Article
Computer Science, Artificial Intelligence
Linjing Wang, Tianlan Mo, Xuetao Wang, Wentao Chen, Qiang He, Xin Li, Shuxu Zhang, Ruimeng Yang, Jialiang Wu, Xuejun Gu, Jun Wei, Peiliang Xie, Linghong Zhou, Xin Zhen
Summary: This study introduced a two-level framework, HF2HM, to integrate diversified classification models by feeding heterogeneous classifiers with homogeneous random-projected training datasets. Results showed the superiority of the proposed HF2HM framework over base classifiers and state-of-the-art benchmark ensemble methods, indicating its potential as a tool for medical decision making in practical clinical settings.
KNOWLEDGE-BASED SYSTEMS
(2021)
Article
Chemistry, Multidisciplinary
Uyeol Park, Yunho Kang, Haneul Lee, Seokheon Yun
Summary: Accurately estimating the cost of a construction project in the early stages is crucial for successful completion. When there is insufficient information to predict construction costs, a machine learning model using past data can be used as an alternative. This study proposes a two-level stacking heterogeneous ensemble algorithm that combines RF, SVM, and CatBoosting. Bayesian optimization with cross-validation is employed to determine the optimal hyperparameter values of the base learners during training. Cost information data provided by the Public Procurement Service in South Korea is used to evaluate ML algorithms and the proposed stacking-based ensemble model. The analysis results show that the two-level stacking ensemble model outperforms the individual ensemble models.
APPLIED SCIENCES-BASEL
(2022)
Article
Computer Science, Artificial Intelligence
Shaker El-Sappagh, Farman Ali, Tamer Abuhmed, Jaiteg Singh, Jose M. Alonso
Summary: This study proposes a novel ensemble learning framework for predicting Alzheimer's disease progression. The framework incorporates heterogeneous base learners into an integrated model and utilizes multimodal time-series data. The proposed model achieves outstanding performance in predicting disease progression and can be implemented in low-cost healthcare environments. The balance between accuracy and diversity is found to be critical in selecting ensemble members. The proposed framework holds promise for efficient information fusion ensembles in medical and non-medical problems.
Article
Computer Science, Information Systems
Esra'a Alshdaifat, Malak Al-hassan, Ahmad Aloqaily
Summary: This paper presents an alternative approach to selecting base classifiers for forming a parallel heterogeneous ensemble by trimming poorly performing classifiers, resulting in a more effective ensemble. Experimental analysis showed that the proposed approach outperformed other state-of-the-art methods in terms of effectiveness and superiority.
Article
Computer Science, Information Systems
Christian Natale Gencarelli, Debora Voltolina, Mohammed Hammouti, Marco Zazzeri, Simone Sterlacchini
Summary: The collaborative open-source IT infrastructure optimizes geological field data collection, integration, validation, and sharing, ensuring operational continuity and leveraging both online and offline field data collection tools for system resilience.
ISPRS INTERNATIONAL JOURNAL OF GEO-INFORMATION
(2022)
Article
Computer Science, Artificial Intelligence
Xiaoning Shen, Di Xu, Liyan Song, Yuchi Zhang
Summary: This paper proposes a heterogeneous multi-project multi-task allocation model based on the group collaboration mode to address the problem of projects consisting of heterogeneous tasks in the Mobile CrowdSensing (MCS) platform. The model distinguishes the roles of group members and incorporates their skills and social competence. The proposed multi-objective fireworks algorithm with dual-feedback ensemble learning framework is used to solve the model and experimental results show its effectiveness compared to other algorithms.
APPLIED SOFT COMPUTING
(2023)
Article
Transportation Science & Technology
Hao Wu, David Levinson
Summary: Ensemble forecasting is a modeling approach that combines data sources, models of different types, and alternative assumptions to utilize all available information in predictions. It aims to improve forecast accuracy and robustness in various fields, but is currently lacking in integration and adoption across disciplines.
TRANSPORTATION RESEARCH PART C-EMERGING TECHNOLOGIES
(2021)
Article
Clinical Neurology
Koichiro Shiba, Adel Daoud, Shiho Kino, Daisuke Nishi, Katsunori Kondo, Ichiro Kawachi
Summary: Understanding the differential mental health effects of traumatic experiences is crucial for identifying vulnerable subpopulations. This study found considerable heterogeneity in the association between disaster-related traumatic experiences and subsequent mental health problems. Some subgroups experienced severe impacts, with the most vulnerable group tending to be from lower socioeconomic backgrounds with preexisting depressive symptoms.
PSYCHIATRY AND CLINICAL NEUROSCIENCES
(2022)
Article
Chemistry, Multidisciplinary
Abed Alanazi, Abdu Gumaei
Summary: This paper proposes a decision fusion ensemble learning (DFEL) model based on a voting technique for detecting malicious websites. By combining the predictions of gradient boosting, extreme gradient boosting, and random forest classifiers, the accuracy of malicious website detection is improved.
APPLIED SCIENCES-BASEL
(2023)
Article
Computer Science, Artificial Intelligence
HaoJie Chen, Guofu Ding, Shengfeng Qin, Jian Zhang
Summary: The study introduced a hyper-heuristic collaborative scheduling approach for project scheduling with random activity durations, proposing a HH-EGP method to address stochastic resource constrained project scheduling problem (SRCPSP). Experimental results demonstrated the advantage of HH-EGP over traditional heuristics and meta-heuristics in solving SRCPSP.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Computer Science, Information Systems
Parnika Bhat, Sunny Behal, Kamlesh Dutta
Summary: This paper proposes a precise dynamic analysis approach to identify a variety of malicious attacks. The proposed method focuses on behavioral analysis of malware and uses features such as system calls, binders, and complex Android objects. By employing feature selection and stacking machine learning algorithms, efficient malware detection and classification with an accuracy rate of 98.08% is achieved.
COMPUTERS & SECURITY
(2023)
Article
Economics
Yufei Xia, Junhao Zhao, Lingyun He, Yinguo Li, Xiaoli Yang
Summary: Peer-to-peer (P2P) lending is an emerging field in FinTech, facing significant credit risk. Special data and macroeconomic variables are found to be powerful predictors of loss given default (LGD) in P2P lending. Utilizing a novel HSE approach can outperform other models in most cases, achieving optimal average ranks.
INTERNATIONAL JOURNAL OF FORECASTING
(2021)
Article
Engineering, Electrical & Electronic
Zhenyi Chen, Yushan Gao, Yanyang Zi, Mingquan Zhang, Chen Li, Zhongmin Xiao
Summary: This article proposes a heterogeneous time-tracking fusion algorithm to monitor the health state of large mechanical equipment based on multisensor information. The algorithm obtains time-domain indexes and instantaneous frequencies of fast-varying harmonic-like signals and applies a dynamic time-tracking function to fuse different varying-rate signals into a dynamic normalized time-varying index representing the health state.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2023)
Article
Automation & Control Systems
Franck Djeumou, Zhe Xu, Murat Cubuktepe, Ufuk Topcu
Summary: We develop a probabilistic control algorithm, GTLProCo, for decentralized control of swarms of agents with heterogeneous dynamics and objectives. GTLProCo significantly improves scalability over existing algorithms and utilizes a recently proposed language called graph temporal logic to express high-level task specifications for the swarm. It controls the density distribution of the swarm in a decentralized and probabilistic manner, synthesizing a time-varying Markov chain under GTL constraints. We demonstrate the effectiveness of GTLProCo in various swarm scenarios and present an efficient scheme for solving the associated optimization problem.
IEEE TRANSACTIONS ON AUTOMATIC CONTROL
(2023)
Article
Computer Science, Interdisciplinary Applications
Jen-Yin Yeh, Chi-Hua Chen
Summary: This study predicts the success of crowdfunding projects by analyzing social media activity, human capital of funders, and online project presentation. It proposes a neural network method based on ensemble machine learning to prevent overfitting. The study shows that the ensemble neural network method achieves the highest accuracy for prediction. It also provides practical implications for project founders and investors by identifying influential features and offering a model to predict crowdfunding success.
JOURNAL OF ENTERPRISE INFORMATION MANAGEMENT
(2022)
Article
Geochemistry & Geophysics
Jose Luis Villaescusa-Nadal, Eric Vermote, Belen Franch, Andres E. Santamaria-Artigas, Jean-Claude Roger, Sergii Skakun
Summary: The goal of this study is to develop accurate and consistent surface reflectance and albedo products for analyzing global albedo trends. Distinguishing between cloud and snow is challenging due to the limitations of the sensor used. To address this issue, the researchers propose an algorithm based on spectral analysis to identify clear land and snow pixels using satellite and reanalysis data.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Environmental Sciences
Sergii Skakun, Jan Wevers, Carsten Brockmann, Georgia Doxani, Matej Aleksandrov, Matej Batic, David Frantz, Ferran Gascon, Luis Gomez-Chova, Olivier Hagolle, Dan Lopez-Puigdollers, Jerome Louis, Matic Lubej, Gonzalo Mateo-Garcia, Julien Osman, Devis Peressutti, Bringfried Pflug, Jernej Puc, Rudolf Richter, Jean-Claude Roger, Pat Scaramuzza, Eric Vermote, Nejc Vesel, Anze Zupanc, Lojze Zust
Summary: This paper summarizes the results of the first Cloud Masking Intercomparison eXercise (CMIX) conducted by the Committee Earth Observation Satellites (CEOS) Working Group on Calibration & Validation (WGCV). Ten cloud detection algorithms developed by different organizations were evaluated and showed good agreement in detecting thick clouds but had higher uncertainties in detecting thin/semi-transparent clouds.
REMOTE SENSING OF ENVIRONMENT
(2022)
Article
Environmental Sciences
Xiaojuan Huang, Yangyang Fu, Jingjing Wang, Jie Dong, Yi Zheng, Baihong Pan, Sergii Skakun, Wenping Yuan
Summary: This study utilized synthetic aperture radar and satellite image data, combined with a time-weighted dynamic time warping method, to successfully generate a winter cereal map of Europe with a spatial resolution of 30 meters. Validation against agricultural census data showed high accuracy of the method. Additionally, the method can identify the distribution of winter cereals two months before harvest, providing timely monitoring and identification of crop growth at a continental level.
Article
Geography, Physical
Belen Franch, Juanma Cintas, Inbal Becker-Reshef, Maria Jose Sanchez-Torres, Javier Roger, Sergii Skakun, Jose Antonio Sobrino, Kristof Van Tricht, Jeroen Degerickx, Sven Gilliams, Benjamin Koetz, Zoltan Szantoi, Alyssa Whitcraft
Summary: Crop calendars are important for monitoring crop development, and existing global products only provide information at national or subnational level. This study presents gridded maps for wheat and maize crop calendars, capturing their spatial variability. The maps are generated using global products and evaluated using a Random Forest algorithm.
GISCIENCE & REMOTE SENSING
(2022)
Article
Geography, Physical
Yiming Zhang, Sergii Skakun, Michael Oluwatosin Adegbenro, Qing Ying
Summary: Worldwide economic development and population growth have led to significant changes in urban land use. This study utilizes a deep learning model trained on a benchmark dataset to map and quantify urban land use changes in the Washington DC-Baltimore area. The results show that a substantial portion of urban land experienced changes, particularly in the construction of commercial and residential buildings.
INTERNATIONAL JOURNAL OF DIGITAL EARTH
(2022)
Article
Geography, Physical
Victor Hugo Rohden Prudente, Sergii Skakun, Lucas Volochen Oldoni, Haron A. M. Xaud, Maristela R. Xaud, Marcos Adami, Ieda Del ' Arco Sanches
Summary: Remote sensing plays a crucial role in the mapping of Land Use and Land Cover (LULC) worldwide, particularly in areas with frequent cloud cover. Combining optical and SAR data improves the accuracy of classification. This study investigates the incorporation of SAR data into the classification process using optical data in the mapping of LULC in the Roraima State, Brazil. The results show that a combination of multi-temporal surface reflectance, vegetation index, backscatter coefficient, and polarization data yields the most accurate LULC mapping.
ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING
(2022)
Article
Environmental Sciences
Jaemin Eun, Sergii Skakun
Summary: The ongoing military conflict in Eastern Ukraine has led to significant changes in land use and economic activities, particularly in agriculture and industry. By analyzing the changes in nighttime light activity, the study evaluates and reflects the socio-economic impacts of human conflicts. The results show a nearly 50% decrease in nighttime light activity in the Donetsk and Luhansk regions from 2012 to 2016. Furthermore, the study finds that the sensitivity to nighttime light losses varies between areas inside and outside cities, and there are noticeable differences in losses attributed to industrial land-use types.
ENVIRONMENTAL RESEARCH LETTERS
(2022)
Article
Environmental Sciences
Christian Abys, Sergii Skakun, Inbal Becker-Reshef
Summary: The Russian wheat industry has experienced significant growth in the past two decades, becoming one of the top global wheat exporters. However, the volatility in wheat production continues to have an impact on the global commodities market and has lasting implications for land use and land cover change.
ENVIRONMENTAL RESEARCH COMMUNICATIONS
(2022)
Article
Information Science & Library Science
Valery Lukinskiy, Vladislav Lukinskiy, Dmitry Ivanov, Boris Sokolov, Darya Bazhina
Summary: This paper proposes a method to assess supplier reliability and order fulfilment performance using discrete distribution analysis, taking into account the impact of parameter probabilities and the number of parameters on the value of a perfect order. The proposed method can be applied in practice to evaluate the reliability of supply and order fulfilment processes, and to evaluate the effectiveness of various operational strategies.
INTERNATIONAL JOURNAL OF INFORMATION MANAGEMENT
(2023)
Article
Environmental Sciences
Georgia Doxani, Eric F. Vermote, Jean-Claude Roger, Sergii Skakun, Ferran Gascon, Alan Collison, Liesbeth De Keukelaere, Camille Desjardins, David Frantz, Olivier Hagolle, Minsu Kim, Jerome Louis, Fabio Pacifici, Bringfried Pflug, Herve Poilve, Didier Ramon, Rudolf Richter, Feng Yin
Summary: The correction of atmospheric effects is crucial for remote sensing applications. A benchmark exercise called ACIX was conducted to assess the performance of different atmospheric correction methods. The results showed that the processors were successful in retrieving aerosol optical depth and water vapor retrievals, but there was still a need for better assessment of uncertainties in surface reflectance retrievals.
REMOTE SENSING OF ENVIRONMENT
(2023)
Article
Environmental Sciences
Erik C. Duncan, Sergii Skakun, Ankit Kariryaa, Alexander Prishchepov
Summary: Unexploded munitions have devastating effects on various aspects, and it is important to detect and map them accurately. This study applies a deep learning approach to identify and map artillery craters in agricultural fields in Eastern Ukraine. The model shows high accuracy in detecting craters, and the reliability improves with larger crater sizes. The estimated crater map reveals a significant number of craters in the region, providing valuable information for demining and assessing the impact of warfare on agriculture and the environment.
SCIENCE OF REMOTE SENSING
(2023)
Article
Environmental Sciences
Leonid Shumilo, Sergii Skakun, Meredith L. Gore, Andrii Shelestov, Nataliia Kussul, George Hurtt, Dmytro Karabchuk, Volodymyr Yarotskiy
Summary: The ongoing Russian-Ukrainian War has significant impacts on the protected areas of the Emerald Network. However, despite the conflict, the implementation of Bern Convention policies and forest management practices in the Luhansk region have helped maintain reforestation rates in Ukrainian-controlled territories.
COMMUNICATIONS EARTH & ENVIRONMENT
(2023)
Proceedings Paper
Geosciences, Multidisciplinary
E. Vermote, J. McCorkel, W. H. Rountree, A. Santamaria-Artigas, S. Skakun, B. Franch, J. C. Roger
Summary: This study validates the surface reflectance from Sentinel-2 produced by LaSRC using an automated camera system (CAMSIS). The results show good agreement between the surface reflectance and NDVI computed from CAMSIS calibrated data and the observations from Sentinel-2, indicating a good performance of LaSRC atmospheric correction.
2022 IEEE INTERNATIONAL GEOSCIENCE AND REMOTE SENSING SYMPOSIUM (IGARSS 2022)
(2022)
Proceedings Paper
Automation & Control Systems
Dmitry Ivanov, Alexandre Dolgui, Boris Sokolov, Marina Ivanova
Summary: Optimal control is a convenient approach for developing supply chain process optimization models and describing process dynamics. While there has been significant research on integrating optimization and simulation methods for supply chain management, the integration of models and algorithms has received less attention. This paper addresses this issue by discussing the challenges and providing implementation guidance for integrating optimization and simulation models and algorithms. The developed theoretical framework is exemplified through an integrated optimization-simulation model of supply chain design and planning with consideration of disruption risks.
Proceedings Paper
Geography, Physical
Mehdi Hosseini, Inbal Becker-Reshef, Ritvik Sahajpal, Pedro Lafluf, Guillermo Leale, Estefania Puricelli, Sergii Skakun, Heather McNairn
Summary: This study evaluated the potential of using Sentinel-1 dual-polarimetric data to forecast soybean yields at a field scale in central Argentina. By extracting polarimetric features and training an Artificial Neural Network (ANN) model, along with an innovative iterative retrieval method, the accuracy of soybean yield prediction was improved.
XXIV ISPRS CONGRESS: IMAGING TODAY, FORESEEING TOMORROW, COMMISSION III
(2022)