, interference ratio and etc. must be considered in vertical handoff process along with traditional RSSI information. In this study, a new artificial neural network based handoff decision algorithm is proposed in order to reduce the handoff latency of smart terminal deployed in aforementioned wireless heterogeneous infrastructures. The prominent parameters data rate, monetary cost and RSSI information are taken as inputs of the developed vertical handoff decision system. Performance results of the proposed system are also compared with those of classical Multiple Attribute Decision Making method Simple Additive Weighting, and of some other artificial intelligence based algorithms. According to the results obtained, the proposed neural network based vertical handoff decision algorithm is able to determine whether a handoff is necessary or not properly, and selects the best candidate access network considering the abovementioned parameters. The results also show that, the neural network based algorithm developed significantly reduces the handoff latency while the number of handoffs, which is another vital performance metric, is still reasonable.     

进化神经网络在信用卡欺诈检测中的应用

自动信用卡欺诈检测是一个重要且有潜力的领域.基于人工神经网络的欺诈检测系统虽能令人满意.但具有良好结构的神经网络是很难构造的.由此提出一种进化方法来自适应地生成用于欺诈检测的神经网络结构.实验结果表明,该进化神经网络可以有效地完成信用卡欺诈检测.     

人工神经网络控制的实时仿真系统

为了观测使用人工神经网络作为控制器的实际的自动控制系统的控制效果,我们开发了一种实时闭环仿真系统.该仿真系统具有两个独立的部分:一部分是被控对象部分,由PC计算机计算被控对象的数学模型来模拟;另一部分为控制器部分,由真正的神经网络硬件实现.两部分由硬件接口电路连接在一起.此仿真系统工作于真正的时间轴中,即数学模型中的时间常数不再仅仅是计算中的参数而是反映真正的时间长度,它满足检验用于实际系统的神经网络控制器性能的需要.实验结果表明,此实时仿真系统对于设计基于人工神经网络的控制系统是一种有用的工具.     

复杂系统拓扑结构对人工神经网络的优化综述

随着人工神经网络在不同领域的成功应用,改变网络结构以优化其性能成为近年来的研究热点。由于人工神经网络具有广泛的连通性和复杂的结构,在获得高性能的同时,设计时间、布线成本和空间成本都更低的稀释型人工神经网络备受关注。复杂系统理论主要考虑结构对网络整体行为的影响,并将其应用于人工神经网络,使其具有更高的效率和更简单的结构。研究表明,复杂随机拓扑结构在连通性较低的情况下也要优于全连接人工神经网络。但根据神经生物学的研究,具有短特征路径长度和无标度分布的高度聚集的神经元更受青睐,且连接成本更低。因此,将小世界和无标度拓扑应用于人工神经网络成了相关领域的研究热点。文中总结和讨论了小世界、无标度和混合复杂网络与传统的全连接和随机结构对人工神经网络性能的影响。