Regression using hybrid Bayesian networks: Modelling landscape–socioeconomy relationships

Modelling environmental systems becomes a challenge when dealing directly with continuous and discrete data simultaneously. The aim in regression is to give a prediction of a response variable given the value of some feature variables. Multiple linear regression models, commonly used in environmental science, have a number of limitations: (1) all feature variables must be instantiated to obtain a prediction, and (2) the inclusion of categorical variables usually yields more complicated models. Hybrid Bayesian networks are an appropriate approach to solve regression problems without such limitations, and they also provide additional advantages. This methodology is applied to modelling landscape–socioeconomy relationships for different types of data (continuous, discrete or hybrid). Three models relating socioeconomy and landscape are proposed, and two scenarios of socioeconomic change are introduced in each one to obtain a prediction. This proposal can be easily applied to other areas in environmental modelling.     

基于预测关系的贝叶斯网络学习算法

在介绍有代表性的贝叶斯网络结构学习算法基础上,给出了变量之间预测能力的概念及估计方法,并证明了预测能力就是预测正确率,在此基础上建立了基于变量之间预测关系的贝叶斯网络结构学习方法,并使用模拟数据进行了对比实验,实验结果显示该算法能够有效地进行贝叶斯网络结构学习。     

Estimation of individual prediction reliability using the local sensitivity analysis

For a given prediction model, some predictions may be reliable while others may be unreliable. The average accuracy of the system cannot provide the reliability estimate for a single particular prediction. The measure of individual prediction reliability can be important information in risk-sensitive applications of machine learning (e.g. medicine, engineering, business). We define empirical measures for estimation of prediction accuracy in regression. Presented measures are based on sensitivity analysis of regression models. They estimate reliability for each individual regression prediction in contrast to the average prediction reliability of the given regression model. We study the empirical sensitivity properties of five regression models (linear regression, locally weighted regression, regression trees, neural networks, and support vector machines) and the relation between reliability measures and distribution of learning examples with prediction errors for all five regression models. We show that the suggested methodology is appropriate only for the three studied models: regression trees, neural networks, and support vector machines, and test the proposed estimates with these three models. The results of our experiments on 48 data sets indicate significant correlations of the proposed measures with the prediction error.     

Adaptive transfer learning in deep neural networks: Wind power prediction using knowledge transfer from region to region and between different task domains

Transfer learning (TL) in deep neural networks is gaining importance because, in most of the applications, the labeling of data is costly and time consuming. Additionally, TL also provides an effective weight initialization strategy for deep neural networks. This paper introduces the idea of adaptive TL in deep neural networks (ATL‐DNN) for wind power prediction. Specifically, we show in case of wind power prediction that adaptive TL of the deep neural networks system can be adaptively modified as regards training on a different wind farm is concerned. The proposed ATL‐DNN technique is tested for short‐term wind power prediction, where continuously arriving information has to be exploited. Adaptive TL not only helps in providing good weight initialization, but also in utilizing the incoming data for effective learning. Additionally, the proposed ATL‐DNN technique is shown to transfer knowledge between different task domains (wind power to wind speed prediction) and from one region to another region. The simulation results show that the proposed ATL‐DNN technique achieves average values of 0.0637, 0.0986, and 0.0984 for the mean absolute error, root mean squared error, and standard deviation error, respectively.     

相似性特征对链路预测的影响与增强

链路预测的主要任务是设计一个能够更加准确地描述给定网络机制的预测算法,从而得到更准确的预测结果。在分析现有研究成果基础上发现,网络的相似性特征对采用的链路预测方法有较大的影响：在节点间标签相似性较低的网络中,提高标签的相似性可以提高预测的准确性;而在节点间标签相似性较高的网络中,则应更加关注结构信息对于链路预测的贡献来提高预测的准确性。随后,通过对标签进行加权处理,提出带权值的标签相似性算法,在低相似性网络中能够提高链路预测的准确性。同时,在较高相似性网络中,将网络的结构信息引入到节点的相似性计算中,并通过偏好链接机制来提升链路预测的准确性。在四个真实网络上的实验结果表明,所提算法相对于标签系统间的余弦相似性（CSTS）算法、偏好链接（PA）等算法取得了最高的准确率。根据网络相似性特征,采用所提出的对应算法进行链路预测能够得到更准确的预测结果。     

基于元路径与节点属性的合著关系预测

在异构社会网络中,合著关系的预测是具有代表性的一类关系预测,与同构网络的链接预测方法在节点表示、网络构造等方面存在较大差异。综合考虑异构社会网络特有的元路径信息和节点属性特征,提出了节点的复合向量化表示：将节点的TF-IDF特征与基于Metapath2vec算法的向量化表示相结合;在元路径的表示上采取元路径中同类型节点归并重构的方法,以提取元路径中同类型节点间的隐含信息;并通过卷积神经网络（CNN）实现学术网络的合著关系预测。实验结果表明,节点的复合向量化表示及重构元路径方法可以更好地表征异构社会网络,与其他方法对比中均获得更好的预测评价指标。     

基于构造性神经网络的时间序列混合预测模型 *

针对传统时间序列预测模型不适应非线性预测而适应非线性预测的BP算法存在收敛速度慢,且容易陷入局部极小等问题,提出一种基于构造性神经网络的时间序列混合预测模型。采用构造性神经网络模型（覆盖算法）得出的类别值对统计时间序列模型的预测值进行修正,建立一种同时考虑时间序列自身周期变化和外生变量因子对时间序列未来变化趋势影响的混合预测模型,涵盖了实际问题的线性和非线性两方面,提高了预测精度。将该模型应用到粮食产量的预测中,取得了较好的预测效果。     

基于构造性神经网络的时间序列混合预测模型 *

针对传统时间序列预测模型不适应非线性预测而适应非线性预测的 BP算法存在收敛速度慢 ,且容易陷入局部极小等问题 ,提出一种基于构造性神经网络的时间序列混合预测模型。采用构造性神经网络模型 (覆盖算法 )得出的类别值对统计时间序列模型的预测值进行修正 ,建立一种同时考虑时间序列自身周期变化和外生变量因子对时间序列未来变化趋势影响的混合预测模型 ,涵盖了实际问题的线性和非线性两方面 ,提高了预测精度。将该模型应用到粮食产量的预测中 ,取得了较好的预测效果。     

神经网络在软件失效预测中的性能比较和研究

在软件开发的早期预测有失效倾向的软件模块,能够极大地提高软件的质量.软件失效预测中的一个普遍问题是数据中噪声的存在.神经网络具有鲁棒性而且对噪声有很强的抑制能力.不同结构的神经网络在训练算法和应用领域都有差异.该文主要就软件失效预测这个应用领域叙述几种适用的网络,并比较这几种网络在训练结果和性能上的差异.上述方法在SDH通信软件的失效预测中得到了成功的应用.试验结果显示虽然MLP、PNN、LVQ网络都能解决这类模式分类问题,但是只有MLP网络训练结果比较稳定,在不同的数据集上训练出的网络都有很好的预测效果.     

一种改进的异构链路协同预测算法研究

现有的链路预测方法仅考虑单种链路类型预测或多种链路类型的独立预测,经常使得预测结果不够准确。为此,研究了异构信息网络中多种链路类型的协同预测问题。根据源节点的相似节点和目标节点的相似节点之间的当前链路信息,提出了同质连接原理,设计了一种针对不同类型节点的相关性指标,用于描述不同类型节点间的链路存在概率,并将其与传统的邻近性指标相结合拓展到异构链路预测中。然后,将异构信息网络中的被标记数据和无标记数据融合起来,提出一种异构链路协同预测算法(Heterogeneous Collective Link Prediction, HCLP),通过获得不同类型链路间的各种复杂关系,结合互补性预测信息,实现多种链路类型的协同预测。基于真实场景的实验结果表明,所提的链路协同预测方法可有效提升异构信息网络的链路预测性能。     

Neural networks and statistical techniques: A review of applications

Neural networks are being used in areas of prediction and classification, the areas where statistical methods have traditionally been used. Both the traditional statistical methods and neural networks are looked upon as competing model-building techniques in literature. This paper carries out a comprehensive review of articles that involve a comparative study of feed forward neural networks and statistical techniques used for prediction and classification problems in various areas of applications. Tabular presentations highlighting the important features of these articles are also provided. This study aims to give useful insight into the capabilities of neural networks and statistical methods used in different kinds of applications.     

基于复杂网络动力学模型的链路预测方法

链路预测是复杂网络中研究缺失连边和未来形成连边的重要组成部分,当前基于网络结构的链路预测方法成果丰富,而基于复杂网络动力学模型的链路预测研究较少。针对无权无向网络,首先构建了复杂网络动力学模型,然后给出了基于复杂网络动力学模型的链路预测节点中心性的量化评价指标,最后通过给出的节点中心性量化指标,提出了由复杂网络动力学模型定义的链路预测方法。通过在真实网络数据集上进行的实验表明,提出的链路预测方法较基准方法有明显的预测精度的提升。     

基于个体选择的动态权重神经网络集成方法研究

神经网络集成技术能有效地提高神经网络的预测精度和泛化能力,已成为机器学习和神经计算领域的一个研究热点。该文针对回归分析问题提出了一种结合应用遗传算法进行个体选择和动态确定结果合成权重的神经网络集成构造方法。在训练出个体神经网络之后,应用遗传算法对个体网络进行选择,然后根据被选择的各个体网络在输入空间上对训练样本的预测误差,应用广义回归网络来动态地确定各个体网络在特定输入空间上的合成权重。实验结果表明,与仅应用个体网络选择或动态确定权重的方法相比,该集成方法基本上能取得更好地预测精度和相近的稳定性。     

Exchange rate forecasting: comparison of various architectures of neural networks

This paper evaluates the predictive accuracy of neural networks in forecasting exchange rate. The multi-layer perceptron (MLP) and radial basis function (RBF) networks with different architectures are used to forecast five exchange rate time series. The results of each prediction are evaluated and compared according to the networks and architectures used. It is found that neural networks can be effectively used in forecasting exchange rate and hence in designing trading strategies. RBF networks performed better than MLP networks in our simulation experiment. This experiment suggests that it is possible to extract information hidden in the exchange rate and predict it into future.     

Centrality prediction in dynamic human contact networks

Real technological, social and biological networks evolve over time. Predicting their future topology has applications to epidemiology, targeted marketing, network reliability and routing in ad hoc and peer-to-peer networks. The key problem for such applications is usually to identify the nodes that will be in more important positions in the future. Previous researchers had used ad hoc prediction functions. In this paper, we evaluate ways of predicting a node’s future importance under three important metrics, namely degree, closeness centrality, and betweenness centrality, using empirical data on human contact networks collected using mobile devices. We find that node importance is highly predictable due to both periodic and legacy effects of human social behaviour, and we design reasonable prediction functions. However human behaviour is not the same in all circumstances: the centrality of students at Cambridge is best correlated both daily and hourly, no doubt due to hourly lecture schedules, while academics at conferences exhibit rather flat closeness centrality, no doubt because conference attendees are generally trying to speak to new people at each break. This highlights the utility of having a number of different metrics for centrality in dynamic networks, so as to identify typical patterns and predict behaviour. We show that the best-performing prediction functions are 25% more accurate on average than simply using the previous centrality value. These prediction functions can be efficiently computed in linear time, and are thus practical for processing dynamic networks in real-time.     

基于三元组结构的有向网链路预测方法

当前链路预测的研究主要集中在无向网络,然而现实世界中存在大量的有向网络,忽略链路的方向会缺失一些重要信息甚至使预测失去意义,而直接将无向网络的预测方法应用于有向网络又存在预测精度降低的问题。为此,提出了一个基于三元组的有向网络链路预测算法,该算法针对有向网络和无向网络三元组结构的不同,应用势理论对三元组进行筛选,通过统计分析不同三元组闭合的可能性,以网络整体三元组闭合指数作为权重计算节点间的相似性。在9个真实数据集上的实验表明,所提方法比基准方法的预测精度提高了4.3%。     

Evolutionary product unit neural networks for short-term wind speed forecasting in wind farms

Combinations of physical and statistical wind speed forecasting models are frequently used in wind speed prediction problems arising in wind farms management. Artificial neural networks can be used in these models as a final step to obtain accurate wind speed predictions. The aim of this work is to determine the potential of evolutionary product unit neural networks (EPUNNs) for improving the accuracy and interpretation of these systems. Traditional neural network and EPUNN approaches have been used to develop different wind speed prediction models. The results obtained using different EPUNN models show that the functional model and the hybrid algorithms proposed provide very accurate prediction compared with standard neural networks used to solve this regression problem. One of the main advantages of the application of these EPUNNs has been the possibility of obtaining some interpretation of the non-linear relation predicted by the model, as will be shown in real data of a wind farm in Spain.     

Predicting grinding burn using artificial neural networks

This paper introduces a method for predicting grinding burn using artificial neural networks (ANN). First, the way to model grinding burn via ANN is presented. Then, as an example, the prediction of grinding burn of ultra-strength steel 300M via ANN is given. Very promising results were obtained.     

Online Text Prediction with Recurrent Neural Networks

Arithmetic coding is one of the most outstanding techniques for lossless data compression. It attains its good performance with the help of a probability model which indicates at each step the probability of occurrence of each possible input symbol given the current context. The better this model, the greater the compression ratio achieved. This work analyses the use of discrete-time recurrent neural networks and their capability for predicting the next symbol in a sequence in order to implement that model. The focus of this study is on online prediction, a task much harder than the classical offline grammatical inference with neural networks. The results obtained show that recurrent neural networks have no problem when the sequences come from the output of a finite-state machine, easily giving high compression ratios. When compressing real texts, however, the dynamics of the sequences seem to be too complex to be learned online correctly by the net.     

A Neural Network Based Model for Prognosis of Early Breast Cancer

The prediction of clinical outcome of patients after breast cancer surgery plays an important role in medical tasks such as diagnosis and treatment planning. Survival estimations are currently performed by clinicians using non-numerical techniques. Artificial neural networks are shown to be a powerful tool for analyzing data sets where there are complicated nonlinear interactions between the input data and the information to be predicted. In this paper, a new estimation to set the maximum bound on prediction accuracy is presented, based on the approximation of the a posteriori probability of Bayes by feed-forward three-layer neural networks. This result is applied to different patients' follow-up time intervals, in order to obtain the best prediction accuracy for the correct classification probability of patient relapse after breast cancer surgery using clinical-pathological data (tumor size, patient age, menarchy age, etc.), which were obtained from the Medical Oncology Service of the Hospital Clinico Universitario of Malaga, Spain. Different network topologies and learning parameters are investigated to obtain the best prediction accuracy. The actual results show as, after training process, the final model is appropriate to make predictions about the relapse probability at different times of follow-up.