Learning to rank refers to machine learning techniques for training a model in a ranking task. Learning to rank has been shown to be useful in many applications of information retrieval, natural language processing, and data mining. Learning to rank can be described by two systems: a learning system and a ranking system. The learning system takes training data as input and constructs a ranking model. The ranking system then makes use of the learned ranking model for ranking prediction. In this paper, a new learning algorithm based on ensemble learning for learning ranking models in information retrieval is proposed. This algorithm iteratively constructs weak learners using a fraction of the training data whose weight distribution is determined based on previous weak learners. The proposed algorithm combines the weak rankers to achieve the final ranking model. This algorithm constructs a ranking model on a fraction of the training data to increase the accuracy and reduce the learning time. Experimental results based on Letor.3 benchmark dataset shows that the proposed algorithm significantly outperforms other ensemble learning algorithms.

Shear wave velocity is one of the important parameters for determining mechanical and petrophysical properties in hydrocarbon reservoirs. Shear waves are usually obtained from dipole sonic imager (DSI) tools or core analysis in the laboratory. However, these methods as common sources for shear wave estimation are time-consuming and costly and thus can only provide information on shear wave in a few drilled wells. To overcome these limitations, different artificial intelligence methods are used to estimate the mentioned parameter through the conventional well logs. In this study, the shear wave velocity was estimated using ensemble learning methods in Asmari Reservoir in the Mansouri oilfield. In this study, shear wave velocity was estimated using ensemble learning methods such as voting, stacking, bagging, and boosting in the Asmari reservoir, and the results were compared with conventional models such as Linear regression (LR), support vector regression (SVR), nearest neighbor algorithm (KNN), decision tree (DT), neural network (ANN) and hybrid methods such as combining neural network with genetic algorithm (ANN-GA) ), particle swarm (ANN-PSO) and fuzzy systems (ANFIS). In order to evaluate and validate the models, correlation coefficient (R2) and root mean square error (RMSE) were used. A comparison of conventional models and hybrid methods with ensemble learning methods showed that ensemble algorithms perform better in shear wave estimation. Considering the testing phase, among the ensemble learning methods, the Catboost model has provided the lowest error (RMSE=0.058) and highest correlation coefficient (R=0.983) can determine shear wave with high accuracy.

Almost all MODIS images have cloudy regions. Cloud detection and discriminating it from similar objects like snow/ice is a necessary pre-processing step before extracting accurate information. Features for cloud detection can be divided into two categories; textural and spectral features. Using textural features in the visible bands make cloud and snow/ice pixels separable, while cloud and snow can have similar temperature. Thermal features have been used for cloud detection in different heights in MODIS cloud mask (MOD35). Recently variety of studies using ensemble learning methods for land cover classification have been done. In the studies, ensemble learning methods used for classification but in this study, a new application has been introduced; for fusion of classifiers having reflectance and thermal features. Also, in previous studies, the effect of changing feature selection method on the performance of ensemble learning methods has not been examined. So, the purpose of this study is the comparison of the performance of fusion of the reflectance and thermal classifiers using two kinds of ensemble learning methods including boosting and Random Forest (RF) for detection of cloud, cirrus and snow/ice pixels, based on the feature selection method applied. First, some of the Visible-Infrared bands (VIR), 1, 2, 8 and 26 in addition to the thermal bands including 20, 22, 31, 32 and 35 of Terra MODIS were calibrated and the reflectance and Brightness Temperature (BT) values were extracted. Also, three indexes, Normalized Difference Vegetation Index (NDVI), Normalized Difference Snow Index (NDSI) and the ratio of bands b1/b2 were computed. Gray Level Co-occurrence Matrix (GLCM) textural features of the above mentioned bands and some of the BT differences including, BT31-BT32, BT33-BT31, BT31-BT20, BT20-BT32, BT22-BT32 were added to the reflectance and thermal input features sets. For selecting the suitable reflectance and thermal features in different kinds of boosting methods including Adaboost.M1, AdaboostSVM, Logitboost and Totalboost, S criteria and Genetic Algorithm (GA) were used, and in RF algorithm in addition to these methods, Recursive Feature Elimination (RFE) and correlation matrix were applied too. After feature selection step, training data were selected manually for cloud, cirrus and snow/ice and fed into the reflectance and thermal classifiers. Classifiers were fused in decision level using the majority vote method. The performance of different cases was compared using producer accuracy, user accuracy and kappa coefficient indexes. For almost all ensemble methods, ignoring which one of the feature selection methods applied, a high cloud and cirrus producer accuracy achieved. RFE and correlation matrix methods in RF algorithm, resulted in 99% and 100% cloud producer accuracy. These values are higher than S criteria and GA. Boosting methods, ignoring the kind of feature selection method, got higher snow/ice producer accuracy than RF algorithm by assigning higher weights to the class that has less training data among classes. Also, boosting methods resulted in higher cirrus user accuracy than RF. Among feature selection methods applied in RF, correlation matrix achieved 91% value for cirrus user accuracy. Finally, agreement to the reference map produced from MOD35 calculated. RF algorithms showed higher agreement with the reference map in comparison to the boosting methods. Highest agreement resulted from RF-RFE with the value of 76% and lowest agreement from Logitboost-GA with the value of 42%.

The purpose of this research is to use the ensemble learning model to combine the predictions of random forest models, short-term long memory and recurrent neural network to provide an algorithmic trading system based on its. In this research, a prediction model based on ensemble machine learning model is presented and its performance is compared with each of the sub-algorithms and real data. In this research, in the first stage, using three machine learning models, the price top and bottom of Bitcoin have been predicted. In the second stage, the outputs of the models are presented as feature variables to the XGboost and LightGBM models to predict the roof and floors. Then, in the third stage, the outputs of the second stage are combined with the collective voting classification pattern to predict the next ceiling and floor. Bitcoin price top and bottom data in the 1-hour time frame from 1/1/2018 to the end of 6/30/2022 have been used as target variables and 31 technical analysis indicators as feature variables for three models in the first stage. Finally, forecast values and algorithmic trading systems were evaluated and compared with real data for 3 models and the introduced ensemble learning model. The obtained results show the improvement of the precision and accuracy of the proposed collective learning model in predicting the top and bottom of Bitcoin, as well as its better performance than the sub-algorithms.

The volume and speed of transactions in financial markets has increased significantly and has undergone extensive changes nowadays. Facing with increasing, decreasing or fluctuating trends in the stock market, determining the right trading strategy is very important. Therefore, complex meta-heuristic models are used for choosing a suitable strategy. In this research, an attempt is made to develop a new method of selecting and optimizing the stock portfolio based on the ensemble learning algorithm and genetics in order to select the best trading strategy to achieve greater returns and less risk. A combination of a six-class support vector machine (SVM) algorithm is used to predict returns and receive a buying signal; besides, a dynamic genetic algorithm is used to optimize trading rules. In this study, collective learning methods including Bagging, one of the algorithms based on Ensemble Learning, have been used to improve the accuracy of classification of returns. Data related to each share and fundamental variables in a daily time interval between years 1390 to 1399 is used as training and test data. The obtained results, comparing to traditional methods, are promising.

In this article, using the market timing approach and homogeneous and inhomogeneous collective learning methods, the purchase, maintenance and sales signal and market forecast are presented based on the basic characteristics, technical characteristics and time series of returns of each company in the 100 days leading to the current day. . Based on this, 208 companies were selected as active companies between 1390 and 1399 To teach data by two-level ensemble learning machine (HHEL) and market trend forecasting based on market timing strategy, use data from 5 years 1390 to 1394 and to test the data as stock portfolio optimization based on stock portfolio maximization and risk minimization. The investment portfolio uses MOPSO and NSGA II algorithms and is compared with the obtained investment portfolio with the buy and hold strategy. The results showed that the MOPSO algorithm achieved the highest stock portfolio yield with 96.437% compared to the NSGA II algorithm with a yield of 91.157% and the same investment method with a yield of 13.058%. Also, the portfolio risk in NSGA II algorithm was much lower than the portfolio risk in MOPSO algorithm with 0.792% and 1.367%, respectively

This study aims to provide a learning organization SEM and the application of done at Tehran University. The research method is descriptive-correlational and analytical method was SEM. The population was University of Tehran staffs on the number of 2730 people. Sample size by using Cochran sampling formula and simple random sampling, were selected 243 people. To collect data, a questionnaire based on the model of Peter Senge (1990) designed by reliability (94/0 = α). Also evaluated and approved content and construct validity. For data analysis, Kolmogorov-Smirnov structural equation modeling, and one-sample t-test, was used. The results showed that the learning organization model has been fitted to the data. Accordingly, the individual features of the paradigm (γ =0/24) and systematic approach (γ =0/52) been effective. Systematic approach to the common cause (β=0/51) and collective learning (β=0/44) and shared vision have also had an impact on collective learning (β=0/17) has had impact. Finally, mental models based on shared vision and collective learning as well as individual features common cause whatsoever. The results showed that the application of learning organization and all aspects of its application in the target population, below the average and only mental patterns was middle.

Spatiotemporal estimation and monitoring of soil moisture based on remote sensing observations (optical and thermal) is challenging due to its physical nature in high vegetation conditions, necessitating improving and increasing the accuracy of soil moisture estimation in these areas. Therefore, this research aimed to develop a new approach to estimating surface soil moisture in agricultural fields with dense vegetation using machine learning algorithms by incorporating optical and thermal remote sensing data and soil physical properties. For this objective, 16 Landsat-8 satellite images and more than 430 control locations were used during the sugarcane crop’s growth period in 2018-2019 at the Hakim Farabi Sugarcane Agro-Industrial company in the Khuzestan province of Iran. A set of 10 scenarios of various unique combinations of the available input variables were developed and then evaluated by five machine learning algorithms, including multiple linear regression (MLR), decision tree-based algorithms (CART and M5P), and ensemble learning-based algorithms (gradient-boosted regression trees (GBRT) and random forest regression (RFR)). According to the results, the highest correlation between input variables and surface soil moisture was observed in Soil Wetness Index (SWI) and Normalized Soil Moisture Index (NSMI) with R values of 0.79 and 0.69, respectively. Also, the highest accuracy of machine learning algorithms based on R2, RMSE, and MAE results was obtained in GBRT (0.99, 0.011, and 0.006) and RFR (0.99, 0.014, and 0.007), respectively. In general, the findings of this research show the importance of using variables based on Landsat-8 remote sensing data in combination with ensemble learning algorithms that can be independent of any ground measurements.

The healthcare insurance industry faces a significant challenge predicting individuals' insurance costs, which are based on complex parameters such as age and physical characteristics. Insurance companies categorize policyholders into high-risk and low-risk groups to manage risks and avoid potential losses. However, the accurate estimation of costs for each individual can be a daunting task. By leveraging data science and machine learning techniques, insurance companies can improve their cost estimation accuracy and better manage risks. This approach can help insurance companies to provide more accurate insurance coverage and pricing for individuals leading to higher customer satisfaction and lower financial losses. To address this challenge, a data science and machine learning-based approach that uses ensemble learning to predict high-risk and low-risk individuals is used. The method involves several steps including data preprocessing, feature engineering, and cross-validation to evaluate the model's performance. The first step involves preprocessing the data by cleaning it, handling missing values, and encoding categorical variables. The second step generates new features using feature engineering techniques such as scaling, normalization, and dimensionality reduction. Next, ensemble learning is used to combine multiple regression methods such as logistic regression, neural networks, support vector machines, random forests, LightGBM, and XGBoost. By combining these methods, the aim is to leverage their strengths and minimize their weaknesses to achieve better prediction accuracy. Finally, the model's performance is evaluated using cross-validation techniques such as k-fold cross-validation. These techniques help to validate the model's accuracy and prevent overfitting. The proposed approach achieves an AUC of 0.73 demonstrating its effectiveness in predicting high-risk and low-risk individuals. In conclusion, the healthcare insurance industry can benefit greatly from data science and machine learning-based approaches. By accurately predicting high-risk and low-risk individuals, insurance companies can better manage risks and provide more accurate coverage and pricing for their customers. This can lead to the improvement of  customer satisfaction and the reduction of financial losses for insurance companies.

Precipitation is one of the most important components of hydrology and meteorology, and the prediction of its values are important in various fields, such as agriculture and the environment. Considering that the occurrence of precipitation and its amount depend on many factors, the modeling of precipitation has many complications. In this research, the meteorological data of Tabriz synoptic station including minimum, maximum and average temperature, relative humidity, air pressure, maximum wind speed and precipitation in the period of 1986-2020 were used. The machine learning and ensemble learning methods including Multi Layer Perceptron (MLP), Random Forest (RF), Ada Boost (AB), Gradient Boost (GB), Extra Trees (ET) and Decision Tree Regression (DTR) models were used for rainfall modeling. 70% of the data was used for training and 30% for testing the models. The statistical criteria of Coefficient of Correlation (R), Root Mean Square Error (RMSE), Mean Absolute Error (MAE) and Kling-Gupta Efficiency (KGE) were used to evaluate the models. According to the results, on daily scale the MLP model with R=0.993, RMSE=0.184 mm, MAE=0.184 mm and KGE=0.82, and the ET model with R=0.986, RMSE=0.324 mm, MAE=0.324 mm and KGE=0.75, respectively, and on monthly scale the MLP model with R=0.999, RMSE=0.153 mm, MAE=0.222 mm KGE=0.88, and the ET model with R=0.981, RMSE=0.266 mm, MAE=0.197 mm and KGE=0.71, respectively, have the highest accuracy. Overall, the results show that machine learning and ensemble learning models perform well in predicting daily and monthly rainfall.