Stability analysis of fractional order fuzzy cellular neural networks with leakage delay and time varying delays

In this paper we investigated the stability of fractional order fuzzy cellular neural networks with leakage delay and time varying delays. Based on Lyapunov theory and applying bounded techniques of fractional calculation, sufficient criterion are established to guarantee the stability. Hybrid feedback control is applied to derive the proposed results. Finally, numerical examples with simulation results are given to illustrate the effectiveness of the proposed method.     

Some methods of training radial basis neural networks in solving the Navier‐Stokes equations

The purpose of this research is to analyze the application of neural networks and specific features of training radial basis functions for solving 2‐dimensional Navier‐Stokes equations. The authors developed an algorithm for solving hydrodynamic equations with representation of their solution by the method of weighted residuals upon the general neural network approximation throughout the entire computational domain. The article deals with testing of the developed algorithm through solving the 2‐dimensional Navier‐Stokes equations. Artificial neural networks are widely used for solving problems of mathematical physics; however, their use for modeling of hydrodynamic problems is very limited. At the same time, the problem of hydrodynamic modeling can be solved through neural network modeling, and our study demonstrates an example of its solution. The choice of neural networks based on radial basis functions is due to the ease of implementation and organization of the training process, the accuracy of the approximations, and smoothness of solutions. Radial basis neural networks in the solution of differential equations in partial derivatives allow obtaining a sufficiently accurate solution with a relatively small size of the neural network model. The authors propose to consider the neural network as an approximation of the unknown solution of the equation. The Gaussian distribution is used as the activation function.     

RNAGCN: RNA tertiary structure assessment with a graph convolutional network

RNAs play crucial and versatile roles in cellular biochemical reactions. Since experimental approaches of determining their three-dimensional (3D) structures are costly and less efficient, it is greatly advantageous to develop computational methods to predict RNA 3D structures. For these methods, designing a model or scoring function for structure quality assessment is an essential step but this step poses challenges. In this study, we designed and trained a deep learning model to tackle this problem. The model was based on a graph convolutional network (GCN) and named RNAGCN. The model provided a natural way of representing RNA structures, avoided complex algorithms to preserve atomic rotational equivalence, and was capable of extracting features automatically out of structural patterns. Testing results on two datasets convincingly demonstrated that RNAGCN performs similarly to or better than four leading scoring functions. Our approach provides an alternative way of RNA tertiary structure assessment and may facilitate RNA structure predictions. RNAGCN can be downloaded from https://gitee.com/dcw-RNAGCN/rnagcn.     

Hyperparameter on-line learning of stochastic resonance based threshold networks

Aiming at training the feed-forward threshold neural network consisting of nondifferentiable activation functions, the approach of noise injection forms a stochastic resonance based threshold network that can be optimized by various gradient-based optimizers. The introduction of injected noise extends the noise level into the parameter space of the designed threshold network, but leads to a highly non-convex optimization landscape of the loss function. Thus, the hyperparameter on-line learning procedure with respective to network weights and noise levels becomes of challenge. It is shown that the Adam optimizer, as an adaptive variant of stochastic gradient descent, manifests its superior learning ability in training the stochastic resonance based threshold network effectively. Experimental results demonstrate the significant improvement of performance of the designed threshold network trained by the Adam optimizer for function approximation and image classification.     

基于毛竹叶片理化参数的刚竹毒蛾危害检测研究

虫害检测算法研究是开展虫害快速、准确监测，制定精准森防检疫措施的重要基础。以毛竹叶片为研究尺度，基于刚竹毒蛾危害下的寄主外部形态与内部生理现象总结，选择并实测叶损量LL、相对叶绿素含量RCC、相对含水量RWC、原始光谱的733.66~898.56 nm值(ρ_{733.66~898.56})、一阶微分光谱的562.95~585.25 nm值(ρ′_{562.95~585.25})与706.18~725.41 nm值(ρ′_{706.18~725.41})等理化参数，随机划分实验组(63组)和验证组(37组)并设计5次重复实验；分别运用Fisher判别分析、BP神经网络、随机森林等三种方法建立刚竹毒蛾危害等级的检测模型，从检测精度、Kappa系数及R2等指标对模型的检测效果予以分析和比较。结果显示，Fisher判别分析、BP神经网络、随机森林的检测精度分别为69.19%，65.41%，83.78%，Kappa系数分别为0.576 9，0.532 4和0.778 8，R2分别为0.722 2，0.582 6和0.870 9，总体而言，三种方法均具备刚竹毒蛾危害的检测能力，随机森林的检测效果最优，Fisher判别分析次之，再次为BP神经网络；从分等级来看，随机森林的检测精度亦优于Fisher判别分析与BP神经网络，但3种方法对中度危害等级的检测精度均有所不足。该成果可为刚竹毒蛾危害及其他病虫害检测算法的选择提供参考，并为进一步建立冠层、遥感影像像元等尺度的虫害检测模型奠定基础。     

贝叶斯优化卷积神经网络公共场所异常声识别*

针对公共场所异常声的感知和识别问题,提出一种基于贝叶斯优化卷积神经网络的识别方法。提取声信号的Gammatone倒谱系数、倍频程功率谱、短时能量和谱质心,组合成声信号的特征图。构建卷积神经网络作为分类器,利用递增的卷积核设置和池化操作处理不同尺度的特征。基于贝叶斯优化算法优化卷积神经网络的模型参数,对包括火苗噼啪声、婴儿啼哭声、烟花燃放声、玻璃破碎声和警报声的5种公共场所异常声进行识别。该方法的识别结果与基于不同的特征提取和分类器方案得到的识别结果进行比较,结果表明该方法的识别效果优于其他特征提取和分类器方案的识别效果。最后分析了该方法在不同信噪比噪声干扰下的识别结果,验证了该方法的有效性。     

Modeling the Effect of Injection Molding Process Parameters on Warpage Using Neural Network Theory

A back propagation artificial neural network (BPANN) prediction model for warpage of injection-molded polypropylene was developed based on an orthogonal design method. The BPANN model was trained by the input and output data obtained from the moldflow software platform simulations. It is proved that the BPANN model can predict the warpage with reasonable accuracy. Utilizing the BPANN model, the effects of the process parameters, packing pressure (Pp), melt temperature (T_me), mold temperature (T_mo), packing time (t_p), cooling time (t_c), and fill pressure (p_f), on the warpage were investigated. The most important process parameter affecting the warpage was Pp, and the second most important was T_me. The rest of the process parameters, T_mo, t_p, t_c, and p_f, were found to be relatively less influential. Warpage increased with elevating T_mo. In contrast, an increase in Pp and T_me caused the warpage to decrease.     

基于级联卷积神经网络的前列腺磁共振图像分类

针对深度学习训练成本高，以及基于磁共振图像的前列腺癌临床诊断需要大量医学常识且极为耗时的问题，本文提出了一种基于级联卷积神经网络（Convolutional Neural Network，CNN）和磁共振图像的前列腺癌（Prostate Cancer，PCa）自动分类诊断方法，该网络以Faster-RCNN作为前网络，对前列腺区域进行提取分割，用于排除前列腺附近组织器官的干扰；以基于ResNet改进的网络结构CNN40_bottleneck作为后网络，用于对前列腺区域病变进行分类.后网络由瓶颈结构串联组成，其中使用批量标准化（Batch Normalization，BN）、全局平均池化（Global Average Pooling，GAP）进行优化.实验结果证明，本文方法对前列腺癌诊断结果较好，而且缩减了训练时间和参数量，有效降低了训练成本.     

Improved neural networks based on genetic algorithm for pulse recognition

Pulse diagnosis is an important part of Chinese medicine and has played an important role in the development of Chinese medical science. However, the pulse is traditionally determined by cutting it off, which leads to a lack of objective standard pulse identification methods and affects their accuracy and feasibility. This research has studied and discussed the processing and identification of four kinds of pulse: normal pulse, wiry pulse, smooth pulse, and thready pulse. Four frequency-domain characteristics of the pulse wave and six kinds of wavelet scale energy characteristic information were extracted, and a three-layer BP (backprocessing) neural network was established. The LM (Levenberg–Marquard) algorithm and a genetic algorithm were used to improve the BP neural network, to train on and predict experimental samples, and to obtain classification accuracies of 90% and 95% respectively. Moreover, improved BP neural network based on a genetic algorithm has shown highly superior performance in terms of convergence speed and low error rate.