Article
Computer Science, Artificial Intelligence
Maxime Amram, Jack Dunn, Ying Daisy Zhuo
Summary: This research proposes an approach for directly learning optimal tree-based prescription policies from data. It combines methods for counterfactual estimation from the causal inference literature with recent advancements in training globally-optimal decision trees. The resulting optimal policy trees demonstrate excellent performance across various problems, while also being interpretable and scalable.
Article
Biochemical Research Methods
Bonil Koo, Je-Keun Rhee
Summary: The study highlighted the importance of using machine learning methods to predict tumor purity, with the results showing that these models accurately predicted tumor purity and were highly correlated with established standard methods. Additionally, a small group of genes were identified to perform well in predicting tumor purity, mainly involved in the immune system.
BRIEFINGS IN BIOINFORMATICS
(2021)
Article
Computer Science, Information Systems
Robert Burduk, Jedrzej Biedrzycki
Summary: This research proposes a novel ensemble learning algorithm based on feature space partitioning, which selects feature subspaces while considering class label imbalance ratio and defines ensemble class labels. The experimental results show that this method outperforms state-of-the-art ensemble methods on various benchmark datasets.
INFORMATION SCIENCES
(2022)
Article
Management
Rafael Blanquero, Emilio Carrizosa, Cristina Molero-Rio, Dolores Romero Morales
Summary: This paper proposes an optimal regression tree model based on a continuous optimization problem, which aims to strike a balance between prediction accuracy and sparsity. The model can fulfill important properties for regression tasks and provide local explanations due to the smoothness of predictions. The computational experience demonstrates the superiority of this approach in terms of prediction accuracy compared to standard benchmark regression methods, and the scalability with respect to sample size is also illustrated.
EUROPEAN JOURNAL OF OPERATIONAL RESEARCH
(2022)
Article
Computer Science, Theory & Methods
Estevao B. Prado, Rafael A. Moral, Andrew C. Parnell
Summary: Bayesian additive regression trees (BART) is a successful tree-based machine learning method used for regression and classification problems. This paper introduces an extension of BART called model trees BART (MOTR-BART), which considers piecewise linear functions at node levels for prediction, capturing local linearities more efficiently and requiring fewer trees for equal or better performance compared to BART.
STATISTICS AND COMPUTING
(2021)
Article
Computer Science, Artificial Intelligence
Inbal Roshanski, Meir Kalech, Lior Rokach
Summary: Decision forests excel in tabular data with well-tuned hyperparameters, but decision trees may become large and complex in practice, resulting in cumbersome rule collections. FACET is a new algorithm that utilizes automated feature engineering methods to drastically reduce the size of decision trees while maintaining and improving accuracy.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Software Engineering
Jack H. Good, Nicholas Gisolfi, Kyle Miller, Artur Dubrawski
Summary: In recent years, progress has been made in the safety verification of machine learning models such as neural networks and tree ensembles. However, the verification of fuzzy decision trees (FDT) has not been studied. FDT presents unique challenges in verification due to the multiplication of input values. An abstraction-refinement algorithm is proposed for the verification of properties of FDTs, which is shown to be NP-Complete and complete in a finite precision setting. The proposed method outperforms existing solvers and algorithms in terms of speed, as demonstrated in benchmark tests on public datasets. The algorithm code, experiments, and demos are available on GitHub at https://github.com/autonlab/fdt_verification.
IEEE TRANSACTIONS ON SOFTWARE ENGINEERING
(2023)
Article
Computer Science, Information Systems
Marcin Czajkowski, Krzysztof Jurczuk, Marek Kretowski
Summary: By integrating the lasso estimator into the tree induction process, the interpretability of the decision tree can be controlled and its overall performance improved.
INFORMATION SCIENCES
(2023)
Article
Computer Science, Artificial Intelligence
Vinicius G. Costa, Carlos E. Pedreira
Summary: This paper reviews the recent advances in Decision Trees (DTs) research, focusing on issues related to fitting training data, generalization, and interpretability, as well as providing an overview of the field, its key concerns, and future trends.
ARTIFICIAL INTELLIGENCE REVIEW
(2023)
Article
Energy & Fuels
Xin Jin, Shihao Li, Haoran Ye, Jin Wang, Yingji Wu, Daihui Zhang, Hongzhi Ma, Fubao Sun, Arivalagan Pugazhendhi, Changlei Xia
Summary: Biodiesel produced through transesterification is a promising and environmentally friendly fuel derived from biomass resources. However, its production is affected by factors such as feedstock type, reaction time, temperature, and catalyst. Machine learning algorithms, including k-nearest neighbor, Support Vector Machine, Random Forest regression, and AdaBoost regression, have been used to predict biodiesel yield, with Random Forest regression showing the most accurate predictions based on lower RMSE values and higher correlation coefficients.
Article
Chemistry, Analytical
Shawhin Talebi, John Waczak, Bharana A. Fernando, Arjun Sridhar, David J. Lary
Summary: This study aims to outline an objective strategy for discovering optimal EEG bands based on signal power spectra. A two-step data-driven methodology is presented to determine the best EEG bands, using a decision tree and an AIC-inspired quality score. This approach improves the characterization of the underlying power spectrum and isolates key spectral components in dedicated frequency bands.
Article
Computer Science, Hardware & Architecture
Yu-Shan Huang, Jie-Hong R. Jiang
Summary: This paper introduces the importance of circuit learning and proposes a graph learning approach to improve the effectiveness of circuit learning. Experimental results demonstrate the superiority of this approach in terms of accuracy, training time, and circuit size.
IEEE TRANSACTIONS ON COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS
(2023)
Article
Psychiatry
Lucas Boettcher, Josefien J. F. Breedvelt, Fiona C. Warren, Zindel Segal, Willem Kuyken, Claudi L. H. Bockting
Summary: Predicting relapse in depression is crucial for clinical practice. Applying machine-learning methods to individual participant data can improve risk predictions. This study found that decision tree classifiers based on multiple predictors can accurately predict relapse risk and contribute to the development of treatment stratification strategies.
Article
Energy & Fuels
Krishna Kumar Gupta, Kanak Kalita, Ranjan Kumar Ghadai, Manickam Ramachandran, Xiao-Zhi Gao
Summary: This paper explores the application of machine learning methods in the biodiesel production process, utilizing three powerful machine learning algorithms and conducting comprehensive studies. The experiments show that both random forest regression and AdaBoost regression perform well in predictive modeling of biodiesel yield and free fatty acid conversion percentage, with AdaBoost regression potentially being the most suitable approach for biodiesel production modeling.
Article
Mathematics, Interdisciplinary Applications
Sarfaraz Serang, Ross Jacobucci, Gabriela Stegmann, Andreas M. Brandmaier, Demi Culianos, Kevin J. Grimm
Summary: SEM Trees allow the construction of decision trees with structural equation models in each node, creating distinct subgroups based on covariate information. By extending to Mplus, a broader group of researchers can fit a wider range of models efficiently. Mplus Trees algorithm is discussed, along with examples illustrating its practical implementation using publicly available data.
STRUCTURAL EQUATION MODELING-A MULTIDISCIPLINARY JOURNAL
(2021)
Article
Computer Science, Interdisciplinary Applications
Loke Kok Foong, Hossein Moayedi
Summary: This study suggested using equilibrium optimization (EO) and vortex search algorithm (VSA) to optimize a multi-layer perceptron neural network (MLPNN) for predicting the factor of safety of a single-layer soil slope. The results demonstrated the applicability and efficiency of artificial intelligence in this field, as well as the higher generalization ability of the hybrid models. Additionally, the EO-based ensemble outperformed the VSA in optimizing the MLPNN, indicating a larger accuracy.
ENGINEERING WITH COMPUTERS
(2022)
Article
Computer Science, Interdisciplinary Applications
Hossein Moayedi, Soheil Ghareh, Loke Kok Foong
Summary: This study proposes two metaheuristic-integrated predictors using the shuffled complex evolution (SCE) and electromagnetic field optimization (EFO) algorithms synthesized with artificial neural network (ANN) for approximating pan evaporation. The results show that the hybrids outperform the single ANN, with EFO algorithm being faster and more accurate in optimization.
ENGINEERING WITH COMPUTERS
(2022)
Article
Engineering, Multidisciplinary
Yinghao Zhao, Loke Kok Foong
Summary: This paper proposes a reliable predictive tool for the electrical power output of combined cycle power plants using novel soft computing methods. By combining the electrostatic discharge algorithm with an artificial neural network, the proposed hybrid outperforms conventional methods in both training and testing phases.
Article
Engineering, Environmental
Hossein Moayedi, Mohammad Ali Salehi Amin Khasmakhi
Summary: This study proposes two hybrid algorithms combining artificial neural networks (ANN) with spatial analysis of forest fire to improve accuracy. By considering multiple factors such as slope aspect, soil type, and rainfall, the study zoned the susceptibility areas for forest fire and demonstrated the spatial interaction between fire and ignition factors using the frequency ratio model.
STOCHASTIC ENVIRONMENTAL RESEARCH AND RISK ASSESSMENT
(2023)
Article
Engineering, Civil
Hossein Moayedi, Nargess Varamini, Mansour Mosallanezhad, Loke Kok Foong, Binh Nguyen Le
Summary: This study proposes and evaluates four optimized artificial neural networks (ICA, IWO, PSO, and LCA) to determine the bearing capacity of driven piles deployed in cohesionless soils. The IWO-MLP model shows high reliability and can be considered as an innovative model in deep foundation engineering.
TRANSPORTATION GEOTECHNICS
(2022)
Article
Automation & Control Systems
Hossein Moayedi, Atefeh Ahmadi Dehrashid, Mohammad Hossein Gholizadeh
Summary: This study proposes three new hybrid techniques, combining traditional Artificial Neural Network with Stochastic Fractal Search (SFS) and Multi-verse optimizer (MVO), for modeling landslide susceptibility in Kurdistan province, Iran. The results demonstrate the significant effectiveness of these models for optimization.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2023)
Article
Environmental Sciences
Saeid Shabani, Saeid Varamesh, Hossein Moayedi, Bao Le Van
Summary: This study modeled and spatially visualized the susceptibility of a forest stand in northern Iran to snowstorm damage using the random forest (RF) and logistic regression (LR) methods. The RF model outperformed the LR model in both training and validation phases, identifying slope, aspect, and wind effect as the variables with the greatest impacts on forest stand sustainability to snowstorm damage. Approximately 30% of the study area was categorized as highly and very highly susceptible to snowstorms. The results can inform forest managers in developing adaptive forest management plans for snowstorm readiness and recovery.
ENVIRONMENTAL SCIENCE AND POLLUTION RESEARCH
(2023)
Article
Engineering, Environmental
Rana Muhammad Adnan Ikram, Atefeh Ahmadi Dehrashid, Binqiao Zhang, Zhihuan Chen, Binh Nguyen Le, Hossein Moayedi
Summary: A landslide susceptibility map is crucial for minimizing damages caused by landslides. This study introduces a novel approach using the cuckoo optimization algorithm (COA) and the SailFish optimizer (SFO) to develop an artificial neural network (ANN) for landslide forecasting. The hybrid model of COA-MLP showed the best performance in landslide detection and can be useful for planners in identifying dangerous locations.
STOCHASTIC ENVIRONMENTAL RESEARCH AND RISK ASSESSMENT
(2023)
Article
Energy & Fuels
Mohammad Reza Jokar, Saeid Shahmoradi, Adil Hussein Mohammed, Loke Kok Foong, Binh Nguyen Le, Sasan Pirouzi
Summary: This paper presents a hybrid renewable system planning that combines wind turbines, bio-waste energy units, stationary and mobile energy storage. The proposed model prioritizes the use of renewable sources and utilizes storage devices to bridge the gap between demand and renewable generation. It addresses uncertainties by employing robust optimization algorithms and demonstrates low computation time and high robustness. The presence of renewable sources contributes to an environmentally friendly system, and smart charging strategies further reduce the planning cost.
JOURNAL OF ENERGY STORAGE
(2023)
Article
Green & Sustainable Science & Technology
Hossein Moayedi, Hasan Yildizhan, Mohammed Al-Bahrani, Bao Le Van
Summary: The issue of energy efficiency is currently a major concern in global politics. The thermal burden a building faces from outside air is determined by the external environment, particularly the wind speed and outside air temperature. Various factors, including the wall's heat transfer coefficients, coating material, inside and outside temperatures, and external surface temperature, influence the heat load of a building. Through a comprehensive assessment, evaluation, and comparison of two artificial approaches (BSA and COA) used for predicting heat loss in green buildings, the study identifies the optimal method based on R-2 and RMSE criteria. Results show that COA predicts energy loss more accurately, with higher R-2 values and lower RMSE values compared to BSA.
SUSTAINABLE ENERGY TECHNOLOGIES AND ASSESSMENTS
(2023)
Article
Environmental Sciences
Hossein Moayedi, Atefeh Ahmadi Dehrashid
Summary: This research optimized an artificial neural network (ANN) using the backtracking search algorithm (BSA) and the Cuckoo optimization algorithm (COA) to predict landslide susceptibility mapping (LSM). The BSA-ANN model performed better in optimizing the structure and computational parameters of the ANN model compared to the COA-ANN model. The collected landslide susceptibility maps are significant for understanding the level of landslide hazard in the studied area.
ENVIRONMENTAL SCIENCE AND POLLUTION RESEARCH
(2023)
Article
Engineering, Civil
Pan Hu, Zohre Moradi, H. Elhosiny Ali, Loke Kok Foong
Summary: This study evaluates the efficiency of three metaheuristic algorithms for optimizing the performance of a multi-layer perceptron system and finds that these algorithms can significantly improve training and prediction accuracy. The proposed models are capable of providing early, inexpensive, and reliable predictions of the compressive strength of concrete.
SMART STRUCTURES AND SYSTEMS
(2022)
Article
Construction & Building Technology
Hossein Moayedi, Amirali Eghtesada, Mohammad Khajehzadehb, Suraparb Keawsawasvongc, Mohammed M. Al-Amidid, Bao Le Van
Summary: This study employs ER-WCA and EO optimization techniques to predict the compressive strength of concrete, and compares their performance with traditional methods. The results show that both optimizers can improve the accuracy of the model, but EO is more time-effective.
STEEL AND COMPOSITE STRUCTURES
(2022)
Article
Engineering, Civil
Huanlong Hu, Mesut Gor, Hossein Moayedi, Abdolreza Osouli, Loke Kok Foong
Summary: A novel metaheuristic search method called black widow optimization (BWO) was used to improve the accuracy of slope stability analysis by training an artificial neural network (ANN) to predict the factor of safety (FOS) of a single-layer cohesive soil slope. The results showed that the application of BWO significantly enhanced the performance of the ANN, resulting in a decrease in the learning root mean square error and an increase in the correlation between the testing data. The proposed BWO-ANN method shows promise for early prediction of FOS in real-world projects.
SMART STRUCTURES AND SYSTEMS
(2022)
Article
Energy & Fuels
Yingna Du, Chen Huang, Wei Jiang, Qiangwei Yan, Yongfei Li, Gang Chen
Summary: In this study, anionic surfactants modified hydrotalcite was used as a flow improver for crude oil under low-temperature conditions. The modified hydrotalcite showed a significant viscosity reduction effect on crude oil. The mechanism of the modified hydrotalcite on viscosity and pour point of crude oil was explored through characterization and analysis of the modified hydrotalcite and oil samples.
Article
Energy & Fuels
Mohammad Saeid Rostami, Mohammad Mehdi Khodaei
Summary: In this study, a hybrid structure, MIL-53(Al)@MWCNT, was synthesized by combining MIL-53(Al) particles and -COOH functionalized multi-walled carbon nanotube (MWCNT). The hybrid structure was then embedded in a polyethersulfone (PES) polymer matrix to prepare a mixed matrix membrane (MMM) for CO2/CH4 and CO2/N2 separation. The addition of MWCNTs prevented MIL-53(Al) aggregation, improved membrane mechanical properties, and enhanced gas separation efficiency.
Article
Energy & Fuels
Yunlong Li, Desheng Huang, Xiaomeng Dong, Daoyong Yang
Summary: This study develops theoretical and experimental techniques to determine the phase behavior and physical properties of DME/flue gas/water/heavy oil systems. Eight constant composition expansion (CCE) tests are conducted to obtain new experimental data. A thermodynamic model is used to accurately predict saturation pressure and swelling factors, as well as the phase boundaries of N2/heavy oil systems and DME/CO2/heavy oil systems, with high accuracy.
Article
Energy & Fuels
Morteza Afkhamipour, Ebad Seifi, Arash Esmaeili, Mohammad Shamsi, Tohid N. Borhani
Summary: Non-conventional amines are being researched worldwide to overcome the limitations of traditional amines like MEA and MDEA. Adequate process and thermodynamic models are crucial for understanding the applicability and performance of these amines in CO2 absorption, but studies on process modeling for these amines are limited. This study used rate-based modeling and Deshmukh-Mather method to model CO2 absorption by DETA solution in a packed column, validated the model with experimental data, and conducted a sensitivity analysis of mass transfer correlations. The study also compared the CO2 absorption efficiency of DETA solution with an ionic solvent [bmim]-[PF6] and highlighted the importance of finding optimum operational parameters for maximum absorption efficiency.
Article
Energy & Fuels
Arastoo Abdi, Mohamad Awarke, M. Reza Malayeri, Masoud Riazi
Summary: The utilization of smart water in EOR operations has gained attention, but more research is needed to understand the complex mechanisms involved. This study investigated the interfacial tension between smart water and crude oil, considering factors such as salt, pH, asphaltene type, and aged smart water. The results revealed that the hydration of ions in smart water plays a key role in its efficacy, with acidic and basic asphaltene acting as intrinsic surfactants. The pH also influenced the interfacial tension, and the aged smart water's interaction with crude oil depended on asphaltene type, salt, and salinity.
Article
Energy & Fuels
Dongao Zhu, Kun Zhu, Lixian Xu, Haiyan Huang, Jing He, Wenshuai Zhu, Huaming Li, Wei Jiang
Summary: In this study, cobalt-based metal-organic frameworks (Co-based MOFs) were used as supports and co-catalysts to confine the NHPI catalyst, solving the leaching issue. The NHPI@Co-MOF with carboxyl groups exhibited stronger acidity and facilitated the generation of active oxygen radicals O2•, resulting in enhanced catalytic activity. This research provides valuable insights into the selection of suitable organic linkers and broadens the research horizon of MOF hybrids in efficient oxidative desulfurization (ODS) applications.
Article
Energy & Fuels
Edwin G. Hoyos, Gloria Amo-Duodu, U. Gulsum Kiral, Laura Vargas-Estrada, Raquel Lebrero, Rail Munoz
Summary: This study investigated the impact of carbon-coated zero-valent nanoparticle concentration on photosynthetic biogas upgrading. The addition of nanoparticles significantly increased microalgae productivity and enhanced nitrogen and phosphorus assimilation. The presence of nanoparticles also improved the quality of biomethane produced.
Article
Energy & Fuels
Yao Xiao, Asma Leghari, Linfeng Liu, Fangchao Yu, Ming Gao, Lu Ding, Yu Yang, Xueli Chen, Xiaoyu Yan, Fuchen Wang
Summary: Iron is added as a flocculant in wastewater treatment and the hydrothermal carbonization (HTC) of sludge produces wastewater containing Fe. This study investigates the effect of aqueous phase (AP) recycling on hydrochar properties, iron evolution and environmental assessment during HTC of sludge. The results show that AP recycling process improves the dewatering performance of hydrochar and facilitates the recovery of Fe from the liquid phase.
Article
Energy & Fuels
He Liang, Tao Wang, Zhenmin Luo, Jianliang Yu, Weizhai Yi, Fangming Cheng, Jingyu Zhao, Xingqing Yan, Jun Deng, Jihao Shi
Summary: This study investigated the influence of inhibitors (carbon dioxide, nitrogen, and heptafluoropropane) on the lower flammability limit of hydrogen and determined the critical inhibitory concentration needed for complete suppression. The impact of inhibitors on explosive characteristics was evaluated, and the inhibitory mechanism was analyzed with chemical kinetics. The results showed that with the increase of inhibitor quantity, the lower flammability limit of hydrogen also increased. The research findings can contribute to the safe utilization of hydrogen energy.
Article
Energy & Fuels
Zonghui Liu, Zhongze Zhang, Yali Zhou, Ziling Wang, Mingyang Du, Zhe Wen, Bing Yan, Qingxiang Ma, Na Liu, Bing Xue
Summary: In this study, high-performance solid catalysts based on phosphotungstic acid (HPW) supported on Zr-SBA-15 were synthesized and evaluated for the one-pot conversion of furfural (FUR) to γ-valerolactone (GVL). The catalysts were characterized using various techniques, and the ratio of HPW and Zr was found to significantly affect the selectivity of GVL. The HPW/Zr-SBA-15 (2-4-15) catalyst exhibited the highest GVL yield (83%) under optimized reaction conditions, and it was determined that a balance between Bronsted acid sites (BAS) and Lewis acid sites (LAS) was crucial for achieving higher catalytic performance. The reaction parameters and catalyst stability were also investigated.
Article
Energy & Fuels
Michael Stoehr, Stephan Ruoff, Bastian Rauch, Wolfgang Meier, Patrick Le Clercq
Summary: As part of the global energy transition, an experimental study was conducted to understand the effects of different fuel properties on droplet vaporization for various conventional and alternative fuels. The study utilized a flow channel to measure the evolution of droplet diameters over time and distance. The results revealed the temperature-dependent effects of physical properties, such as boiling point, liquid density, and enthalpy of vaporization, and showed the complex interactions of preferential vaporization and temperature-dependent influences of physical properties for multi-component fuels.
Article
Energy & Fuels
Yuan Zhuang, Ruikang Wu, Xinyan Wang, Rui Zhai, Changyong Gao
Summary: Through experimental validation and optimization of the chemical kinetic model, it was found that methanol can accelerate the oxidation reaction of ammonia, and methanol can be rapidly oxidized at high concentration. HO2 was found to generate a significant amount of OH radicals, facilitating the oxidation of methanol and ammonia. Rating: 7.5/10.
Article
Energy & Fuels
Radwan M. EL-Zohairy, Ahmed S. Attia, A. S. Huzayyin, Ahmed I. EL-Seesy
Summary: This paper presents a lab-scale experimental study on the impact of diethyl ether (DEE) as an additive to waste cooking oil biodiesel with Jet A-1 on combustion and emission features of a swirl-stabilized premixed flame. The addition of DEE to biodiesel significantly affects the flame temperature distribution and emissions. The W20D20 blend of DEE, biodiesel, and Jet A-1 shows similar flame temperature distribution to Jet A-1 and significantly reduces UHC, CO, and NOx emissions compared to Jet A-1.
Article
Energy & Fuels
Jiang Bian, Ziyuan Zhao, Yang Liu, Ran Cheng, Xuerui Zang, Xuewen Cao
Summary: This study presents a novel method for ammonia separation using supersonic flow and develops a mathematical model to investigate the condensation phenomenon. The results demonstrate that the L-P nucleation model accurately characterizes the nucleation process of ammonia at low temperatures. Numerical simulations also show that increasing pressure and concentration can enhance ammonia condensation efficiency.
Article
Energy & Fuels
Shiyuan Pan, Xiaodan Shi, Beibei Dong, Jan Skvaril, Haoran Zhang, Yongtu Liang, Hailong Li
Summary: Integrating CO2 capture with biomass-fired combined heat and power (bio-CHP) plants is a promising method for achieving negative emissions. This study develops a reliable data-driven model based on the Transformer architecture to predict the flowrate and CO2 concentration of flue gas in real time. The model validation shows high prediction accuracy, and the potential impact of meteorological parameters on model accuracy is assessed. The results demonstrate that the Transformer model outperforms other models and using near-infrared spectral data as input features improves the prediction accuracy.