Article
Multidisciplinary Sciences
Ahmed Youssef Ali Amer
Summary: This study introduces GLocal-LS-SVM, a novel machine learning algorithm that combines localised and global learning to address challenges associated with decentralised data sources, large datasets, and input-space-related issues. The algorithm employs multiple local LS-SVM models to extract informative support vectors from each local region in the input space, which are then merged to train the global model. Experimental results demonstrate that GLocal-LS-SVM achieves comparable or superior classification performance compared to standard LS-SVM and state-of-the-art models, while also outperforming standard LS-SVM in terms of computational efficiency.
Article
Spectroscopy
Masoumeh Valaee, Mahmoud Reza Sohrabi, Fereshteh Motiee
Summary: In this study, two chemometrics methods, PLS and LS-SVM, were used to determine the content of zidovudine (ZDV) and lamivudine (LMV) in synthetic mixtures and anti-HIV pharmaceutical formulation. The results showed that both methods achieved accurate determination of the two components with good recovery rates. The comparison with HPLC as a reference technique demonstrated the reliability of the chemometrics approaches for routine analysis and quality control of the drug.
SPECTROCHIMICA ACTA PART A-MOLECULAR AND BIOMOLECULAR SPECTROSCOPY
(2023)
Article
Computer Science, Artificial Intelligence
Barenya Bikash Hazarika, Deepak Gupta
Summary: This paper introduces a new support vector machine (SVM) model and an improved least squares SVM model to address class imbalance learning (CIL) in binary classification problems. The algorithms assign weights to samples based on their class distributions during training to reduce the effects of CIL.
NEURAL COMPUTING & APPLICATIONS
(2021)
Article
Mathematics, Applied
Sanpeng Zheng, Renzhong Feng
Summary: The variable projection (VP) method is a classical and effective approach for solving the separable nonlinear least squares (SNLLS) problem. While the classical VP method has been applied to one-output radial basis function neural networks (ORBFNN), this study proposes a new VP method for general radial basis function neural networks (GRBFNN) that can have multiple output neurons. The new VP method transforms the SSE minimization problem of GRBFNN into a lower-dimensional optimization problem, and theoretical analysis shows that the stationary points of the lower-dimensional problem are equivalent to those of the original objective function. Numerical experiments demonstrate that minimizing the new objective function leads to faster convergence, smaller training errors, and smaller testing errors compared to minimizing the original objective function.
APPLIED MATHEMATICS AND COMPUTATION
(2023)
Article
Thermodynamics
Tiago de Oliveira Nogueira, Gilderlanio Barbosa Alves Palacio, Fabricio Damasceno Braga, Pedro Paulo Nunes Maia, Elineudo Pinho de Moura, Carla Freitas de Andrade, Paulo Alexandre Costa Rocha
Summary: This study combines DFA with SVM and RBFK methods for imbalance level classification in wind turbines, with RBFK showing excellent performance in different rotation speeds.
Article
Computer Science, Artificial Intelligence
Xiaoxi Zhao, Saiji Fu, Yingjie Tian, Kun Zhao
Summary: The QTLS method proposed in this paper combines QTSELF with LSSVM, imposes different penalties on samples based on their locations, and enhances model robustness. Its generalization capacity is investigated using Rademacher complexity theory, and extensive experiments confirm its effectiveness.
KNOWLEDGE-BASED SYSTEMS
(2022)
Article
Automation & Control Systems
Jiao Zhu, Sugen Chen, Yufei Liu, Cong Hu
Summary: This study proposes a novel energy-based structural least squares twin support vector clustering algorithm (ESLSTWSVC), which improves clustering performance and efficiency by introducing within-class covariance matrix and solving system of linear equations.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2024)
Article
Computer Science, Artificial Intelligence
Shadfar Davoodi, Hung Vo Thanh, David A. Wood, Mohammad Mehrad, Valeriy S. Rukavishnikov
Summary: Emissions of carbon dioxide contribute to global warming. Carbon geological sequestration in saline aquifers is a feasible solution to reduce atmospheric CO2 buildup. Machine-learning models combined with optimizers can accurately predict CO2 trapping indexes and improve simulation efficiency.
APPLIED SOFT COMPUTING
(2023)
Article
Computer Science, Artificial Intelligence
Barenya Bikash Hazarika, Deepak Gupta
Summary: This paper proposed two novel methods for class imbalance learning, which improve the model training efficiency through weighting and least squares principles, and carried out simulations on imbalanced datasets to compare model performance.
NEURAL PROCESSING LETTERS
(2022)
Article
Operations Research & Management Science
Hossein Moosaei, Fatemeh Bazikar, Milan Hladik
Summary: In this work, an efficient parametric nu-support vector regression model with Universum data (UPar-nu-SVR) is proposed, which overcomes some common disadvantages of previous methods. Two approaches are suggested to solve the quadratic programming problem caused by including unlabeled samples. Furthermore, a least squares parametric nu-support vector regression model with Universum data (LS-UPar-nu-SVR) is also proposed to improve the generalization performance.
ANNALS OF OPERATIONS RESEARCH
(2023)
Article
Computer Science, Artificial Intelligence
Chien-Feng Kung, Pei-Yi Hao
Summary: This study proposes a novel formulation called fuzzy hyperplane based least squares support vector machine (FH-LS-SVM) by using fuzzy set theory for LS-SVM. The FH-LS-SVM assigns fuzzy membership degrees to data vectors based on their importance and fuzzifies the parameters for the hyperplane. The proposed method captures the ambiguity in real-world classification tasks and decreases the effect of noise.
NEURAL PROCESSING LETTERS
(2023)
Article
Multidisciplinary Sciences
Abdullah Elen, Selcuk Bas, Cemil Kozkurt
Summary: An adaptive kernel function based on the Gaussian kernel is designed in this study, which performs well in SVM and is compared with traditional linear, polynomial and Gaussian kernels.
ARABIAN JOURNAL FOR SCIENCE AND ENGINEERING
(2022)
Article
Engineering, Industrial
Te Xu, Yongxia Liu, Lixin Tang, Chang Liu
Summary: Kriging interpolation is a widely used spatial interpolation method in data analytics and environmental variable prediction. It provides the best linear unbiased prediction of intermediate values by searching for data distribution regularity and predicting regionalised variable value. This paper proposes an improved Kriging interpolation algorithm using learning kernels based on Estimation of Distribution Algorithms (EDAs) and Least-Squares Support Vector Machine (LSSVM). Experimental results based on real-world environmental variables demonstrate the effectiveness of the proposed algorithm.
INTERNATIONAL JOURNAL OF PRODUCTION RESEARCH
(2022)
Article
Computer Science, Artificial Intelligence
B. Richhariya, M. Tanveer
Summary: Universum based twin support vector machines use prior information about data distribution, leading to better generalization performance. However, in practice, data points may have varying importance, requiring the use of fuzzy membership functions. The proposed fuzzy universum least squares twin support vector machine (FULSTSVM) addresses this issue by providing weights based on membership values for both data samples and universum data, resulting in improved performance compared to existing algorithms.
NEURAL COMPUTING & APPLICATIONS
(2022)
Article
Computer Science, Artificial Intelligence
Chunmei Wang, Huiru Wang, Zhijian Zhou
Summary: In this paper, a novel algorithm called REDINPSVM is proposed to improve the performance of DINPSVM. The algorithm introduces a regularization term to achieve structural risk minimization, applies the k-nearest neighbor method to eliminate some redundant constraints, and uses the least squares technique to accelerate computation. Comprehensive experimental results on various datasets demonstrate the validity of the proposed method.
APPLIED INTELLIGENCE
(2022)
Article
Computer Science, Interdisciplinary Applications
Juncheng Gao, Menad Nait Amar, Mohammad Reza Motahari, Mahdi Hasanipanah, Danial Jahed Armaghani
Summary: This paper introduces two new prediction tools for peak shear strength in rock slopes, utilizing neural networks and meta-heuristic algorithms. The RBFNN-GWO model demonstrated superior accuracy and convergence speed compared to other models, offering efficient support for rock engineers in slope design processes.
ENGINEERING WITH COMPUTERS
(2022)
Article
Computer Science, Interdisciplinary Applications
Rabar H. Faraj, Azad A. Mohammed, Ahmed Mohammed, Khalid M. Omer, Hemn Unis Ahmed
Summary: The study proposed three different models to predict the compressive strength of self-compacting concrete mixtures with or without nano-silica, and found that the multi-logistic model (MLR) outperformed other models for forecasting performance; sensitivity analysis demonstrated that curing time is the most influencing variable for predicting the compressive strength of self-compacting concrete modified with nano-silica.
ENGINEERING WITH COMPUTERS
(2022)
Article
Energy & Fuels
Cuthbert Shang Wui Ng, Ashkan Jahanbani Ghahfarokhi, Menad Nait Amar
Summary: In the petroleum domain, optimizing hydrocarbon production is crucial for economic prospects and meeting global energy demand. This paper demonstrates the development of proxies using a machine learning technique (LSTM) for a 3D reservoir model, and their successful application in production optimization. The proxies show high accuracy and computational efficiency compared to numerical reservoir simulation.
Article
Computer Science, Interdisciplinary Applications
Mahdi Hasanipanah, Mehdi Jamei, Ahmed Salih Mohammed, Menad Nait Amar, Ouaer Hocine, Khaled Mohamed Khedher
Summary: Rock mass deformation modulus (E-m) is a crucial parameter for designing surface and underground rock engineering constructions. In-situ test methods have been proposed to determine the deformability level of jointed rock mass, but they are expensive and time-consuming. This study presents three advanced and efficient machine-learning methods for predicting E-m, with the CFNN-LMA model performing the best among the three models evaluated.
EARTH SCIENCE INFORMATICS
(2022)
Article
Chemistry, Multidisciplinary
Mehdi Jamei, Ahmed Salih Mohammed, Iman Ahmadianfar, Mohanad Muayad Sabri Sabri, Masoud Karbasi, Mahdi Hasanipanah
Summary: This study proposes a linear genetic programming (LGP) model for estimating brittleness index (BI) in deep underground projects and validates it using local weighted linear regression (LWLR) and KStar approaches. The results show that the LGP model outperforms other methods in estimating BI.
APPLIED SCIENCES-BASEL
(2022)
Article
Multidisciplinary Sciences
Chiya Yousef Rahimzadeh, Azeez Abdullah Barzinjy, Ahmed Salih Mohammed, Samir Mustafa Hamad
Summary: The usage of green synthesis method for producing nanoparticles has gained popularity in the scientific community. The method is favored for being environmentally-friendly and utilizing non-toxic materials. This study successfully synthesized nano-silica (SiO2) nanoparticles using Rhus coriaria L. extract and sodium metasilicate under reflux conditions. The green-synthesized SiO2 nanoparticles showed superior stability, enhanced thermal properties, and a larger surface area compared to chemically synthesized ones.
Article
Construction & Building Technology
Nzar Shakr Piro, Ahmed Salih Mohammed, Samir M. Hamad, Rawaz Kurda, Bootan S. Qader
Summary: To preserve the environment and conserve natural resources, steel slag recovery has been used to partially replace fine and coarse aggregate in concrete. This study focused on predicting the compressive strength of concrete with steel slag aggregate replacement, using various models. The results showed that the curing time had the most significant impact on the compressive strength, and the artificial neural network (ANN) model performed the best in predicting the compressive strength.
STRUCTURAL CONCRETE
(2023)
Article
Computer Science, Interdisciplinary Applications
Cuthbert Shang Wui Ng, Menad Nait Amar, Ashkan Jahanbani Ghahfarokhi, Lars Struen Imsland
Summary: Machine Learning has made significant contributions to reservoir engineering, specifically in reservoir simulation. The combination of ML and metaheuristic algorithms shows great potential for developing proxy models in reservoir simulation and optimization studies.
COMPUTERS & CHEMICAL ENGINEERING
(2023)
Article
Materials Science, Textiles
Muhammad Ahsan Gulzar, Babar Ali, Osama Barakat, Marc Azab, Ahmed M. Najemalden, Ahmed Salih Mohammed, Yasser Alashker
Summary: This study aims to develop eco-friendly and ductile concrete by incorporating ground granulated blast furnace slag (GGBS) and jute fiber (JF). Two concrete families were produced with 0% and 25% GGBS as partial replacements for cement, and JF was added at 0%, 0.25%, and 0.5% volume fractions. The effects of a plasticizer on jute fiber reinforced concrete (JFRC) with GGBS were also studied. Various properties including compressive strength, splitting tensile strength, flexural strength, water absorption, chloride ion penetration depth, and electrical resistivity were examined. The results showed that increasing JF content decreased compressive strength but improved splitting tensile strength and flexural strength. The positive effect of JF on compressive strength was observed in mixes with or without a plasticizer. The negative effects of the hydrophilic nature of JF on water absorption and chloride ion penetration resistance were mitigated by using GGBS and controlling workability.
JOURNAL OF NATURAL FIBERS
(2023)
Article
Chemistry, Multidisciplinary
Ahmed Salih Mohammed, Wael Emad, Warzer Sarwar Qadir, Rawaz Kurda, Kawan Ghafor, Raed Kadhim Faris
Summary: This study tested the effect of three water-reducer additives on the workability and compressive strength of concrete. The results showed that adding water-reducer additives can increase the compressive strength of concrete by 8% to 186%, depending on the type of additive and cement content. The study aimed to establish mathematical models to predict the compressive strength of concrete containing water-reducer additives and investigate the impact of mix proportion on compressive strength.
APPLIED SCIENCES-BASEL
(2023)
Article
Chemistry, Physical
Yuzhen Wang, Mahdi Hasanipanah, Ahmad Safuan A. Rashid, Binh Nguyen Le, Dmitrii Vladimirovich Ulrikh
Summary: The accurate estimation of rock strength is crucial for rock-based projects, and this study applied advanced machine learning techniques to predict unconfined compressive strength (UCS) based on non-destructive tests and petrographic studies. The results showed that the extreme gradient boosting tree model outperformed the random forest model in terms of accuracy and error for UCS prediction.
Article
Green & Sustainable Science & Technology
Xiaohua Ding, Mehdi Jamei, Mahdi Hasanipanah, Rini Asnida Abdullah, Binh Nguyen Le
Summary: Using explosive material to fragment rock masses is a common method in surface mines, but it can lead to environmental problems such as flyrock. This study develops hybrid models for predicting flyrock using neural networks and optimization algorithms. The models were tested using data from granite quarry sites in Malaysia, and the results showed that the LSSVM-WOA model was the most accurate in predicting flyrock values.
Article
Green & Sustainable Science & Technology
Yuzhen Wang, Mohammad Rezaei, Rini Asnida Abdullah, Mahdi Hasanipanah
Summary: The intact rock elastic modulus (E) is a key parameter in the design of projects related to rock mechanics and engineering geology. This study aims to evaluate the effectiveness of two meta-heuristic-driven approaches, ANFIS-DE and ANFIS-FA, in predicting E. The ANFIS-FA model outperformed ANFIS-DE, ANFIS, and NN models in terms of predicting E value, based on data collected from the Azad and Bakhtiari dam sites in Iran. Sensitivity analysis showed that P-wave velocity had a larger influence on E compared to other independent variables.
Article
Green & Sustainable Science & Technology
Cumaraswamy Vipulanandan, Ahmed Salih Mohammed, Praveen Ramanathan
Summary: This study analyzed the stress intensity factor (K-I) and bond strength tests of oil well cement (class H) with a water-to-cement ratio (w/c) of 0.38. The mechanical properties of the cement paste were tested and qualified, including compressive and flexural strengths. The relationship between elastic modulus and axial strain was obtained using the Vipulanandan p-q model. The bonding strength between the cement and steel tube representing the casing in the borehole was determined at different curing times.
Article
Green & Sustainable Science & Technology
Ahmad Khalil Mohammed, A. M. T. Hassan, Ahmed Salih Mohammed
Summary: To mitigate the negative environmental effects of cement production, the construction industry is adopting eco-friendly approaches, such as using alternative and recycled materials and reducing carbon emissions in concrete production. This study focuses on investigating the factors influencing the compressive strength of concrete containing ground granulated blast furnace slag (GGBFS) at 28 days of age. Statistical modeling techniques were employed to comprehensively analyze the effects of temperature, water-to-binder ratio, GGBFS-to-binder ratio, fine aggregate, coarse aggregate, and superplasticizer on the compressive strength.