An improved scheme for direct adaptive control of dynamical systems using backpropagation neural networks

This paper presents an improved direct control architecture for the on-line learning control of dynamical systems using backpropagation neural networks. The proposed architecture is compared with the other direct control schemes. In this scheme the neural network interconnection strengths are updated based on the output error of the dynamical system directly, rather than using a transformed version of the error employed in other schemes. The ill effects of the controlled dynamics on the on-line updating of the network weights are moderated by including a compensating gain layer. An error feedback is introduced to improve the dynamic response of the control system. Simulation studies are performed using the nonlinear dynamics of an underwater vehicle and the promising results support the effectiveness of the proposed scheme.     

An adaptive step size for backpropagation using linear lowerbounding functions

An adaptive step size is presented for the backpropagation algorithm in feedforward neural nets using linear lower bounding functions. Basically, a linear lower bounding function (LLBF) for a given function over an interval is a linear function that lies below the given function and matches the original function value at one end point. To search for an adaptive step size, an LLBF for the error function, which is expressed in terms of the step size, is derived. Since the error in a neural net can never be smaller than zero, it is plausible not to take a step larger than the step size when the associated LLBF reaches zero. In the paper, an adaptive learning algorithm based on the above idea is given. Numerical examples are used to illustrate its feasibility and to compare it with some previous results     

一种优化的LTPB网络带宽分配方法及实时性能分析

研究LTPB（线性令牌传递总线）通信网络中带宽分配方法对网络实时性能的影响。提出了一种优化的带宽分配方法和基于“最差情形下可达负载率”的网络实时性能分析方法，证明了该带宽分配方法优于所有传统的分配方法，最后举例说明了结论的正确性。     

Space instrument neural network for real-time data analysis

A simple software implementation of an artificial neural network (ANN) was used to analyze up to 200 autocorrelation functions (ACFs) per second within the Shuttle Potential and Return Electron Experiment (SPREE) flown on the Shuttle STS46 mission, July 31, 1992. As all ACF data are stored onboard until postmission, this facility provided ground-based experimenters with their only access to ACF data in real time for optimum instrument control. ACFs contain data either as waveforms or as radar echoes. Operating directly on the ACF, the neural network identifies the type of data, ascertains the wave frequency or radar peak separation, and provides a score or measure of significance of its decision. An effective 16:1 data reduction is achieved and the data interpretation performance is comparable to that achieved by an expert data analyst. Erroneous analysis accounts for less than 1% of data analyzed     

Asymptotic performance analysis of subspace adaptive algorithmsintroduced in the neural network literature

In the neural network literature, many algorithms have been proposed for estimating the eigenstructure of covariance matrices. We first show that many of these algorithms, when presented in a common framework, show great similitudes with the gradient-like stochastic algorithms usually encountered in the signal processing literature. We derive the asymptotic distribution of these different recursive subspace estimators. A closed-form expression of the covariances in distribution of eigenvectors and associated projection matrix estimators are given and analyzed. In particular, closed-form expressions of the mean square error of these estimators are given. It is found that these covariance matrices have a structure very similar to those describing batch estimation techniques. The accuracy of our asymptotic analysis is checked by numerical simulations, and it is found to be valid not only for a “small” step size but in a very large domain. Finally, convergence speed and deviation from orthonormality of the different algorithms are compared, and several tradeoffs are analyzed     

An adaptive network traffic prediction approach for LDoS attacks detection

Low‐rate denial of service (LDoS) attacks reduce throughput and degrade quality of service (QoS) of network services by sending out attack packets with relatively low average rate. LDoS attack flows are difficult to detect from normal traffic since it has the property of low average rate. The research on network traffic analysis and modeling shows that network traffic measurement data are irregular nonlinear time series. To characterize and analyze network traffic between attack and non‐attack situations, the adaptive normal and abnormal ν‐support vector regression (ν‐SVR) prediction models are constructed on the basis of the reconstructed phase space. In this paper, the dimension of reconstructed phase space for ν‐SVR is optimized by Bayesian information criteria method, and the parameter in the radial basis function is adaptively adjusted by minimizing the within‐class distance and maximizing the between‐class distance in the feature space. The nonthreshold decision function is obtained through calculating the prediction error of adaptive normal and abnormal ν‐SVR prediction models, which is adopted to detect LDoS attacks. Experiments in NS‐2 environment show that the adaptive ν‐SVR prediction model can effectively predict the network traffic measurement time series, and the probability distribution of time series generated by the adaptive ν‐SVR prediction model is quite similar to that of the network traffic measurement data. Experiments also clearly demonstrate the superiority of the proposed approach in LDoS attacks detection.     

基于深度神经网络的自适应波束形成算法

阵列天线接收到的期望信号和干扰信号,其入射的到达角度(Angle of Arrival,AOA)总是快速变化的,而传统波束形成算法计算量大,无法实时计算。针对这一问题,提出了一种基于深度神经网络的自适应波束形成(Deep Neural Network Adaptive Beamforming,DNNABF)算法,用入射信号AOA组成的向量作为网络输入,网络输出逼近最小方差无失真响应(Minimum Variance Distortionless Response,MVDR)算法求得的权矢量。仿真结果表明,卷积神经网络(Convolutional Neural Network,CNN)与DNNABF方法都能准确拟合MVDR算法权矢量,可在入射信号AOA快速变化时自适应地形成波束和零陷,但DNN计算速度相对MVDR有将近6.5倍的提升,训练模型时间也远低于CNN。     

An improved neural network realization for reliability analysis

In this paper we present an improved neural network training algorithm and architecture for reliability analysis of a simplex system and a TMR system which includes the effects of permanent fault and intermittent fault. A fully-connected three-layer neural network represents a discrete-time n-state reliability Markov model of a fault-tolerant system. The desired reliability of the system is fed into the neural network, and when the neural network converges, the design parameters are retrieved from the weights of the neural network. Finally, the simulation results show that the proposed method converges faster than other methods, especially in the case of the state number of the Markov model, which increases. This technique is also suitable for any system.     

AN ALGORITHM FOR VLSI CHANNEL ROUTING USING NEURAL NETWORK

The algorithm for VLSI channel routing using Hopfield neural model is discussed inthis paper.The basic methods of mapping VLSI channel routing problem to Hopfield neural net-work,constructing energy function,setting initial neural status,and selecting various parametersare proposed.Finally,some experimental results are given.     

Simulation of SiGe:C HBTs using neural network and adaptive neuro-fuzzy inference system for RF applications

In this article, the small-signal equivalent circuit model of SiGe:C heterojunction bipolar transistors (HBTs) has directly been extracted from S-parameter data. Moreover, in this article, we present a new modelling approach using ANFIS (adaptive neuro-fuzzy inference system), which in general has a high degree of accuracy, simplicity and novelty (independent approach). Then measured and model-calculated data show an excellent agreement with less than 1.68?×?10^?5% discrepancy in the frequency range of higher than 300 GHz over a wide range of bias points in ANFIS. The results show ANFIS model is better than ANN (artificial neural network) for redeveloping the model and increasing the input parameters.     

Real-time parallel adaptive neural network control for nonlinear servomechanisms—an approach using direct adaptive techniques

In this paper, a parallel adaptive neural network control system applicable to nonlinear dynamical systems of the type commonly encountered in many practical position control servomechanisms is developed. The controller is based on the use of direct adaptive techniques and an approach of using an additional parallel neural network to provide adaptive enhancements to a basic fixed neural network-based nonlinear controller. Properties of the proposed new controller are discussed in the paper and it is shown that if Gaussian radial basis function networks are used for the additional parallel neural network, uniformly stable adaptation is assured and asymptotic tracking of the position reference signal is achieved. The effectiveness of the proposed adaptive neural network control system is demonstrated in real-time implementation experiments for position control in a servomechanism with asymmetrical loading and changes in the load.     

基于自适应对比度增强的红外小目标检测网络

红外探测系统需要尽早发现目标以便及时拦截,但是红外图像上的小目标检测是一个挑战十足的任务。为了提高检测准确率,提出一种基于自适应对比度增强的红外小目标检测方法。为了利用自注意力机制和卷积各自的优势,设计了一个高效的特征提取网络和一个面向小目标的检测头。同时为了解决实际应用中出现的弱目标,在检测子网络前添加了一个图像预处理子网络,该模块可以自适应地调节图像对比度。在红外空中小目标数据集上的实验表明,提出的方法能达到93.76%的检测精度,与经典的检测方法相比,能够更好地平衡检测精度和召回率,证明了方法的巨大应用潜力。     

Nonlocal low-rank matrix completion for image interpolation using edge detection and neural network

In this paper, we propose a nonlocal low-rank matrix completion method using edge detection and neural network to effectively exploit the nonlocal inter-pixel correlation for image interpolation and other possible applications. We first interpolate the images using some basic techniques, such as bilinear and edge-directed methods. Then, each image patch is categorized as smooth regions, edge regions, or texture regions and adaptive interpolating mechanisms are applied to each specific type of regions. Finally, for each specific type of regions, neural networks and low-rank matrix completion are employed to accurately update the results. An iteratively re-weighted minimization algorithm is used to solve the low-rank energy minimization function. Our experiments on benchmark images clearly indicate that the proposed method produces much better results than some existing algorithms using a variety of image quality metric in terms of both objective image quality assessment and subjective quality assessment.     

红外目标检测网络改进半监督迁移学习方法

针对红外数据集规模小,标记样本少的特点,提出了一种红外目标检测网络的半监督迁移学习方法,主要用于提高目标检测网络在小样本红外数据集上的训练效率和泛化能力,提高深度学习模型在训练样本较少的红外目标检测等场景当中的适应性。文中首先阐述了在标注样本较少时无标注样本对提高模型泛化能力、抑制过拟合方面的作用。然后提出了红外目标检测网络的半监督迁移学习流程:在大量的RGB图像数据集中训练预训练模型,后使用少量的有标注红外图像和无标注红外图像对网络进行半监督学习调优。另外,文中提出了一种特征相似度加权的伪监督损失函数,使用同一批次样本的预测结果相互作为标注,以充分利用无标注图像内相似目标的特征分布信息;为降低半监督训练的计算量,在伪监督损失函数的计算中,各目标仅将其特征向量邻域范围内的预测目标作为伪标注。实验结果表明,文中方法所训练的目标检测网络的测试准确率高于监督迁移学习所获得的网络,其在Faster R-CNN上实现了1.1%的提升,而在YOLO-v3上实现了4.8%的显著提升,验证了所提出方法的有效性。     

一种实时雷达脉冲信号检测算法及其性能分析

提出了一种新的实时雷达脉冲信号检测算法,该算法首先将数据分为两路,对一路进行单点滑动、取共轭,然后与另一路信号相乘,再累加、取模,最后与门限比较,得到检测结果。算法具有递推和流水结构,硬件实现时只需一个复数乘法器、一个复数加法器、一个复数减法器和一个复数取模运算器。在此采用一阶扰动分析,推导了算法起始点检测误差的解析式,给出了算法性能边界,仿真结果验证了理论推导的正确性。与其他信号检测算法相比,该算法结构规整,易于硬件应用,可实现实时检测。     

Tamper video detection and localization using an adaptive segmentation and deep network technique

In this work we have explored the hybrid deep learning architecture for recognizing the tampering from the videos. This hybrid architecture explores the features from the authentic videos to categorize the tampered portions from the forged videos. Initially, the process begins by compressing the input video using the Discrete cosine transform (DCT) based double compression approach. Then, the filtering process is carried out to improve the quality of compressed frame using the bilateral filtering. Then, the modified segmentation approach is applied to segment the frames into different regions. The features from these segmented portions are extracted and fed into hybrid DNN-AGSO (deep neural network- Adaptivf RELATED WORKSe Galactic Swarm Optimization) using Gabor wavelet transform (GWT) technique. Three different datasets are used to evaluate the overall performance they are, VTD, MFC-18, and VIRAT by MATLAB platform. The recognition rate achieved by VTD, MFC-18, and VIRAT datasets are 96%, 95.2%, and 93.47% respectively.     

Fault detection and diagnosis in asymmetric multilevel inverter using artificial neural network

The increased component requirement to realise multilevel inverter (MLI) fallout in a higher fault prospect due to power semiconductors. In this scenario, efficient fault detection and diagnosis (FDD) strategies to detect and locate the power semiconductor faults have to be incorporated in addition to the conventional protection systems. Even though a number of FDD methods have been introduced in the symmetrical cascaded H-bridge (CHB) MLIs, very few methods address the FDD in asymmetric CHB-MLIs. In this paper, the gate-open circuit FDD strategy in asymmetric CHB-MLI is presented. Here, a single artificial neural network (ANN) is used to detect and diagnose the fault in both binary and trinary configurations of the asymmetric CHB-MLIs. In this method, features of the output voltage of the MLIs are used as to train the ANN for FDD method. The results prove the validity of the proposed method in detecting and locating the fault in both asymmetric MLI configurations. Finally, the ANN response to the input parameter variation is also analysed to access the performance of the proposed ANN-based FDD strategy.     

Epileptic seizures detection in EEG using DWT-based ApEn and artificial neural network

There are numerous neurological disorders such as dementia, headache, traumatic brain injuries, stroke, and epilepsy. Out of these epilepsy is the most prevalent neurological disorder in the human after stroke. Electroencephalogram (EEG) contains valuable information related to different physiological state of the brain. A scheme is presented for detecting epileptic seizures from EEG data recorded from normal subjects and epileptic patients. The scheme is based on discrete wavelet transform (DWT) analysis and approximate entropy (ApEn) of EEG signals. Seizure detection is performed in two stages. In the first stage, EEG signals are decomposed by DWT to calculate approximation and detail coefficients. In the second stage, ApEn values of the approximation and detail coefficients are calculated. Significant differences have been found between the ApEn values of the epileptic and the normal EEG allowing us to detect seizures with 100 % classification accuracy using artificial neural network. The analysis results depicted that during seizure activity, EEG had lower ApEn values compared to normal EEG. This gives that epileptic EEG is more predictable or less complex than the normal EEG. In this study, feed-forward back-propagation neural network has been used for classification and training algorithm for this network that updates the weight and bias values according to Levenberg–Marquardt optimization technique.     

Fault detection in analogue circuits using hybrid evolutionary algorithm and neural network

With the development of analog integrated circuits technology and due to the complexity, and various types of faults that occur in analog integrated circuits, fault detection is a new idea, has been studied in recent decades. In this paper a three amplifier state variable filter is used as circuit under test (CUT) and, a hybrid neural network is proposed for soft fault diagnosis of the CUT. Genetic algorithm (GA) has the powerful ability of searching the global optimal solution, and back propagation (BP) algorithm has the feature of rapid convergence on the local optima. The hybrid of two algorithm will improve the evolving speed of neural network. GA-BP scheme adopts GA to search the optimal combination of weights in the solution space, and then uses BP algorithm to obtain the accurate optimal solution quickly. Experiment results show that the proposed GA-BP scheme is more efficient and effective than BP algorithm.     

Prediction of matching condition for a microstrip subsystem using artificial neural network and adaptive neuro-fuzzy inference system

In this paper, a subsystem consisting of a microstrip bandpass filter and a microstrip low noise amplifier (LNA) is designed for WLAN applications. The proposed filter has a small implementation area (49 mm²), small insertion loss (0.08 dB) and wide fractional bandwidth (FBW) (61%). To design the proposed LNA, the compact microstrip cells, an field effect transistor, and only a lumped capacitor are used. It has a low supply voltage and a low return loss (–40 dB) at the operation frequency. The matching condition of the proposed subsystem is predicted using subsystem analysis, artificial neural network (ANN) and adaptive neuro-fuzzy inference system (ANFIS). To design the proposed filter, the transmission matrix of the proposed resonator is obtained and analysed. The performance of the proposed ANN and ANFIS models is tested using the numerical data by four performance measures, namely the correlation coefficient (CC), the mean absolute error (MAE), the average percentage error (APE) and the root mean square error (RMSE). The obtained results show that these models are in good agreement with the numerical data, and a small error between the predicted values and numerical solution is obtained.