Objective: In the present era, businesses have developed to a large extent which has, in turn, forced them to manage their resources and expenditures wisely for the sake of competition. This is mainly because the competitive market has severely reduced the flexibility of companies, which means that their ability respond to different economic situations has reduced and this puts most firms at the constant risk of bankruptcy and contraction. Therefore, in this study, we have tried to predict the bankruptcy of manufacturing companies through preventing the occurrence of such risks. Methods: In this study, the "Kernel Extreme Learning Machine" has been used as one of the artificial intelligence models for predicting bankruptcy. Given that Machine Learning methods require an optimization algorithm we have used one of the most up-to-date, "Gray Wolf Algorithm" which has been introduced in 2014. Results: The above model has been implemented on the 136 samples that were collected from the Tehran Stock Exchange between 2015 and 2018. All of the performance evaluation criteria including the classification, accuracy, type error, second-order error and area under the ROC curve showed better performance than the genetic algorithm which was presented and its significance was confirmed by t-test. Conclusion: Considering the gray wolf algorithm’ s high accuracy and its performance compared to the genetic algorithm, it is necessary to use the gray wolf algorithm to predict the bankruptcy of Iranian manufacturing companies either for investment purposes and for validation purposes, or for using internal management of the company.

With a largeamount of multimedia content in the web, storage and retrieval of them by classical Learning methods dealt with some major challenges like memory restriction. These limitations in some of Learning algorithms like SVM and ANN is so serious that these algorithms cannot be employed in large-scale Learning context. Kernel Extreme Learning Machine ((KELM)) algorithm is one of the powerful methods in Machine Learning. Learning phase of this method is based on constructing Kernel matrix of labeled instances and calculating inverse of it. So, employing this method in large scale Learning context with a lot of labeled instances is not feasible. In this research to overcome limitation of employing the (KELM) in large-scale multi-label Learning, a new approach is proposed. The proposed approach is based on prototype selection in neighborhood of each training instance. By using the proposed approach, the size of training set is reduced. So, classical Learning methods can be applied on reduced training set. Since multimedia contents are basically multi-label, the proposed prototype selection approach is based on multi-label domains like automatic image annotation. Experimental results on NUS-WIDE large-scale multi-label image set and three other versions include Object, Scene and Lite indicated the effectiveness of the proposed approach in solving the limitation of employing (KELM) method in large-scale multi-label Learning.

The financial time series data is a highly nonlinear signal and hence difficult to predict precisely. The prediction accuracy can be improved by linearizing the signal. In this paper the nonlinear data sample is linearized by decomposing it into several IMFs. A hybrid multi-layer decomposition technique is developed. The decomposition proposed in this paper is the combination of both EMD and VMD methods. As a new contribution to the previous literature in this study the VMD is used to further decompose the higher frequency signals obtained from the EMD based decomposed signal. In the result analysis it is observed that the double decomposition improves the prediction accuracy. This is a new introduction in the field of stock market prediction. The prediction accuracy of the proposed model is performed by applying it to three different stock markets for predicting the closing price. Historical data (closing price) is implemented to obtain 1 day ahead predicted closing price. Comparative analysis of different previously implemented methods like BPNN, SVM, ANN and ELM, along with the proposed method is performed. GA is implemented for optimizing the Kernel factors. It is observed that the proposed hybrid model outperformed the other methods.

In this study, the groundwater level of the Kabodarahang aquifer located in Hamadan Province, Iran, is simulated using MODFLOW, Extreme Learning Machine (ELM), and Wavelet-Extreme Learning Machine (WA-ELM) Models. The correlation coefficient and scatter index values for the MODFLOW model are calculated 0. 917 and 0. 0004, respectively. Then, by different input combination and using the stepwise selection, 10 different models are introduced for the ELM and WA-ELM models with different lags. By evaluating all activation functions of the ELM model, the sigmoid activation function predicts groundwater level values with more accuracy. Also, Daubechies2 is selected as the mother wavelet of the WA-ELM models. According to different numerical models results, the WA-ELM model is selected as the superior model in prediction of groundwater level. For the superior model, the correlation coefficient and Nash-Sutcliffe efficiency coefficient are calculated 0. 959 and 0. 915, respectively. These values for ELM model was respectively computed as 0. 828 and 0. 672.

Accurate electricity price forecasting gives a capability to make better decisions in the electricity market environment when this market is complicated due to severe fl uctuations. The main purpose of a prediction model is to forecast future prices. For doing this, the predicted variable (as output) and historical data (as input) should be close to each other. Machine Learning is known as one of the most successful ways of forecasting time series. Extreme Learning Machine (ELM) is a feed-forward neural network with one hidden layer. Hence, in this paper, an Extreme Learning Machine has been used for predicting electricity prices in a medium-term time horizon. The real data of New York City electricity market has been utilized to simulate and predict the electricity price in four seasons of the year. Finally, the fi ndings are compared with multi-layer perceptron (MLP) results, which prove the effi ciency of the model.

Interval data are usually applied where inaccuracy and variability must be considered. This paper presents a Learning method for Interval Extreme Learning Machine (IELM) in classification. IELM has two steps similar to well known ELM. At first weights connecting the input and the hidden layers are generated randomly and in the second step, ELM uses the Moore–Penrose generalized inverse to determine the weights connecting the hidden and output layers. In order to use Moore–Penrose generalized inverse for determining second layer weights in IELM, this paper proposes four classification methods to handle symbolic interval data based on ELM. The first one uses a midpoint of intervals for each feature value then it applies a classic ELM. The second one considers each feature value as a pair of quantitative features and implements a conjoint for classic Extreme Learning Machine. The third one represents interval features by their vertices and performs a classic Extreme Learning Machine as well. The fourth one takes each interval as a pair of quantitative features after that two separated classic Extreme Learning Machines are performed on these features and combines the results accordingly. Algorithms are tested on the synthetic and real datasets. A synthetic dataset is applied to determine the number of hidden layer nodes in an IELM. The classification error rate is considered as a comparison criterion. The error rate obtained for each proposed methods is 19.1667%, 15% , 6.5358% and 18.3333% respectively. Experiments demonstrate the usefulness of these classifiers to classify symbolic interval data.

Streaming data refers to data that is continuously generated in the form of fast streams with high volumes. This kind of data often runs into evolving environments where a change may affect the data distribution. Because of a wide range of real-world applications of data streams, performance improvement of streaming analytics has become a hot topic for researchers. The proposed method integrates online ensemble Learning into Extreme Machine Learning to improve the data stream classification performance. The proposed incremental method does not need to access the samples of previous blocks. Also, regarding the AdaBoost approach, it can react to concept drift by the component weighting mechanism and component update mechanism. The proposed method can adapt to the changes, and its performance is leveraged to retain high-accurate classifiers. The experiments have been done on benchmark datasets. The proposed method can achieve 0. 90% average specificity, 0. 69% average sensitivity, and 0. 87% average accuracy, indicating its superiority compared to two competing methods.

In this study, the discharge coe cient of rectangular and circular side ori ces was estimated using the Extreme Learning Machine method. Furthermore, in this study for evaluating the ability of di erent ELM models the Monte Carlo simulations are used. The Monte Carlo simulation is a comprehensive classi cation of computational algorithms which uses the random sampling procedure for calculating numerical results. The main idea of this method is based on solving problems which might be actual in nature using random decision-making processes. The Monte Carlo methods are usually used for simulating physical and mathematical systems not solvable with other methods. The Monte Carlo simulation is generally used by probability distribution to solve various problems such as numerical optimization and numerical integration. The k-fold cross validation method is also used for examining the performance of the above models. In this method, the main sample is randomly divided into k sub-samples with the same size. Among k sub-samples, a sub-sample is used as the validation data and the remaining as the test data of the model. Then, the validation process repeats k times (equal to the number of layers) and each of k sub-samples is used exactly once as validation data. In this study, the experimental values obtained by Hussein et al. (2010) and Hussein et al. (2011) are used for validating the results of the numerical models. Their experimental model consisted of a rectangular channel with the length, the width and the height of 9. 15m, 0. 5m and 0. 6m, respectively. They installed the circular and rectangular ori ces at a distance of 5m from the inlet of the main channel on the side wall. In the next stage, the most optimized number of hidden neurons was chosen equal to 30 and the results of all activation functions of the Extreme Learning Machine were examined and the sigmoid activation function is selected for simulating the discharge coe cient. Subsequently, two modeling combinations were introduced using the input parameters as well as  ve di erent Extreme Learning Machine models were developed. The analysis of the modeling results showed that the model with the shape coe cient has more accuracy. The superior model is a function of all input parameters and reasonably estimates values of the discharge coe cient. For example, the values of R and MAPE for this model are estimated 0. 990 and 0. 223, respectively. The results of the superior model were also compared with the empirical equations and it was shown that this model has more accuracy. Also, the partial derivative sensitivity analysis (PDSA) was run for all input parameters.