Article
Computer Science, Artificial Intelligence
Giang Ngo, Rodney Beard, Rohitash Chandra
Summary: In this paper, an evolutionary bagged ensemble learning method is proposed, which enhances the diversity of bags using evolutionary algorithms. The experimental results show that this method outperforms traditional ensemble learning methods on various benchmark datasets.
Article
Automation & Control Systems
Jianyuan Sun, Hui Yu, Guoqiang Zhong, Junyu Dong, Shu Zhang, Hongchuan Yu
Summary: In this article, a new random forests algorithm called random Shapley forests (RSFs) is proposed, which uses the Shapley value to evaluate the importance of each feature. The experiments conducted on benchmark and real-world datasets demonstrate that RSFs outperform or are at least comparable to existing consistent RFs, original RFs, and support vector machines.
IEEE TRANSACTIONS ON CYBERNETICS
(2022)
Article
Business, Finance
Lu Wang, Rui Wu, WeiChun Ma, Weiju Xu
Summary: Based on the relationship between the global soybean market and weather, this study aims to fill the gap in soybean volatility forecasting under weather information. By using extended GARCH-MIDAS approaches and adding weather variables, we find that models incorporating bagging-based weather information outperform those with raw weather indicators or without weather information. Our conclusions are robust to further tests, and our novel bagging-related GARCH-MIDAS-W-MBB model provides fresh insights into soybean volatility forecasting.
INTERNATIONAL REVIEW OF FINANCIAL ANALYSIS
(2023)
Article
Statistics & Probability
Wei Peng, Tim Coleman, Lucas Mentch
Summary: Random forests are a popular off-the-shelf supervised learning algorithm, and this research establishes convergence rates for random forests and other supervised learning ensembles, providing a quantitative measure for the speed of convergence.
ELECTRONIC JOURNAL OF STATISTICS
(2022)
Article
Health Care Sciences & Services
Dong-Hwa Jeong, Se-Eun Kim, Woo-Hyeok Choi, Seong-Ho Ahn
Summary: This study aims to classify physical activities in daily life using machine learning methods. By extracting features and applying sampling methods, the data imbalance issue was successfully addressed. The results showed that methods like random forest and adaptive boosting performed well in PA classification.
Article
Soil Science
Kamal Nabiollahi, Ruhollah Taghizadeh-Mehrjardi, Aram Shahabi, Brandon Heung, Alireza Amirian-Chakan, Masoud Davari, Thomas Scholten
Summary: In a study conducted in Kurdistan Province, Iran, a combination of random forests and covariate data was used to assess the spatial variability of salinity and sodicity in agricultural salt-affected land. The results showed that optimization algorithms helped improve the accuracy of predictions, and identified groundwater table, categorical maps, salinity index, and multi-resolution ridge top flatness as important covariates for predicting soil salinity and sodicity.
Article
Computer Science, Interdisciplinary Applications
Xiaotie Chen, David L. Woodruff
Summary: This software utilizes sampled data to obtain a consistent sample-average solution and estimate confidence intervals for the optimality gap using bootstrap and bagging, without the need for considering the underlying distribution of the samples.
INFORMS JOURNAL ON COMPUTING
(2023)
Article
Engineering, Multidisciplinary
Manuel Zumbado-Corrales, Juan Esquivel-Rodriguez
Summary: Electron Microscopy Maps are crucial for studying bio-molecular structures, describing envelopes of proteins within cells. Segmentation and Evolutionary-Optimized Segmentation algorithms are used to improve the identification of protein regions, aiding in drug design and functional understanding.
Article
Computer Science, Artificial Intelligence
Victor Acena, Isaac Martin de Diego, Ruben R. Fernandez, Javier M. Moguerza
Summary: This study introduces a new ensemble framework called MOE, which effectively combines stable and unstable machine learning algorithms in constructing predictive models. By using resampling techniques and weighted random bootstrap sampling, the framework constructs slightly overfitted base learners, thereby improving the predictive ability.
KNOWLEDGE-BASED SYSTEMS
(2022)
Article
Geography, Physical
Binh Thai Pham, Abolfazl Jaafari, Trung Nguyen-Thoi, Tran Van Phong, Huu Duy Nguyen, Neelima Satyam, Md Masroor, Sufia Rehman, Haroon Sajjad, Mehebub Sahana, Hiep Van Le, Indra Prakash
Summary: This study developed highly accurate ensemble machine learning models for spatial prediction of rainfall-induced landslides in the Uttarkashi district, India. The D-REPT model was identified as the most accurate, providing insights for engineers and modelers to develop more advanced predictive models.
INTERNATIONAL JOURNAL OF DIGITAL EARTH
(2021)
Article
Automation & Control Systems
Victor Henrique Alves Ribeiro, Roberto Santana, Gilberto Reynoso-Meza
Summary: This paper proposes two novel machine learning algorithms to improve the automatic target recognition system for unmanned aerial vehicles. These models make use of the stochastic procedure of Random Forests and employ the novel Random Vector Functional Link Tree or Extreme Learning Tree for decision split. Experimental results show that the proposed algorithms outperform other state-of-the-art ensemble learning techniques in terms of predictive performance and computational complexity.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2023)
Article
Engineering, Environmental
Xudong Hu, Cheng Huang, Hongbo Mei, Han Zhang
Summary: A novel machine learning ensemble model, BRSNBtree, was proposed to predict landslide susceptibility in Zigui County of the Three Gorges Reservoir Area. The results showed that the distance to rivers was the most important factor in predicting landslide susceptibility, and BRSNBtree outperformed other methods in terms of prediction performance.
BULLETIN OF ENGINEERING GEOLOGY AND THE ENVIRONMENT
(2021)
Article
Plant Sciences
Guangman Song, Quan Wang, Jia Jin
Summary: Understanding the uncertainty in parameterization of Vcmax and Jmax is crucial for predicting carbon fluxes. Recent studies have shown that the relationship between Vcmax and Jmax varies depending on species and leaf traits. Our analysis in cool-temperate forest stands in Japan revealed that leaf traits, particularly LMA, significantly influence the regression, leading to improved model predictions.
PLANT PHYSIOLOGY AND BIOCHEMISTRY
(2021)
Article
Computer Science, Artificial Intelligence
Arthur Hoarau, Arnaud Martin, Jean-Christophe Dubois, Yolande Le Gall
Summary: This paper proposes an Evidential Decision Tree and an Evidential Random Forest, which can handle uncertain and imprecise predictions and can predict rich labels. Experimental results showed better performance for the presented methods compared to other evidential models and recent Cautious Random Forests in handling noisy data and effectively uncertainly and imprecisely labeled datasets. The proposed models also offer better robustness and the ability to predict rich labels, which can be used in other approaches such as active learning.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Information Systems
A. Dmitry Devyatkin, G. Oleg Grigoriev
Summary: This paper proposes an algorithm for training kernel decision trees and random forests, which overcomes the limitations of traditional methods in dealing with multidimensional sparse data. Experimental results show that the proposed algorithm outperforms other methods in various tasks, and the selected regularization technique helps reduce overfitting.