A pruning algorithm with L1/2regularizer for extreme learning machine简

Compared with traditional learning methods such as the back propagation （BP） method, extreme learning machine provides much faster learning speed and needs less human intervention, and thus has been widely used. In this paper we combine the L1/2 regularization method with extreme learning machine to prune extreme learning machine. A variable learning coefficient is employed to prevent too large a learning increment. A numerical experiment demonstrates that a network pruned by L1/2 regularization has fewer hidden nodes but provides better performance than both the original network and the network pruned by L2 regularization.     

Assessing safety at work using an adaptive neuro-fuzzy inference system (ANFIS) approach aided by partial least squares structural equation modeling (PLS-SEM)

The main objective of this research was to apply an adaptive neuro-fuzzy inference system (ANFIS) approach aided by Partial Least Squares Structural Equation Modeling (PLS-SEM) to assess safety at work, defined as employee propensity to follow safety regulations, including safe work practices at the workplace. A survey with seven main components: 1) use of mobile technology, 2) tacit safety knowledge, 3) explicit safety knowledge, 4) attitudes toward safety: psychological aspects, 5) attitudes toward safety: emotional aspects, 6) safety culture: behavioral aspects, and 7) safety culture: psychological aspects, was used for this purpose. Workers from three manufacturing companies located in southeastern Poland completed a paper-based survey. PLS-SEM, combined with an adaptive neuro-fuzzy inference system (ANFIS) method, was used to develop the study model and determine its main components. The results showed that tacit safety knowledge, attitudes toward safety: psychological aspects, attitudes toward safety: emotional aspects, safety culture: behavioral aspects, safety culture: psychological aspects, and the use of mobile technology were significant factors influencing the perceived safety at work. Moreover, the results of the ANFIS modeling approach showed that behavioral aspects of safety culture were the most critical predictor of the perceived safety at work.     

自适应混沌粒子群算法对极限学习机参数的优化

针对极限学习机（ELM）在处理非线性数据时效果不理想,并且ELM的参数随机化不利于模型泛化的特点,提出了一种改进的极限学习机算法。结合自适应混沌粒子群（ACPSO）算法对ELM的参数进行优化,以增强算法的稳定性,提高ELM对基因表达数据分类的精度。在UCI基因数据集上进行仿真实验,实验结果表明,与探测粒子群-极限学习机（DPSO-ELM）、粒子群-极限学习机（PSO-ELM）等算法相比,自适应混沌粒子群-极限学习机（ACPSO-ELM）算法具有较好的稳定性、可靠性,且能有效提高基因分类精度。     

Forecasting reservoir monthly runoff via ensemble empirical mode decomposition and extreme learning machine optimized by an improved gravitational search algorithm

Monthly streamflow prediction plays a significant role in reservoir operation and water resource management. Hence, this research tries to develop a hybrid model for accurate monthly streamflow prediction, where the ensemble empirical mode decomposition (EEMD) is firstly used to decompose the original streamflow data into a finite amount of intrinsic mode functions (IMFs) and a residue; and then the extreme learning machine (ELM) is employed to forecast each IMFs and the residue, while an improved gravitational search algorithm (IGSA) based on elitist-guide evolution strategies, selection operator and mutation operator is used to select the parameters of all the ELM models; finally, the summarized predicated results for all the subcomponents are treated as the final forecasting result. The hybrid method is applied to forecast the monthly runoff of Three Gorges in China, while four quantitative indexes are used to test the performances of the developed forecasting models. The results show that EEMD can effectively separate the internal characteristics of the original monthly runoff, and the hybrid model is able to make an obvious improvement over other models in hydrological time series prediction.     

半监督极限学习机及其在近红外光谱数据分析中的应用

当数据集中包含的训练信息不充分时,监督的极限学习机较难应用,因此将半监督学习应用到极限学习机,提出一种半监督极限学习机分类模型;但其模型是非凸、非光滑的,很难直接求其全局最优解。为此利用组合优化方法,将提出的半监督极限学习机化为线性混合整数规划,可直接得到其全局最优解。进一步,利用近红外光谱技术,将半监督极限学习机应用于药品和杂交种子的近红外光谱数据的模式分类。与传统方法相比,在不同的光谱区域的数值实验结果显示:当数据集中包含训练信息不充分时,提出的半监督极限学习机提高了模型的推广能力,验证了所提出方法的可行性和有效性。     

粒子群优化混合核极限学习机的构造煤厚度预测方法

在构造煤厚度的预测中,针对预测精度不高的问题,提出利用粒子群优化（PSO）算法优化极限学习机（ELM）的方法来对构造煤厚度进行预测。首先,利用主成分分析（PCA）对三维地震属性进行降维处理,在降低地震属性的维数的同时消除变量之间的相关性。然后,构建全局多项式核函数和局部高斯径向基核函数混合核极限学习机（HKELM）模型,并利用PSO算法优化HKELM的核参数。同时,针对PSO算法存在容易陷入局部最优的问题,在PSO算法中加入模拟退火的思想和随迭代次数减小的惯性权重,以及基于反向学习的变异操作,使PSO算法可以更容易跳出局部极小值点,得到更优结果。此外,为了增强模型的泛化能力,在核函数的基础上加入L2正则项,有效地避免了噪声和异常点对模型泛化性能的影响。最后,将预测模型应用到阳煤集团新景矿区芦南二采区中部15#煤层中,预测得到的采区构造煤厚度与实际地质资料具有较高的一致性。实验结果表明,利用改进PSO算法优化HKELM构建构造煤厚度预测模型的预测误差较小,可以推广用于实际采区的构造煤厚度预测。     

Application of adaptive neuro-fuzzy inference system for vigilance level estimation by using wavelet-entropy feature extraction

This paper presents the application of adaptive neuro-fuzzy inference system (ANFIS) model for estimation of vigilance level by using electroencephalogram (EEG) signals recorded during transition from wakefulness to sleep. The developed ANFIS model combined the neural network adaptive capabilities and the fuzzy logic qualitative approach. This study comprises of three stages. In the first stage, three types of EEG signals (alert signal, drowsy signal and sleep signal) were obtained from 30 healthy subjects. In the second stage, for feature extraction, obtained EEG signals were separated to its sub-bands using discrete wavelet transform (DWT). Then, entropy of each sub-band was calculated using Shannon entropy algorithm. In the third stage, the ANFIS was trained with the back-propagation gradient descent method in combination with least squares method. The extracted features of three types of EEG signals were used as input patterns of the three ANFIS classifiers. In order to improve estimation accuracy, the fourth ANFIS classifier (combining ANFIS) was trained using the outputs of the three ANFIS classifiers as input data. The performance of the ANFIS model was tested using the EEG data obtained from 12 healthy subjects that have not been used for the training. The results confirmed that the developed ANFIS classifier has potential for estimation of vigilance level by using EEG signals.     

State of charge estimation for Li-ion battery based on model from extreme learning machine

Lithium-ion (Li-ion) battery state of charge (SOC) estimation is important for electric vehicles (EVs). The model-based state estimation method using the Kalman filter (KF) variants is studied and improved in this paper. To establish an accurate discrete model for Li-ion battery, the extreme learning machine (ELM) algorithm is proposed to train the model using experimental data. The estimation of SOC is then compared using four algorithms: extended Kalman filter (EKF), unscented Kalman filter (UKF), adaptive extended Kalman filter (AEKF) and adaptive unscented Kalman filter (AUKF). The comparison of the experimental results shows that AEKF and AUKF have better convergence rate, and AUKF has the best accuracy. The comparison from the radial basis function neural network (RBF NN) model also verifies that the ELM model has lighter computation load and smaller estimation error in SOC estimation process. In general, the performance of Li-ion battery SOC estimation is improved by the AUKF algorithm applied on the ELM model.     

Predicting survival of individual patients with esophageal cancer by adaptive neuro-fuzzy inference system approach

Since esophageal cancer has no symptoms in the early stage, it is usually not detected until advanced stages in which treatment is challenging. Integrated treatment provided by a multidisciplinary team is crucial for maximizing the prognosis and survival of patients with esophageal cancer. Currently, clinicians must rely on the cancer staging system for diagnosis and treatment. An accurate and easily applied system for predicting the prognosis of esophageal cancer would be useful for comparing different treatment strategies and for calculating cancer survival probability. This study presents a hazard modeling and survival prediction system based on adaptive neuro-fuzzy inference system (ANFIS) to assist clinicians in prognostic assessment of patients with esophageal cancer and in predicting the survival of individual patients. Expert knowledge was used to construct the fuzzy rule based prognosis inference system for esophageal cancer. Fuzzy logic was used to process the values of input variables rather than categorizing values as normal or abnormal based on cutoffs. After transformation and expansion, censored survival data could be used by the ANFIS for training to establish the risk model for accurately predicting individual survival for different time intervals or for different treatment modalities. Actual values for serum C-reactive protein, albumin, and time intervals were input into the model for use in predicting the survival of individual patients for different time intervals. The curves obtained by the ANFIS approach were fitted to those obtained using the actual values. The comparison results show that the ANFIS is a practical, effective, and accurate method of predicting the survival of esophageal cancer patients.     

PID补偿的完全在线序贯极限学习机控制器在输入扰动系统自适应控制中的应用

针对输入受外界扰动的系统在实现自适应控制难的问题,提出一种比例-积分-微分（PID）补偿的完全在线序贯极限学习机（FOS-ELM）控制器设计方法。首先,建立系统的动态线性模型,采用FOS-ELM算法设计控制器并学习其参数;其次,计算系统的实际输出误差,结合系统的控制误差,设计所需补偿的PID增量参数;最后,对PID补偿的FOS-ELM控制器参数在线调整并用于系统控制。在发动机空气燃油比（AFR）控制系统模型上进行实验,实验结果表明上述方法在实现自适应控制的同时降低了系统扰动输入带来的干扰,提高了系统有效控制率,在正负干扰系数为0.2时,其有效控制率从不足53%提高到93%以上。同时该方法易于实现,具有很强的鲁棒性和实用价值。     

Predicting the occurrence of adverse events using an adaptive neuro-fuzzy inference system (ANFIS) approach with the help of ANFIS input selection

This study presents an adaptive neuro-fuzzy inference system (ANFIS) approach performed to estimate the number of adverse events where the dependent variables are adverse events leading to four types of variables: number of people killed, wounded, hijacked and total number of adverse events. Fourteen infrastructure development projects were selected based on allocated budgets values at different time periods, population density, and previous month adverse event numbers selected as independent variables. Firstly, number of independent variables was reduced by using ANFIS input selection approach. Then, several ANFIS models were performed and investigated for Afghanistan and the whole country divided into seven regions for analysis purposes. Performances of models were assessed and compared based on the mean absolute errors. The difference between observed and estimated value was also calculated within \({\pm }1\) range with values around 90 %. We included multiple linear regression (MLR) model results to assess the predictive power of the ANFIS approach, in comparison to a traditional statistical approach. When the model accuracy was calculated according to the performance metrics, ANFIS showed greater predictive accuracy than MLR analysis, as indicated by experimental results. As a result of this study, we conclude that ANFIS is able to estimate the occurrence of adverse events according to economical infrastructure development project data.     

An easy-to-use MATLAB program (MamLand) for the assessment of landslide susceptibility using a Mamdani fuzzy algorithm

In this study, landslide susceptibility mapping using a completely expert opinion-based approach was applied for the Sinop (northern Turkey) region and its close vicinity. For this purpose, an easy-to-use program, “MamLand,” was developed for the construction of a Mamdani fuzzy inference system and employed in MATLAB. Using this newly developed program, it is possible to construct a landslide susceptibility map based on expert opinion. In this study, seven conditioning parameters characterising topographical, geological, and environmental conditions were included in the FIS. A landslide inventory dataset including 351 landslide locations was obtained for the study area. After completing the data production stage of the study, the data were processed using a soft computing approach, i.e., a Mamdani-type fuzzy inference system. In this system, only landslide conditioning data were assessed, and landslide inventory data were not included in the assessment approach. Thus, a file depicting the landslide susceptibility degrees for the study area was produced using the Mamdani FIS. These degrees were then exported into a GIS environment, and a landslide susceptibility map was produced and assessed in point of statistical interpretation. For this purpose, the obtained landslide susceptibility map and the landslide inventory data were compared, and an area under curve (AUC) obtained from receiver operating characteristics (ROC) assessment was carried out. From this assessment, the AUC value was found to be 0.855, indicating that this landslide susceptibility map, which was produced in a data-independent manner, was successful.     

Stability of iterative learning control with data dropouts via asynchronous dynamical system

In this paper, the stability of iterative learning control with data dropouts is discussed. By the super vector formulation, an iterative learning control (ILC) system with data dropouts can be modeled as an asynchronous dynamical system with rate constraints on events in the iteration domain. The stability condition is provided in the form of linear matrix inequalities (LMIS) depending on the stability of asynchronous dynamical systems. The analysis is supported by simulations.     

使用深度学习和不同频率维度的脑功能性连接对轻微认知障碍的诊断

准确诊断轻微认知障碍（MCI）对于阿尔兹海默症（AD）的预防和治疗十分关键,目前常使用深度学习和静息态功能核磁共振成像（rs-fMRI）对MCI进行辅助诊断。皮尔逊（Pearson）相关法和加窗的皮尔逊（WP）相关法能在时间维度上表示脑功能性连接（FC）,但不能将不同频率维度上的信息进行分解表示。针对这一问题,提出将不同频率维度的FC系数作为现有深度学习的特征输入的方法,以提高MCI分类准确率。首先将被试的数据进行拼接后进行多通道经验模态分解（MEMD）,然后通过切割求得不同频率维度上的FC系数,最后使用VGG16和长短期记忆（LSTM）网络进行测试。实验结果表明,使用所提出的FC系数时,MCI的分类准确率最高可达84.33%,相较使用传统FC系数时的准确率提高了18.33~21.00个百分点;而且不同频率维度的FC系数对MCI有着不同的分辨率。     

An adaptive network-based fuzzy inference system (ANFIS) for the forecasting: The case of close price indices

The close price prediction model of the Zagreb Stock Exchange Crobex® index is presented in this paper. For the input/output data plan modeling the Crobex® index close price historical data are retrieved from the Zagreb Stock Exchange official internet pages. The prediction model is created in the way that for each of 5 days in advance it predicts the Crobex® close price. The prediction model is generated based on the input/output data plan by means of the adaptive neuro-fuzzy inference system method, representing the fuzzy inference system. It is of the essence to point out that for each day a separate fuzzy inference system is created by means of the adaptive neuro-fuzzy inference system method based on the same set of input/output data, the only difference being that for every separate fuzzy inference system different subsets for training and checking are used so that input variables are differently created. The input/output data set represents the historical data of the Crobex® index close price from 4 November 2010 to 24 January 2012 and the Crobex® index close price is predicted for the subsequent 5 days, the first day of prediction being 25 January 2012. After that the above mentioned input/output data set is shifted 5 days in advance and the Crobex® index close price is predicted in advance for the next 5 days starting with the last day of the input/output data set. In that way the Crobex® index close prices are predicted until 19 October 2012 based on the Crobex® index close price historical data. At the end of the paper qualitative and quantitative estimates are presented for the given approach of predicting the Crobex® index close price showing that the approach is useful for predicting within its limits.     

Robust design of iterative learning control for a batch process described by 2D Roesser system with packet dropouts and time‐varying delays

Batch process, working as a best choice for low‐volume and high‐value products in manufacturing, has been widely used in chemical industries. The actuator faults and time delays often occur in practical production. This paper develops an iterative learning control (ILC) design for a batch process described by two‐dimensional (2D) Roesser system with packet dropouts and time‐varying delays. The phenomenon of actuator faults is regarded as an arbitrary stochastic sequence satisfying the Bernoulli random binary distribution. Firstly, the ILC design for a batch process is transformed into stability analysis for a 2D stochastic system with time‐varying delays. Secondly, for analyzing the stability of 2D stochastic systems, we derive the stability condition in terms of linear matrix inequality. Then, we give a procedure to get the control gain for the ILC design. An injection modeling process as an example with simulations in different cases of data dropout is given to demonstrate the validity of the proposed method. Furthermore, the proposed method has a better result by comparing the existing methods.