Quantum neural network

In the neural network theory content-addressable memories are defined by patterns that are attractors of the dynamical rule of the system. This paper develops a quantum neural network starting from a classical neural network Hamiltonian and using a Schrödinger-like equation. It then shows that such a system exhibits probabilistic memory storage characteristics analogous to those of the dynamical attractors of classical systems.     

应用导数荧光光谱和概率神经网络鉴别合成色素

实验测量了食品色素胭脂红、苋菜红、诱惑红和工业色素苏丹红Ⅳ溶液分别在波长为300,400,440和380 nm的光激发下产生的荧光光谱.对这4种红色素的各8个溶液样本选取60个发射波长值所对应的荧光强度作为网络特征参数,训练、建立概率神经网络.据此,对32个色素溶液样本进行种类识别.为解决原始荧光光谱重叠造成识别准确率不高的问题,应用导数荧光光谱,将二阶导数光谱数据作为网络特征参数,建立网络,进行识别,识别准确率达100%.由此,提出了应用二阶导数荧光光谱结合概率神经网络对合成色素方便、快捷、准确地进行种     

概率神经网络在胃镜样品红外光谱检测中的应用

将人工神经网络方法应用于人体胃镜样品红外光谱检测,以克服常规线性判别分析方法的局限性,从而提高了胃镜样品判别的准确率。概率神经网络是一种适用模式分类的径向基神经网络,采用样本的先验概率和最优判定原则对新的样本进行分类,具有识别率高、训练速度快、不会陷入局部极值等优点。文章采用概率神经网络进行胃镜样品红外光谱模式识别,将预处理后的胃镜样品光谱进行主成分分析,将得分值作为输入,建立概率神经网络判别模型。文中选取118例胃镜离体样品进行红外光谱判别分析,其中正常胃组织19例,胃炎组织64例,胃癌35例,选取其中59例样品建立概率神经网络校正模型,其余样品作为预测集来检验模型。实验结果表明,正常、炎症及癌症胃镜样品检测的总体准确率达到81.4%,对胃镜样品的判别取得了较好的结果。     

基于PNN网络和Freeman链码的抽油机井工况诊断

利用概率神经网络(PNN)对抽油机井工况进行诊断,建立了抽油机井工况诊断的概率神经网络模型。对示功图提取特征值的质量好坏直接影响识别效率和可靠性,提出了用Freeman链码对等效的电流示功图提取特征参数,进行预处理,建立抽油机典型工况的链码特征样本库。将Freeman链码作为特征向量,利用MATLAB对网络进行训练。结果表明,Freeman链码能够有效的识别各种典型工况示功图,并且该概率神经网络学习速度快、诊断准确率高,可用于抽油机井工况的实时监测和诊断。     

Exo-atmospheric target discrimination using probabilistic neural network

Exo-atmospheric targets are especially difficult to distinguish using currently available techniques, because all target parts follow the same spatial trajectory. The feasibility of distinguishing multiple type components of exo-atmospheric targets is demonstrated by applying the probabilistic neural network. Differences in thermal behavior and time-varying signals of space-objects are analyzed during the selection of features used as inputs of the neural network. A novel multi-colorimetric technology is introduced to measure precisely the temporal evolutional characteristics of temperature and emissivity-area products. To test the effectiveness of the recognition algorithm, the results obtained from a set of synthetic multispectral data set are presented and discussed. These results indicate that the discrimination algorithm can obtain a remarkable success rate.     

Theory of the dynamics of the Hopfield model of associative memory

We present an analysis of the parallel dynamics of the Hopfield model of the associative memory of a neural network without recourse to the replica formalism. A probabilistic method based on the signal-to-noise ratio is employed to obtain a simple recursion relation for the zero temperature as well as the finite temperature dynamics of the network. The fixed points of the recursion relation and their basins of attraction are found to be in fairly satisfactory agreement with the numerical simulations of the model. We also present some new numerical results which support our recursion relation and throw light on the nature of the ensemble of the network states which are optimized with respect to single spin flips.     

基于邻域粗糙集和并行神经网络的故障诊断

针对目前使用神经网络诊断故障时出现的输入向量选择困难、网络结构复杂、对并发故障诊断效果不好等问题,提出了基于邻域粗糙集和并行神经网络的故障诊断方法。先利用邻域粗糙集对初始征兆进行约简,留下有价值的征兆作为神经网络的输入向量,然后针对每种故障类型设计一个神经网络。用多个训练好的神经网络来并行诊断故障,综合每个神经网络的结果给出最终的诊断结论。用转子实验台的实验数据对这种故障诊断方法进行验证,结果显示该方法能优化神经网络结构,且神经网络具有训练速度快、诊断正确率高的特点。     

Probabilistic Models with Deep Neural Networks

Recent advances in statistical inference have significantly expanded the toolbox of probabilistic modeling. Historically, probabilistic modeling has been constrained to very restricted model classes, where exact or approximate probabilistic inference is feasible. However, developments in variational inference, a general form of approximate probabilistic inference that originated in statistical physics, have enabled probabilistic modeling to overcome these limitations: (i) Approximate probabilistic inference is now possible over a broad class of probabilistic models containing a large number of parameters, and (ii) scalable inference methods based on stochastic gradient descent and distributed computing engines allow probabilistic modeling to be applied to massive data sets. One important practical consequence of these advances is the possibility to include deep neural networks within probabilistic models, thereby capturing complex non-linear stochastic relationships between the random variables. These advances, in conjunction with the release of novel probabilistic modeling toolboxes, have greatly expanded the scope of applications of probabilistic models, and allowed the models to take advantage of the recent strides made by the deep learning community. In this paper, we provide an overview of the main concepts, methods, and tools needed to use deep neural networks within a probabilistic modeling framework.     

Designing neural networks that process mean values of random variables

We develop a class of neural networks derived from probabilistic models posed in the form of Bayesian networks. Making biologically and technically plausible assumptions about the nature of the probabilistic models to be represented in the networks, we derive neural networks exhibiting standard dynamics that require no training to determine the synaptic weights, that perform accurate calculation of the mean values of the relevant random variables, that can pool multiple sources of evidence, and that deal appropriately with ambivalent, inconsistent, or contradictory evidence.     

概率神经网络及FAAS在植物药分类研究中的应用

用火焰原子吸收法（FAAS）测定了植物药中Fe、Mg、Mn、Cu、Zn和Ca元素的含量,采用主成分分析法对所测数据进行预处理,结合概率神经网络模型对中药功效类别进行识别预测研究,取得了较满意的结果。     

The application of neural networks to forecast fuzzy time series

Fuzzy time series models have been applied to handle nonlinear problems. To forecast fuzzy time series, this study applies a backpropagation neural network because of its nonlinear structures. We propose two models: a basic model using a neural network approach to forecast all of the observations, and a hybrid model consisting of a neural network approach to forecast the known patterns as well as a simple method to forecast the unknown patterns. The stock index in Taiwan for the years 1991–2003 is chosen as the forecasting target. The empirical results show that the hybrid model outperforms both the basic and a conventional fuzzy time series models.     

Winsorization for Robust Bayesian Neural Networks

With the advent of big data and the popularity of black-box deep learning methods, it is imperative to address the robustness of neural networks to noise and outliers. We propose the use of Winsorization to recover model performances when the data may have outliers and other aberrant observations. We provide a comparative analysis of several probabilistic artificial intelligence and machine learning techniques for supervised learning case studies. Broadly, Winsorization is a versatile technique for accounting for outliers in data. However, different probabilistic machine learning techniques have different levels of efficiency when used on outlier-prone data, with or without Winsorization. We notice that Gaussian processes are extremely vulnerable to outliers, while deep learning techniques in general are more robust.     

Probabilistic Autoencoder Using Fisher Information

Neural networks play a growing role in many scientific disciplines, including physics. Variational autoencoders (VAEs) are neural networks that are able to represent the essential information of a high dimensional data set in a low dimensional latent space, which have a probabilistic interpretation. In particular, the so-called encoder network, the first part of the VAE, which maps its input onto a position in latent space, additionally provides uncertainty information in terms of variance around this position. In this work, an extension to the autoencoder architecture is introduced, the FisherNet. In this architecture, the latent space uncertainty is not generated using an additional information channel in the encoder but derived from the decoder by means of the Fisher information metric. This architecture has advantages from a theoretical point of view as it provides a direct uncertainty quantification derived from the model and also accounts for uncertainty cross-correlations. We can show experimentally that the FisherNet produces more accurate data reconstructions than a comparable VAE and its learning performance also apparently scales better with the number of latent space dimensions.     

基于PCA和PNN的高甘油三脂血清荧光光谱识别

基于主成分分析和概率神经网络,提出了一种有效识别高甘油三脂血清荧光光谱的新方法.研究测量了正常和高甘油三脂血清在290 nm和350 nm激发光下产生的荧光光谱,并分别以3种采样间隔(1 nm、2 nm和5 nm)提取荧光强度作为样品的初始特征;利用主成分分析法对初始特征进行分析,以累积可信度大于95%的主成分作为样品特征;构建了4层概率神经网络,并分析了平滑系数和采样间隔对识别效果的影响.实验结果表明,当采样间隔采用5 nm,平滑系数位于0.26~0.92区间时,正常和高甘油三脂血清样品的识别率可达到95%和100%.     

Fusion of visible and infrared imagery for face recognition

In recent years face recognition has received substantial attention, but still remained very challenging in real applications. Despite the variety of approaches and tools studied, face recognition is not accurate or robust enough to be used in uncontrolled environments. Infrared (IR) imagery of human faces offers a promising alternative to visible imagery, however, IR has its own limitations. In this paper, a scheme to fuse information from the two modalities is proposed. The scheme is based on eigenfaces and probabilistic neural network (PNN), using fuzzy integral to fuse the objective evidence supplied by each modality. Recognition rate is used to evaluate the fusion scheme. Experimental results show that the scheme improves recognition performance substantially.     

A powerful approach to real-time mobile objects tracking in crowded environments utilizing 3D video analysis of fixed camera,particle filter and neural network

A powerful approach in the area of real-time mobile objects tracking in crowded environments, utilizing 3D video frames analysis is now taken into real consideration, as a candidate to be improved. The method presented here is able to track a number of real-time mobile objects in the real complex situations in the presence of occlusion, overlapping and various shifts. This is a development of probabilistic estimation theory via particle filter. In one such case, the whole of chosen new features of mobile objects, which are unconsidered in the present probabilistic estimation, should first be analyzed through a novel neural network. Subsequently, the probabilistic estimation in each one of frames may be made in a better outcome, as long as all the mentioned components are integrated. Evaluation of the proposed approach through PETS-09 database has been finally carried out, once the results with respect to a number of standard benchmark procedures indicate that 12% accuracy improvement is acquired.     

A neural-computation method of predicting the early interaural cross-correlation coefficient (IACCE3) for auditoria

A method of predicting the early interaural cross-correlation coefficient (IACC_E3) in unoccupied concert halls has been investigated using neural network analysis. Constructional and acoustical data for 36 unoccupied concert halls, in various countries, were utilized for the neural network analyses. A neural network for calculating IACC_E3 has been embedded in a standard spreadsheet application so that designers and researchers, without access to specialized neural network software can use the results of the present work. Investigations using the neural network model have shown that IACC_E3 predictions are within the subjective difference limen, which is 0.075±0.008. Five concert halls were used to assess the neural network analysis method and the errors between measured and predicted (1−IACC_E3) ranged from −0.05 to 0.02. These results indicate that there is a good basis for using trained neural networks to predict IACC_E3.     

Modeling Information Popularity Dynamics via Branching Process on Micro-Blog Network

Predicting and modeling of items popularity on web 2.0 have attracted great attention of many scholars. From the perspective of information competition, we propose a probabilistic model using the branching process to characterize the process in which micro-blogging gains its popularity. The model is analytically tractable and can reproduce several characteristics of empirical micro-blogging data on Sina micro-blog, the most popular microblogging system in China. We find that the information competition on micro-blog network leads to the decay of information popularity obeying power law distribution with exponent about 1.5, and the value is similar to the exponent of degree distribution of micro-blog network. Furthermore, the mean popularity is decided by the probability of innovating a new message. Our work presents evidence supporting the idea that two distinct factors affect information popularity: information competition and social network structure.     

Chaotic diagonal recurrent neural network

We propose a novel neural network based on a diagonal recurrent neural network and chaos,and its structure and learning algorithm are designed.The multilayer feedforward neural network,diagonal recurrent neural network,and chaotic diagonal recurrent neural network are used to approach the cubic symmetry map.The simulation results show that the approximation capability of the chaotic diagonal recurrent neural network is better than the other two neural networks.