Article
Computer Science, Information Systems
Abdullah Mohammed Rashid, Habshah Midi, Waleed Dhhan Slwabi, Jayanthi Arasan
Summary: This paper introduces a new robust iteratively reweighted SIMPLS based on nu-Support Vector Regression, referred to as SVR-RWSIMPLS. It proves to be more efficient, robust, and faster in computational running times compared to the traditional RWSIMPLS, especially when multiple leverage points and vertical outliers are present. The proposed diagnostic plot is also successful in accurately classifying observations into different groups.
Article
Chemistry, Applied
Kiran Raj Bukkarapu, Anand Krishnasamy
Summary: This study developed SVR models based on FTIR spectra of biodiesel and biodiesel-diesel blends to predict important engine fuel properties. The models showed good performance in predicting the blend proportion, viscosity, cetane number, and calorific value. Compared with other regression models, SVR was found to be the most suitable approach.
FUEL PROCESSING TECHNOLOGY
(2022)
Article
Computer Science, Artificial Intelligence
Chen Wang, Guohua Peng, Wei Lin
Summary: The proposed RLML-LSR method aims to enhance the robustness and performance of scene recognition through constructing a local discriminative metric function and utilizing least square regression regularization. Extensive experiments on various datasets demonstrate the effectiveness and robustness of the method.
Article
Materials Science, Paper & Wood
Wan Sieng Yeo, Woei Jye Lau
Summary: The study developed a multi-output LSSVR (MLSSVR) model using bleaching process variables and results from two different case studies to predict the WI of cotton. The predictive accuracy of the MLSSVR model was measured by RMSE, MAE, and R-2, showing that it outperformed other regression models in predicting WI with significantly lower RMSE and MAE values, and highest R-2 values up to 0.9999.
Article
Neurosciences
Yuanhao Li, Badong Chen, Gang Wang, Natsue Yoshimura, Yasuharu Koike
Summary: This study proposes a new robust variant for PLSR, called PMCR, which optimizes the regression coefficients using the maximum correntropy criterion and calculates robust projectors using half-quadratic optimization. Experimental results demonstrate that PMCR outperforms existing methods in noisy, inter-correlated, and high-dimensional decoding tasks, improving the decoding robustness for brain-computer interfaces.
FRONTIERS IN NEUROSCIENCE
(2023)
Article
Management
Saiji Fu, Yingjie Tian, Long Tang
Summary: This paper proposes a general framework that can smoothly and adaptively bound any non-negative function, and introduces a robust regression model called BLSSVR. BLSSVR mitigates the effects of noise and outliers by limiting the maximum loss. With appropriate parameters, the bounded least squares loss grows faster than its unbounded form in the initial stage, which enables BLSSVR to assign larger weights to non-outlier points. The Nesterov accelerated gradient algorithm is employed to optimize BLSSVR. Extensive experiments demonstrate the superiority of BLSSVR over benchmark methods.
EUROPEAN JOURNAL OF OPERATIONAL RESEARCH
(2023)
Article
Geochemistry & Geophysics
Bin Wang, Jie Yu, Yu Chen, Zhisheng Zhao, Chao Liu
Summary: A robust CTLS algorithm for universal 3-D symmetric transformation was proposed in this study, which can efficiently deal with various 3-D symmetric transformation problems influenced by outliers. By introducing the equivalent weight principle of robust estimation, the algorithm's universality and efficiency were enhanced.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Instruments & Instrumentation
Ahmad Asghari, Alireza Adl, Peyman Ghajarbeygi, Sina Darzi
Summary: The study proposed the use of FTIR spectroscopy combined with SVR as an efficient method for precise determination of Benzalkonium chloride in aqueous samples, with SVR model having higher prediction capability and lower error compared to PLS model.
INFRARED PHYSICS & TECHNOLOGY
(2021)
Article
Computer Science, Artificial Intelligence
Eufrasio de A. Lima Neto, Paulo C. Rodrigues
Summary: SVD is a widely used algorithm for dimensionality reduction and principal component analysis, but it is not suitable for data contaminated with outlying observations. To overcome this limitation, a kernel robust SVD algorithm is proposed, which operates in the original space and applies a robust linear regression framework to obtain robust estimates for the singular values and singular vectors. Simulation results show that the proposed algorithm outperforms classical and robust SVD algorithms. The merits of the proposed algorithm are also illustrated in an image recovery application.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Anton Bjorklund, Andreas Henelius, Emilia Oikarinen, Kimmo Kallonen, Kai Puolamaki
Summary: This paper introduces a robust regression method that can effectively handle datasets with outliers. By approximating the data using a sparse linear model, better robustness to outliers can be achieved. An efficient approximation algorithm called SLISE is presented to solve the NP-hard problem, and the method is demonstrated on both synthetic and real-world regression problems.
DATA MINING AND KNOWLEDGE DISCOVERY
(2022)
Article
Computer Science, Interdisciplinary Applications
Jale Tezcan, Claudia C. Marin-Artieda
Summary: In this paper, a computationally simple and efficient formulation is proposed to convert acceleration signal to displacement signal. The previously derived approach based on Least Square Support Vector Machine is reformulated as a convolution operation between the acceleration signal and an explicitly defined windowing function. The new formulation inherits the benefits of the original method while significantly reducing the computational effort.
ADVANCES IN ENGINEERING SOFTWARE
(2023)
Article
Chemistry, Analytical
Min-Jee Kim, Hye-In Lee, Jae-Hyun Choi, Kyoung Jae Lim, Changyeun Mo
Summary: This study investigated the spectral characteristics of topsoil from four major rivers in the Republic of Korea and developed a machine learning-based model to predict soil organic matter (SOM) content using spectroscopic techniques. The results identified the important wavelength of SOM in topsoil and confirmed the predictability of SOM content, which could be used for the construction of a national topsoil database.
Article
Pharmacology & Pharmacy
Yu-Wen Wu, Giang Huong Ta, Yi-Chieh Lung, Ching-Feng Weng, Max K. Leong
Summary: Topical and transdermal drug delivery is a safe and effective route of administration. Skin permeability is a critical parameter in drug discovery and development. This study used two QSAR models to predict skin permeability and uncover the permeation mechanism, and found that the synergy between interpretable PLS and predictive HSVR models is useful for drug discovery and development.
Article
Environmental Sciences
Daqing Liu, Chenglian Feng, Yu Qiao, Jindong Wang, Yingchen Bai, Fengchang Wu
Summary: This study quantified the structure-toxicity relationships of OPEs using a partial least square (PLS) regression model, and predicted the acute toxicity of OPEs without data based on the regression results. The results indicated that the influence of physicochemical properties on OPE toxicity was not significant, and acute toxicity was mainly influenced by autocorrelation coefficients. According to the prediction results, CDP may pose a high risk while halogenated alkyl-substituted OPEs may be less toxic. These findings could provide valuable insights for environmental management and risk assessment of emerging chemicals like OPEs.
SCIENCE OF THE TOTAL ENVIRONMENT
(2023)
Article
Chemistry, Analytical
Geonwoo Kim, Hoonsoo Lee, Insuck Baek, Byoung-Kwan Cho, Moon S. Kim
Summary: A SWIR HSI system and optimized model were developed for rapid detection of BPO particles in wheat flour, showing high potential in discriminating BPO particles efficiently and allowing for quantitative evaluation.
SENSORS AND ACTUATORS B-CHEMICAL
(2022)
Article
Automation & Control Systems
Haifei Peng, Jian Long, Cheng Huang, Shibo Wei, Zhencheng Ye
Summary: This paper proposes a novel multi-modal hybrid modeling strategy (GMVAE-STA) that can effectively extract deep multi-modal representations and complex spatial and temporal relationships, and applies it to industrial process prediction.
CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS
(2024)