for Journals by Title or ISSN
for Articles by Keywords
help
  Subjects -> COMPUTER SCIENCE (Total: 2050 journals)
    - ANIMATION AND SIMULATION (30 journals)
    - ARTIFICIAL INTELLIGENCE (99 journals)
    - AUTOMATION AND ROBOTICS (100 journals)
    - CLOUD COMPUTING AND NETWORKS (64 journals)
    - COMPUTER ARCHITECTURE (9 journals)
    - COMPUTER ENGINEERING (10 journals)
    - COMPUTER GAMES (16 journals)
    - COMPUTER PROGRAMMING (27 journals)
    - COMPUTER SCIENCE (1196 journals)
    - COMPUTER SECURITY (46 journals)
    - DATA BASE MANAGEMENT (14 journals)
    - DATA MINING (32 journals)
    - E-BUSINESS (22 journals)
    - E-LEARNING (28 journals)
    - ELECTRONIC DATA PROCESSING (22 journals)
    - IMAGE AND VIDEO PROCESSING (39 journals)
    - INFORMATION SYSTEMS (109 journals)
    - INTERNET (94 journals)
    - SOCIAL WEB (51 journals)
    - SOFTWARE (34 journals)
    - THEORY OF COMPUTING (8 journals)

COMPUTER SCIENCE (1196 journals)                  1 2 3 4 5 6 | Last

Showing 1 - 200 of 872 Journals sorted alphabetically
3D Printing and Additive Manufacturing     Full-text available via subscription   (Followers: 20)
Abakós     Open Access   (Followers: 4)
ACM Computing Surveys     Hybrid Journal   (Followers: 22)
ACM Journal on Computing and Cultural Heritage     Hybrid Journal   (Followers: 8)
ACM Journal on Emerging Technologies in Computing Systems     Hybrid Journal   (Followers: 11)
ACM Transactions on Accessible Computing (TACCESS)     Hybrid Journal   (Followers: 3)
ACM Transactions on Algorithms (TALG)     Hybrid Journal   (Followers: 15)
ACM Transactions on Applied Perception (TAP)     Hybrid Journal   (Followers: 5)
ACM Transactions on Architecture and Code Optimization (TACO)     Hybrid Journal   (Followers: 9)
ACM Transactions on Autonomous and Adaptive Systems (TAAS)     Hybrid Journal   (Followers: 7)
ACM Transactions on Computation Theory (TOCT)     Hybrid Journal   (Followers: 12)
ACM Transactions on Computational Logic (TOCL)     Hybrid Journal   (Followers: 3)
ACM Transactions on Computer Systems (TOCS)     Hybrid Journal   (Followers: 17)
ACM Transactions on Computer-Human Interaction     Hybrid Journal   (Followers: 14)
ACM Transactions on Computing Education (TOCE)     Hybrid Journal   (Followers: 5)
ACM Transactions on Design Automation of Electronic Systems (TODAES)     Hybrid Journal   (Followers: 3)
ACM Transactions on Economics and Computation     Hybrid Journal  
ACM Transactions on Embedded Computing Systems (TECS)     Hybrid Journal   (Followers: 3)
ACM Transactions on Information Systems (TOIS)     Hybrid Journal   (Followers: 19)
ACM Transactions on Intelligent Systems and Technology (TIST)     Hybrid Journal   (Followers: 7)
ACM Transactions on Interactive Intelligent Systems (TiiS)     Hybrid Journal   (Followers: 3)
ACM Transactions on Multimedia Computing, Communications, and Applications (TOMCCAP)     Hybrid Journal   (Followers: 9)
ACM Transactions on Reconfigurable Technology and Systems (TRETS)     Hybrid Journal   (Followers: 6)
ACM Transactions on Sensor Networks (TOSN)     Hybrid Journal   (Followers: 7)
ACM Transactions on Speech and Language Processing (TSLP)     Hybrid Journal   (Followers: 8)
ACM Transactions on Storage     Hybrid Journal  
ACS Applied Materials & Interfaces     Full-text available via subscription   (Followers: 27)
Acta Automatica Sinica     Full-text available via subscription   (Followers: 2)
Acta Universitatis Cibiniensis. Technical Series     Open Access  
Ad Hoc Networks     Hybrid Journal   (Followers: 11)
Adaptive Behavior     Hybrid Journal   (Followers: 11)
Advanced Engineering Materials     Hybrid Journal   (Followers: 28)
Advanced Science Letters     Full-text available via subscription   (Followers: 9)
Advances in Adaptive Data Analysis     Hybrid Journal   (Followers: 7)
Advances in Artificial Intelligence     Open Access   (Followers: 15)
Advances in Calculus of Variations     Hybrid Journal   (Followers: 2)
Advances in Catalysis     Full-text available via subscription   (Followers: 5)
Advances in Computational Mathematics     Hybrid Journal   (Followers: 18)
Advances in Computer Science : an International Journal     Open Access   (Followers: 15)
Advances in Computing     Open Access   (Followers: 2)
Advances in Data Analysis and Classification     Hybrid Journal   (Followers: 51)
Advances in Engineering Software     Hybrid Journal   (Followers: 27)
Advances in Geosciences (ADGEO)     Open Access   (Followers: 13)
Advances in Human Factors/Ergonomics     Full-text available via subscription   (Followers: 23)
Advances in Human-Computer Interaction     Open Access   (Followers: 19)
Advances in Materials Sciences     Open Access   (Followers: 14)
Advances in Operations Research     Open Access   (Followers: 12)
Advances in Parallel Computing     Full-text available via subscription   (Followers: 6)
Advances in Porous Media     Full-text available via subscription   (Followers: 5)
Advances in Remote Sensing     Open Access   (Followers: 43)
Advances in Science and Research (ASR)     Open Access   (Followers: 4)
Advances in Technology Innovation     Open Access   (Followers: 5)
AEU - International Journal of Electronics and Communications     Hybrid Journal   (Followers: 8)
African Journal of Information and Communication     Open Access   (Followers: 6)
African Journal of Mathematics and Computer Science Research     Open Access   (Followers: 4)
AI EDAM     Hybrid Journal  
Air, Soil & Water Research     Open Access   (Followers: 11)
AIS Transactions on Human-Computer Interaction     Open Access   (Followers: 5)
Algebras and Representation Theory     Hybrid Journal   (Followers: 1)
Algorithms     Open Access   (Followers: 11)
American Journal of Computational and Applied Mathematics     Open Access   (Followers: 5)
American Journal of Computational Mathematics     Open Access   (Followers: 4)
American Journal of Information Systems     Open Access   (Followers: 5)
American Journal of Sensor Technology     Open Access   (Followers: 4)
Anais da Academia Brasileira de Ciências     Open Access   (Followers: 2)
Analog Integrated Circuits and Signal Processing     Hybrid Journal   (Followers: 7)
Analysis in Theory and Applications     Hybrid Journal   (Followers: 1)
Animation Practice, Process & Production     Hybrid Journal   (Followers: 5)
Annals of Combinatorics     Hybrid Journal   (Followers: 3)
Annals of Data Science     Hybrid Journal   (Followers: 11)
Annals of Mathematics and Artificial Intelligence     Hybrid Journal   (Followers: 12)
Annals of Pure and Applied Logic     Open Access   (Followers: 2)
Annals of Software Engineering     Hybrid Journal   (Followers: 13)
Annual Reviews in Control     Hybrid Journal   (Followers: 6)
Anuario Americanista Europeo     Open Access  
Applicable Algebra in Engineering, Communication and Computing     Hybrid Journal   (Followers: 2)
Applied and Computational Harmonic Analysis     Full-text available via subscription   (Followers: 1)
Applied Artificial Intelligence: An International Journal     Hybrid Journal   (Followers: 13)
Applied Categorical Structures     Hybrid Journal   (Followers: 2)
Applied Clinical Informatics     Hybrid Journal   (Followers: 2)
Applied Computational Intelligence and Soft Computing     Open Access   (Followers: 11)
Applied Computer Systems     Open Access   (Followers: 2)
Applied Informatics     Open Access  
Applied Mathematics and Computation     Hybrid Journal   (Followers: 33)
Applied Medical Informatics     Open Access   (Followers: 10)
Applied Numerical Mathematics     Hybrid Journal   (Followers: 5)
Applied Soft Computing     Hybrid Journal   (Followers: 16)
Applied Spatial Analysis and Policy     Hybrid Journal   (Followers: 4)
Applied System Innovation     Open Access  
Architectural Theory Review     Hybrid Journal   (Followers: 3)
Archive of Applied Mechanics     Hybrid Journal   (Followers: 5)
Archive of Numerical Software     Open Access  
Archives and Museum Informatics     Hybrid Journal   (Followers: 130)
Archives of Computational Methods in Engineering     Hybrid Journal   (Followers: 5)
arq: Architectural Research Quarterly     Hybrid Journal   (Followers: 7)
Artifact     Hybrid Journal   (Followers: 2)
Artificial Life     Hybrid Journal   (Followers: 7)
Asia Pacific Journal on Computational Engineering     Open Access  
Asia-Pacific Journal of Information Technology and Multimedia     Open Access   (Followers: 1)
Asian Journal of Computer Science and Information Technology     Open Access  
Asian Journal of Control     Hybrid Journal  
Assembly Automation     Hybrid Journal   (Followers: 2)
at - Automatisierungstechnik     Hybrid Journal   (Followers: 1)
Australian Educational Computing     Open Access   (Followers: 1)
Automatic Control and Computer Sciences     Hybrid Journal   (Followers: 4)
Automatic Documentation and Mathematical Linguistics     Hybrid Journal   (Followers: 5)
Automatica     Hybrid Journal   (Followers: 11)
Automation in Construction     Hybrid Journal   (Followers: 6)
Autonomous Mental Development, IEEE Transactions on     Hybrid Journal   (Followers: 9)
Basin Research     Hybrid Journal   (Followers: 5)
Behaviour & Information Technology     Hybrid Journal   (Followers: 53)
Big Data and Cognitive Computing     Open Access   (Followers: 2)
Biodiversity Information Science and Standards     Open Access  
Bioinformatics     Hybrid Journal   (Followers: 281)
Biomedical Engineering     Hybrid Journal   (Followers: 15)
Biomedical Engineering and Computational Biology     Open Access   (Followers: 13)
Biomedical Engineering, IEEE Reviews in     Full-text available via subscription   (Followers: 19)
Biomedical Engineering, IEEE Transactions on     Hybrid Journal   (Followers: 35)
Briefings in Bioinformatics     Hybrid Journal   (Followers: 43)
British Journal of Educational Technology     Hybrid Journal   (Followers: 142)
Broadcasting, IEEE Transactions on     Hybrid Journal   (Followers: 10)
c't Magazin fuer Computertechnik     Full-text available via subscription   (Followers: 1)
CALCOLO     Hybrid Journal  
Calphad     Hybrid Journal  
Canadian Journal of Electrical and Computer Engineering     Full-text available via subscription   (Followers: 14)
Capturing Intelligence     Full-text available via subscription  
Catalysis in Industry     Hybrid Journal   (Followers: 1)
CEAS Space Journal     Hybrid Journal   (Followers: 2)
Cell Communication and Signaling     Open Access   (Followers: 2)
Central European Journal of Computer Science     Hybrid Journal   (Followers: 5)
CERN IdeaSquare Journal of Experimental Innovation     Open Access   (Followers: 2)
Chaos, Solitons & Fractals     Hybrid Journal   (Followers: 3)
Chemometrics and Intelligent Laboratory Systems     Hybrid Journal   (Followers: 14)
ChemSusChem     Hybrid Journal   (Followers: 7)
China Communications     Full-text available via subscription   (Followers: 7)
Chinese Journal of Catalysis     Full-text available via subscription   (Followers: 2)
CIN Computers Informatics Nursing     Full-text available via subscription   (Followers: 11)
Circuits and Systems     Open Access   (Followers: 15)
Clean Air Journal     Full-text available via subscription   (Followers: 1)
CLEI Electronic Journal     Open Access  
Clin-Alert     Hybrid Journal   (Followers: 1)
Cluster Computing     Hybrid Journal   (Followers: 1)
Cognitive Computation     Hybrid Journal   (Followers: 4)
COMBINATORICA     Hybrid Journal  
Combinatorics, Probability and Computing     Hybrid Journal   (Followers: 4)
Combustion Theory and Modelling     Hybrid Journal   (Followers: 14)
Communication Methods and Measures     Hybrid Journal   (Followers: 12)
Communication Theory     Hybrid Journal   (Followers: 20)
Communications Engineer     Hybrid Journal   (Followers: 1)
Communications in Algebra     Hybrid Journal   (Followers: 3)
Communications in Computational Physics     Full-text available via subscription   (Followers: 2)
Communications in Partial Differential Equations     Hybrid Journal   (Followers: 3)
Communications of the ACM     Full-text available via subscription   (Followers: 52)
Communications of the Association for Information Systems     Open Access   (Followers: 16)
COMPEL: The International Journal for Computation and Mathematics in Electrical and Electronic Engineering     Hybrid Journal   (Followers: 3)
Complex & Intelligent Systems     Open Access   (Followers: 1)
Complex Adaptive Systems Modeling     Open Access  
Complex Analysis and Operator Theory     Hybrid Journal   (Followers: 2)
Complexity     Hybrid Journal   (Followers: 6)
Complexus     Full-text available via subscription  
Composite Materials Series     Full-text available via subscription   (Followers: 8)
Computación y Sistemas     Open Access  
Computation     Open Access  
Computational and Applied Mathematics     Hybrid Journal   (Followers: 2)
Computational and Mathematical Methods in Medicine     Open Access   (Followers: 2)
Computational and Mathematical Organization Theory     Hybrid Journal   (Followers: 2)
Computational and Structural Biotechnology Journal     Open Access   (Followers: 2)
Computational and Theoretical Chemistry     Hybrid Journal   (Followers: 9)
Computational Astrophysics and Cosmology     Open Access   (Followers: 1)
Computational Biology and Chemistry     Hybrid Journal   (Followers: 11)
Computational Chemistry     Open Access   (Followers: 2)
Computational Cognitive Science     Open Access   (Followers: 2)
Computational Complexity     Hybrid Journal   (Followers: 4)
Computational Condensed Matter     Open Access  
Computational Ecology and Software     Open Access   (Followers: 9)
Computational Economics     Hybrid Journal   (Followers: 9)
Computational Geosciences     Hybrid Journal   (Followers: 15)
Computational Linguistics     Open Access   (Followers: 23)
Computational Management Science     Hybrid Journal  
Computational Mathematics and Modeling     Hybrid Journal   (Followers: 8)
Computational Mechanics     Hybrid Journal   (Followers: 5)
Computational Methods and Function Theory     Hybrid Journal  
Computational Molecular Bioscience     Open Access   (Followers: 2)
Computational Optimization and Applications     Hybrid Journal   (Followers: 7)
Computational Particle Mechanics     Hybrid Journal   (Followers: 1)
Computational Research     Open Access   (Followers: 1)
Computational Science and Discovery     Full-text available via subscription   (Followers: 2)
Computational Science and Techniques     Open Access  
Computational Statistics     Hybrid Journal   (Followers: 14)
Computational Statistics & Data Analysis     Hybrid Journal   (Followers: 30)
Computer     Full-text available via subscription   (Followers: 94)
Computer Aided Surgery     Hybrid Journal   (Followers: 6)
Computer Applications in Engineering Education     Hybrid Journal   (Followers: 8)
Computer Communications     Hybrid Journal   (Followers: 10)
Computer Engineering and Applications Journal     Open Access   (Followers: 5)
Computer Journal     Hybrid Journal   (Followers: 9)
Computer Methods in Applied Mechanics and Engineering     Hybrid Journal   (Followers: 23)
Computer Methods in Biomechanics and Biomedical Engineering     Hybrid Journal   (Followers: 12)
Computer Methods in the Geosciences     Full-text available via subscription   (Followers: 2)
Computer Music Journal     Hybrid Journal   (Followers: 19)

        1 2 3 4 5 6 | Last

Journal Cover Chemometrics and Intelligent Laboratory Systems
  [SJR: 0.697]   [H-I: 92]   [14 followers]  Follow
    
   Hybrid Journal Hybrid journal (It can contain Open Access articles)
   ISSN (Print) 0169-7439
   Published by Elsevier Homepage  [3175 journals]
  • A tool for simulating multi-response linear model data
    • Authors: Raju Rimal; Trygve Almøy; Solve Sæbø
      Pages: 1 - 10
      Abstract: Publication date: Available online 23 February 2018
      Source:Chemometrics and Intelligent Laboratory Systems
      Author(s): Raju Rimal, Trygve Almøy, Solve Sæbø
      Data science is generating enormous amounts of data, and new and advanced analytical methods are constantly being developed to cope with the challenge of extracting information from such “big-data”. Researchers often use simulated data to assess and document the properties of these new methods, and in this paper we present an extension to the R-package simrel, which is a versatile and transparent tool for simulating linear model data with an extensive range of adjustable properties. The method is based on the concept of relevant components, and is equivalent to the newly developed envelope model. It is a multi-response extension of R-package simrel which is available in R-package repository CRAN, and as simrel the new approach is essentially based on random rotations of latent relevant components to obtain a predictor matrix X , but in addition we introduce random rotations of latent components spanning a response space in order to obtain a multivariate response matrix Y . The properties of the linear relation between X and Y are defined by a small set of input parameters which allow versatile and adjustable simulations. Sub-space rotations also allow for generating data suitable for testing variable selection methods in multi-response settings. The method is implemented as an update to the R-package simrel.

      PubDate: 2018-02-26T08:05:50Z
      DOI: 10.1016/j.chemolab.2018.02.009
      Issue No: Vol. 176 (2018)
       
  • Neural component analysis for fault detection
    • Authors: Haitao Zhao
      Pages: 11 - 21
      Abstract: Publication date: Available online 16 February 2018
      Source:Chemometrics and Intelligent Laboratory Systems
      Author(s): Haitao Zhao
      Principal component analysis (PCA) is largely adopted for chemical process monitoring and numerous PCA-based systems have been developed to solve various fault detection and diagnosis problems. Since PCA-based methods assume that the monitored process is linear, nonlinear PCA models, such as autoencoder models and kernel principal component analysis (KPCA), has been proposed and applied to nonlinear process monitoring. However, KPCA-based methods need to perform eigen-decomposition (ED) on the kernel Gram matrix whose dimensions depend on the number of training data. Moreover, prefixed kernel parameters cannot be most effective for different faults which may need different parameters to maximize their respective detection performances. Autoencoder models lack the consideration of orthogonal constraints which is crucial for PCA-based algorithms. To address these problems, this paper proposes a novel nonlinear method, called neural component analysis (NCA), which intends to train a feedforward neural work with orthogonal constraints such as those used in PCA. NCA is a unified model including a nonlinear encoder and a linear decoder. NCA can adaptively learn its parameters through backpropagation and the dimensionality of the nonlinear features has no relationship with the number of training samples. Extensive experimental results on the Tennessee Eastman (TE) benchmark process show the superiority of NCA in terms of missed detection rate (MDR) and false alarm rate (FAR). The source code of NCA can be found in https://github.com/haitaozhao/Neural-Component-Analysis.git.

      PubDate: 2018-02-26T08:05:50Z
      DOI: 10.1016/j.chemolab.2018.02.001
      Issue No: Vol. 176 (2018)
       
  • Parzen window distribution as new membership function for ANFIS algorithm-
           Application to a distillation column faults prediction
    • Authors: Alaa Daher; Ghaleb Hoblos; Mohamad Khalil; Yahya Chetouani
      Pages: 1 - 12
      Abstract: Publication date: 15 April 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 175
      Author(s): Alaa Daher, Ghaleb Hoblos, Mohamad Khalil, Yahya Chetouani
      The distillation column is one of the most important unit operations used in the chemical engineering. The continuous distillation process is largely used in many applications such as petrochemical production, natural gas processing, and petroleum refineries, and many others. Corrective maintenance of the chemical reactors represents a consequential problem because it is very costly and it disrupts production for long periods of time. In addition, most of the time, this may lead to harmful effects and disastrous results. The most common solution has been to rely on preventive maintenance. Unfortunately, this has been both expensive and inadequate. Therefore, the optimal solution is to resort to predictive maintenance that involves the design of a pre-crash control system and a higher ex-ante understanding of the future path of the reactor. This research paper aims to propose the Adaptive Neuro Fuzzy Inference System (ANFIS) as a superior technique that can forecast the future path of the distillation column system. In addition, this paper will propose Parzen windows distribution as a new membership function in order to improve ANFIS performance either by reducing consumption time and making processing closer to real-time application, or by minimizing the root means square error (RMSE) between real and predictive data. This methodology was tested on real experimental data obtained from a distillation column with the aim of predicting failures that may possibly occur during the automated continuous distillation process. A comparative study was necessary in order to properly select the superior membership function that can be used for the ANFIS algorithm when ANFIS is applied to the distillation column data. Results demonstrated the importance of the proposed technique since it proved to be highly successful in terms of reducing time consumed. Additionally, Parzen windows had the smallest RMSE for many signals in both normal and degraded modes.

      PubDate: 2018-02-26T08:05:50Z
      DOI: 10.1016/j.chemolab.2018.01.002
      Issue No: Vol. 175 (2018)
       
  • Time fractional super-diffusion model and its application in
           peak-preserving smoothing
    • Authors: Yuanlu Li; Min Jiang; Fawang Liu
      Pages: 13 - 19
      Abstract: Publication date: 15 April 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 175
      Author(s): Yuanlu Li, Min Jiang, Fawang Liu
      The super-diffusion model is suggested for peak-preserving smoothing. In this model, the time derivative on the left of the classical diffusion model is replaced with the time fractional derivative. Because of the weight property of the fractional derivative, the super-diffusion model can further improve the smooth performance of the classical nonlinear diffusion model. An explicit difference scheme and an implicit difference scheme are given. Then some comparisons between the proposed model and the classical nonlinear diffusion model are done. The results indicate the proposed model outperforms the classic nonlinear diffusion model. In the end, the proposed method is used to smooth a nuclear magnetic resonance spectroscopy and a mass spectrometry.

      PubDate: 2018-02-26T08:05:50Z
      DOI: 10.1016/j.chemolab.2018.02.005
      Issue No: Vol. 175 (2018)
       
  • Multi-grade principal component analysis for fault detection with multiple
           production grades
    • Authors: Le Zhou; Junghui Chen; Beiping Hou; Zhihuan Song
      Pages: 20 - 29
      Abstract: Publication date: 15 April 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 175
      Author(s): Le Zhou, Junghui Chen, Beiping Hou, Zhihuan Song
      In many chemical industries, a production line usually produces various products with different grades to meet the demands of the worldwide market. A process with multiple grades is not suitable to be described using a traditional single model. In this paper, a multi-grade principal component analysis (MGPCA) model is proposed for multi-grade process modeling and fault detection purposes. The proposed MGPCA can use the measurements from different grades with unequal sizes and to extract the essential information from the multi-grade process. The model is derived in a probabilistic framework and the corresponding parameters are estimated by the expectation-maximization algorithm. Finally, a simulated case and a real industrial polyethylene process with multiple grades are tested to evaluate the property of the proposed method.

      PubDate: 2018-02-26T08:05:50Z
      DOI: 10.1016/j.chemolab.2018.02.003
      Issue No: Vol. 175 (2018)
       
  • Classification of 1-methylcyclopropene treated apples by fluorescence
           fingerprint using partial least squares discriminant analysis with
           stepwise selectivity ratio variable selection method
    • Authors: Vipavee Trivittayasil; Mizuki Tsuta; Satoshi Kasai; Yosuke Matsuo; Yasuyo Sekiyama; Toshihiko Shoji; Ryoko Aiyama; Mito Kokawa; Junichi Sugiyama
      Pages: 30 - 36
      Abstract: Publication date: 15 April 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 175
      Author(s): Vipavee Trivittayasil, Mizuki Tsuta, Satoshi Kasai, Yosuke Matsuo, Yasuyo Sekiyama, Toshihiko Shoji, Ryoko Aiyama, Mito Kokawa, Junichi Sugiyama
      In this study, we investigated the potential of using fluorescence fingerprint (FF) for nondestructive identification of apples treated with 1-methylcyclopropene (1-MCP). In total, 442 apples of two cultivars (Fuji and Orin) and different storage times (0, 4, 5, 6, and 8 months) were assessed. The classification model used in this study was built using partial least squares discriminant analysis (PLSDA) with the stepwise selectivity ratio (SR) method. The stepwise SR method is a recursive variable selection method proposed in this study. FF was capable of classifying 1-MCP-treated apples with accuracies of 91.23%, 89.74%, and 90.17% for calibration, cross-validation, and validation results, respectively. PLSDA with the stepwise SR method could identify four aggregations of wavelength conditions, which are important to the classification. In addition, a non-targeted approach was taken to screen the metabolites characterizing 1-MCP-treated and control apples by liquid chromatography-mass spectrometry (LC/MS) and nuclear magnetic resonance (NMR) spectroscopy. The observed difference in metabolic profiles may contribute to the difference in the fluorescence profiles of 1-MCP treated and control apples.

      PubDate: 2018-02-26T08:05:50Z
      DOI: 10.1016/j.chemolab.2018.02.004
      Issue No: Vol. 175 (2018)
       
  • Kernel-Partial Least Squares regression coupled to pseudo-sample
           trajectories for the analysis of mixture designs of experiments
    • Authors: Raffaele Vitale; Daniel Palací-López; Harmen H.M. Kerkenaar; Geert J. Postma; Lutgarde M.C. Buydens; Alberto Ferrer
      Pages: 37 - 46
      Abstract: Publication date: 15 April 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 175
      Author(s): Raffaele Vitale, Daniel Palací-López, Harmen H.M. Kerkenaar, Geert J. Postma, Lutgarde M.C. Buydens, Alberto Ferrer
      This article explores the potential of Kernel-Partial Least Squares (K-PLS) regression for the analysis of data proceeding from mixture designs of experiments. Gower's idea of pseudo-sample trajectories is exploited for interpretation purposes. The results show that, when the datasets under study are affected by severe non-linearities and comprise few observations, the proposed approach can represent a feasible alternative to classical methodologies (i.e. Scheffé polynomial fitting by means of Ordinary Least Squares - OLS - and Cox polynomial fitting by means of Partial Least Squares - PLS). Furthermore, a way of recovering the parameters of a Scheffé model (provided that it holds and has the same complexity as the K-PLS one) from the trend of the aforementioned pseudo-sample trajectories is illustrated via a simulated case-study.

      PubDate: 2018-02-26T08:05:50Z
      DOI: 10.1016/j.chemolab.2018.02.002
      Issue No: Vol. 175 (2018)
       
  • A new strategy of least absolute shrinkage and selection operator coupled
           with sampling error profile analysis for wavelength selection
    • Authors: Ruoqiu Zhang; Feiyu Zhang; Wanchao Chen; Heming Yao; Jiong Ge; Shengchao Wu; Ting Wu; Yiping Du
      Pages: 47 - 54
      Abstract: Publication date: 15 April 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 175
      Author(s): Ruoqiu Zhang, Feiyu Zhang, Wanchao Chen, Heming Yao, Jiong Ge, Shengchao Wu, Ting Wu, Yiping Du
      A new strategy based on sampling error profile analysis (SEPA) combined with least absolute shrinkage and selection operator (SEPA-LASSO) was proposed. LASSO has been proven to be effective for multivariate calibration with automatic variable selection for high-dimensional data. However, in the previous research, the critical process of multivariate calibration by LASSO was an optimization of 1-norm turning parameter for a fixed sample set without considering the behaviors of variable selection by different subsets of samples. In the present work, Monte Carlo Sampling (MCS), the core of SEPA framework, is used to investigate various sub-models. Least angle regression (LAR) is used to solve LASSO, and various LAR iteration including certain number of variables could be obtained instead of choosing the numerical values of 1-norm turning parameters. SEPA-LASSO algorithm consists of plenty of loops. Under the SEPA framework and LAR algorithm, a number of LASSO sub-models with the same dimensions are built by MCS in each loop, the vote rule is used to determine the importance of variables and select them to build variable subsets. After running the loops, several subsets of variables are obtained and their error profile is used to choose the optimal subset of variables. The performance of SEPA-LASSO was evaluated by three near-infrared (NIR) datasets. The results show that the model built by SEPA-LASSO has excellent predictability and interpretability, compared with some commonly used multivariate calibration methods, such as principal component regression (PCR) and partial least squares (PLS), as well as some wavelength selection methods including LASSO, moving window partial least squares regression (MWPLSR), Monte Carlo uninformative variable elimination (MC-UVE), ordered homogeneity pursuit lasso (OHPL) and stability competitive adaptive reweighted sampling (SCARS).

      PubDate: 2018-02-26T08:05:50Z
      DOI: 10.1016/j.chemolab.2018.02.007
      Issue No: Vol. 175 (2018)
       
  • Midpoint-radii principal component analysis -based EWMA and application to
           air quality monitoring network
    • Authors: M. Mansouri; M.-F. Harkat; M. Nounou; H. Nounou
      Pages: 55 - 64
      Abstract: Publication date: 15 April 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 175
      Author(s): M. Mansouri, M.-F. Harkat, M. Nounou, H. Nounou
      Monitoring air quality is crucial for the safety of humans and the environment. Moreover, real world data collected from air quality network is often affected by different types of errors as measurement noise and variability of pollutant concentrations. The uncertainty in the data, which is strictly connected to the above errors, may be treated by considering interval-valued data analysis. In practical cases of measured data, the true value cannot be measured and the collected data on a process are only approximations given by sensors, and are thus imprecise. This is due mainly to the uncertainties induced by measurement errors or determined by specific experimental conditions. Thus, the main aim of this paper is to develop an enhanced monitoring of air quality network by taking into account the uncertainties on the data. To do that, we develop a new monitoring technique that merges the advantages of Midpoint-radii PCA (MRPCA) method with exponentially weighted moving average (EWMA) chart, in order to enhance sensor fault detection technique of air quality monitoring process. MRPCA is the most popular interval multivariate statistical method, able to tackle the issue of uncertainties on the models and one way to improve the fault detection abilities. On the other hand, the EWMA statistic allows an exponential weighted average to successive observations and able to detect small and moderate faults. The developed MRPCA-based EWMA method relies on using MRPCA as a modeling framework for fault detection and EWMA as a detection chart. The proposed MRPCA-based EWMA scheme is illustrated using a simulation example and applied for sensor fault detection of an air quality monitoring network. The monitoring performances of the developed technique are compared to the classical monitoring techniques. MRPCA model performances are compared with the interval PCA models: complete-information principal component analysis (CIPCA) and Centers PCA (CPCA). The MRPCA-based EWMA monitoring performances are compared to MRPCA-based Shewhart, generalized likelihood ratio test (GLRT) and squared prediction error (SPE) techniques.

      PubDate: 2018-02-26T08:05:50Z
      DOI: 10.1016/j.chemolab.2018.01.016
      Issue No: Vol. 175 (2018)
       
  • Efficient and robust analysis of interlaboratory studies
    • Authors: Jaap Molenaar; Wim P. Cofino; Paul J.J.F. Torfs
      Pages: 65 - 73
      Abstract: Publication date: 15 April 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 175
      Author(s): Jaap Molenaar, Wim P. Cofino, Paul J.J.F. Torfs
      In this paper we present the ab-initio derivation of an estimator for the mean and variance of a sample of data, such as obtained from proficiency tests. This estimator has already been used for some time in this kind of analyses, but a thorough derivation together with a detailed analysis of its properties is missing until now. The estimator uses the information contained in data including uncertainty, represented via probability density functions (pdfs). An implementation of the approach is given that can be used if the uncertainty information is not available; the so-called normal distribution approach (NDA). The present estimation procedure is based on calculating the centroid of the ensemble of pdfs. This centroid is obtained by solving the eigenvalue problem for the so-called similarity matrix. Elements of this matrix measure the similarity (or overlap) between different pdfs in terms of the Bhattacharyya coefficient. Since evaluation of an eigenvalue problem is standard nowadays, the method is extremely fast. The first and second moments of the centroid pdf are used to obtain the mean and variance of the dataset. The properties of the estimator are extensively analyzed. We derive its variance and show the connection between the present estimator and Principal Component Analysis. Furthermore, we study its behavior in several limiting cases, as met in data that are very coherent or very incoherent, and check its consistency. In particular, we investigate how sensitive the estimator is for outliers, investigating its breakdown point. In the normal distribution approach the breakdown point of the estimator is shown to be optimal, i.e., 50%. The largest eigenvalue(s) of the similarity matrix appear(s) to provide important information. If the largest eigenvalue is close to the dimension of the matrix, this indicates that the data are very coherent, so they lie close to each other with similar uncertainties. If there are two (or more) largest eigenvalues with (nearly) equal values, this indicates that the data fall apart in two (or more) clusters.

      PubDate: 2018-02-26T08:05:50Z
      DOI: 10.1016/j.chemolab.2018.01.003
      Issue No: Vol. 175 (2018)
       
  • Stochastic cross validation
    • Authors: Lu Xu; Hai-Yan Fu; Mohammad Goodarzi; Chen-Bo Cai; Qiao-Bo Yin; Ya Wu; Bang-Cheng Tang; Yuan-Bin She
      Pages: 74 - 81
      Abstract: Publication date: 15 April 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 175
      Author(s): Lu Xu, Hai-Yan Fu, Mohammad Goodarzi, Chen-Bo Cai, Qiao-Bo Yin, Ya Wu, Bang-Cheng Tang, Yuan-Bin She
      Cross validation (CV) is by far one of the most commonly used methods to estimate model complexity for partial least squares (PLS). In this study, stochastic cross validation (SCV) was proposed as a novel CV strategy, where the percent of left-out objects (PLOO) was defined as a changeable random number. We proposed two SCV strategies, namely, SCV with uniformly distributed PLOO (SCV-U) and SCV with normally distributed PLOO (SCV-N). SCV-U is actually a hybrid of leave-one-out CV (LOOCV), k-fold CV and Monte Carlo CV (MCCV). The rationale behind SCV-N is that the probability of large perturbations of the original training set will be small. SCV is expected to provide more flexibility for data splitting to explore and learn from the data set and evaluate internally a built model. SCV-U and SCV-N were used for PLS calibrations of three real data sets as well as a simulated data set and they were compared with LOOCV, k-fold CV and MCCV. Given a training and external validation set, different CV techniques were repeatedly used to evaluate the optimal model complexity and the prediction results were compared. The results indicate that SCV-U and SCV-N could provide useful alternatives to the traditional CV methods and SCV is less sensitive to the values of PLOO.

      PubDate: 2018-02-26T08:05:50Z
      DOI: 10.1016/j.chemolab.2018.02.008
      Issue No: Vol. 175 (2018)
       
  • Dynamic soft sensors with active forward-update learning for selection of
           useful data from historical big database
    • Authors: Lester Lik Teck Chan; Qing-Yang Wu; Junghui Chen
      Pages: 87 - 103
      Abstract: Publication date: Available online 5 February 2018
      Source:Chemometrics and Intelligent Laboratory Systems
      Author(s): Lester Lik Teck Chan, Qing-Yang Wu, Junghui Chen
      Conventional static soft sensor is incapable of handling the dynamic of processes. With abundance of data, the problem of variable correlations and a large number of samples are encountered; moreover, the quality of the data for the construction of the soft sensors can be crucial for performance. An active learning strategy based on a latent variable model (LVM) to select representative data for efficient development of the dynamic soft sensor model is proposed. The uncertainty information for data selection is provided by the Gaussian process (GP) model. The developed LVM with the auxiliary GP model can handle the process dynamic. An active forward-update scheme which can update the soft sensor model in advance is proposed to reflect the current status of the process and improve the prediction performance without waiting for the quality measurements. Two case studies are done to demonstrate the features and the applicability of the proposed method.

      PubDate: 2018-02-26T08:05:50Z
      DOI: 10.1016/j.chemolab.2018.01.015
      Issue No: Vol. 175 (2018)
       
  • Highly-overlapped, recursive partial least squares soft sensor with state
           partitioning via local variable selection
    • Authors: Dominic V. Poerio; Steven D. Brown
      Pages: 104 - 115
      Abstract: Publication date: Available online 15 February 2018
      Source:Chemometrics and Intelligent Laboratory Systems
      Author(s): Dominic V. Poerio, Steven D. Brown
      We report the use of a soft sensor ensemble based on recursive partial least squares with a large number of overlapping models. The proposed method uses process memory attenuation in the ensemble by varying the number of training samples included in each model, while always including the most recent samples, which are usually the most relevant for prediction of new samples, and also ensures that no local models are invalidated due to drift. To achieve state partitioning of the process data, covariance-based variable selection is performed on each of the model regions to ensure that only variables most relevant to the dominant process state are included in each of the local models. This approach yields a distribution of predictions from the models, permitting a prediction of the target property from the summary statistics of the observed prediction distribution. The effectiveness of the proposed method is demonstrated by testing against a conventional global soft sensor as well as a state-localized soft sensor, both with and without variable selection, on two soft sensing applications developed from real industrial processes employing various model updating frequencies. Results from the experiments demonstrate that the proposed method tends to outperform a global soft sensor in most cases, and is highly competitive with the compared state-localized soft sensor, indicating that the proposed method achieves accurate state partitioning.

      PubDate: 2018-02-26T08:05:50Z
      DOI: 10.1016/j.chemolab.2018.02.006
      Issue No: Vol. 175 (2018)
       
  • Hierarchical mixture of linear regressions for multivariate spectroscopic
           calibration
    • Authors: Chenhao Cui; Tom Fearn
      Pages: 1 - 14
      Abstract: Publication date: 15 March 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 174
      Author(s): Chenhao Cui, Tom Fearn
      This paper investigates the use of the hierarchical mixture of linear regressions (HMLR) and variational inference for multivariate spectroscopic calibration. The performance of HMLR is compared to the classical methods: partial least squares regression (PLSR), and PLS embedded locally weighted regression (LWR) on three different NIR datasets, including a publicly accessible one. In these tests, HMLR outperformed the other two benchmark methods. Compared to LWR, HMLR is parametric, which makes it interpretable and easy to use. In addition, HMLR provides a novel calibration scheme to build a two-tier PLS regression model automatically. This is especially useful when the investigated constituent covers a large range.

      PubDate: 2018-02-05T06:51:13Z
      DOI: 10.1016/j.chemolab.2017.12.013
      Issue No: Vol. 174 (2018)
       
  • Ensemble deep kernel learning with application to quality prediction in
           industrial polymerization processes
    • Authors: Yi Liu; Chao Yang; Zengliang Gao; Yuan Yao
      Pages: 15 - 21
      Abstract: Publication date: 15 March 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 174
      Author(s): Yi Liu, Chao Yang, Zengliang Gao, Yuan Yao
      For predicting the melt index (MI) in industrial polymerization processes, traditional data-driven empirical models do not utilize the information in a large amount of the unlabeled data. To overcome this data-rich-but-information-poor (DRIP) problem in polymer industries, an ensemble deep kernel learning (EDKL) model is proposed. With an unsupervised learning stage, the deep brief network is adopted to extract useful information from the available data. Then, a kernel learning regression model is formulated to obtain a nonlinear relationship between the extracted features and MI values. Moreover, a bagging-based ensemble strategy is integrated into the deep kernel learning method to enhance the reliability of the prediction model. The industrial MI prediction results demonstrate the advantages of the developed EDKL model as compared with conventional supervised soft sensors (e.g., partial least squares and support vector regression) that only use the limited labeled data.

      PubDate: 2018-02-05T06:51:13Z
      DOI: 10.1016/j.chemolab.2018.01.008
      Issue No: Vol. 174 (2018)
       
  • Improving prediction of extracellular matrix proteins using evolutionary
           information via a grey system model and asymmetric under-sampling
           technique
    • Authors: Muhammad Kabir; Saeed Ahmad; Muhammad Iqbal; Zar Nawab Khan Swati; Zi Liu; Dong-Jun Yu
      Pages: 22 - 32
      Abstract: Publication date: 15 March 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 174
      Author(s): Muhammad Kabir, Saeed Ahmad, Muhammad Iqbal, Zar Nawab Khan Swati, Zi Liu, Dong-Jun Yu
      Extracellular Matrix proteins (ECMP) play vigorous part in performing various biological functions including cell migration, adhesion, proliferation, differentiation. Furthermore, embryonic development, angiogenesis, gene expression, and tumor growth are also regulated by ECMP. In view of this incredible significance, precise and reliable identification of ECMP through computational techniques is highly requisite. Although, previous works made substantial improvement, however, accurately predicting ECMP from primary protein sequence is still at the infant stage due to the rapid growth of proteins samples in online databases. In the current study, a novel sequence-based prediction method called TargetECMP has been proposed, which is based on the evolutionary information extracted via a grey system model. It utilizes asymmetric under-sampling approach for splitting the benchmark dataset into eleven subsets in order to avoid class imbalance problem. Jackknife cross-validation test is performed with support vector machine (SVM) on each subset of data and then ensemble majority voting is utilized to integrate outputs of SVM against each subset. The experimental results achieved by TargetECMP outperformed the existing predictor on both benchmark dataset and independent dataset. Owning to best prediction results provided by TargetECMP, it is demonstrated that the analysis will provide novel insights into basic research, drug discovery and academia in general and function of extracellular matrix proteins in particular.

      PubDate: 2018-02-05T06:51:13Z
      DOI: 10.1016/j.chemolab.2018.01.004
      Issue No: Vol. 174 (2018)
       
  • Multivariate comparison of classification performance measures
    • Authors: Davide Ballabio; Francesca Grisoni; Roberto Todeschini
      Pages: 33 - 44
      Abstract: Publication date: 15 March 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 174
      Author(s): Davide Ballabio, Francesca Grisoni, Roberto Todeschini
      The assessment of the classification performance can be based on class indices, such as sensitivity, specificity and precision, which describe the classification results achieved on each modelled class. However, in several situations, it is useful to represent the global classification performance with a single number. Therefore, several measures have been introduced in literature to deal with this assessment, accuracy being the most known and used. These metrics have been proposed to generally face binary classification tasks and can behave differently depending on the classification scenario. In this study, different global measures of classification performances are compared by means of results achieved on an extended set of real multivariate datasets. The systematic comparison is carried out through multivariate analysis. Further investigations are then derived on specific indices to understand how the presence of unbalanced classes and the number of modelled classes can influence their behaviour. Finally, this work introduces a set of benchmark values based on different random classification scenarios. These benchmark thresholds can serve as the initial criterion to accept or reject a classification model on the basis of its performance.

      PubDate: 2018-02-05T06:51:13Z
      DOI: 10.1016/j.chemolab.2017.12.004
      Issue No: Vol. 174 (2018)
       
  • Alternate deflation and inflation of search space in reweighted sampling:
           An effective variable selection approach for PLS model
    • Authors: Biswanath Mahanty
      Pages: 45 - 55
      Abstract: Publication date: 15 March 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 174
      Author(s): Biswanath Mahanty
      Based on assessment of randomized sub-model populations generated through reweighted binary matrix sampling (BMS), an innovative variable selection strategy for PLS regression model, called alternate deflation and inflation of search space (ADISS) is proposed. Normalized regression coefficients of best PLS sub-models population is used to formulate the weight vector for re-weighted BMS. Unlike the most existing algorithm, ADISS alternatively shifts between forward selection (inflation) and backward elimination (deflation) of variable space, minimizing the risk of accidental loss of informative variables. Compared with methods such as competitive adaptive reweighted sampling (CARS), variable iterative space shrinkage approach (VISSA), or Monte Carlo uninformative variable elimination (MC-UVE), proposed method showed lower cross-validation or prediction error for two different benchmark NIR data sets. ADISS frequently selects nearly the same sets of variables across multiple independent runs, that signifies stability of the output. The unsupervised execution, termination and projection of final variable set from the algorithm is important advantage while considering for large scale data.

      PubDate: 2018-02-05T06:51:13Z
      DOI: 10.1016/j.chemolab.2018.01.005
      Issue No: Vol. 174 (2018)
       
  • A novel convolutional neural network based approach to predictions of
           process dynamic time delay sequences
    • Authors: Bo Yang; Hongguang Li
      Pages: 56 - 61
      Abstract: Publication date: 15 March 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 174
      Author(s): Bo Yang, Hongguang Li
      It is practical that correlated process variables always involve dynamic time-delay sequences. In this paper, a novel convolutional neural network (CNN) based approach is proposed to predict dynamic time delay sequences. Firstly, according to the calculating similarities between correlated process variables, the time delay sequence is extracted offline using a dynamic time delay analysis by elastic windows (EW-DTDA) method. In addition, through an additional correlation analysis between the time delay sequence and process variables data, the process variables majorly influencing the time delay sequences can be obtained. Finally, a deep learning CNN model between the extracted time delay sequence and the obtained majorly influencing variables is constructed to predict the time delay sequence online. In order to validate the effectiveness of the proposed method, the method is applied to a real distillation column for analyzing dynamic time delay sequences, the simulation results conformed the effectiveness of the proposed approach.

      PubDate: 2018-02-05T06:51:13Z
      DOI: 10.1016/j.chemolab.2018.01.012
      Issue No: Vol. 174 (2018)
       
  • Noisy matrix completion on a novel neural network framework
    • Abstract: Publication date: 15 June 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 177
      Author(s): Samuel Mercier, Ismail Uysal
      A novel matrix completion algorithm based on the iterative application of neural networks is presented. It is shown that Bayesian regularization provides proper protection against overfitting, more so than early-stopping or a combination of both. The flexibility to increase the size of the hidden layer provides a better description of increasingly nonlinear relationships between the known and missing values in the data with a limited loss in generalization ability. The proposed neural network algorithm provides a more accurate estimation of missing values than current matrix completion algorithms based on iterative regression approaches or PCA applications for many datasets with fractions of missing values from 5 to 40%. The neural network algorithm performs particularly well on datasets where the number of observations significantly exceeds the number of features.

      PubDate: 2018-04-15T09:57:29Z
       
  • Fitness partition-based multi-objective differential evolutionary
           algorithm and its application to the sodium gluconate fermentation process
           
    • Abstract: Publication date: 15 June 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 177
      Author(s): Zhan Guo, Xuefeng Yan
      The operating conditions of the fermentation process of sodium gluconate play a key role in the quality and quantity of its production. The fermentation process is highly nonlinear and dynamic, and several objects must be considered. To implement a global and efficient optimization for the fermentation process, a fitness partition-based multi-objective differential evolutionary algorithm (FPMDE) is proposed. In the FPMDE algorithm, the information in the target space, which expresses some superiority message, is used to guide the evolutionary process. Namely, according to the fitness values, the target space is divided into some sub-region, and then some optimal directions are extracted for individuals to search for the optimal region and finally approximate the Pareto front. Experimental results on 20 benchmark functions show its advantage in convergence and diversity compared with 5 other state-of-art algorithms. Further, three objective functions for the fermentation process of sodium gluconate are proposed, and the FPMDE algorithm is applied to obtain its Pareto front; the conversion rates and utilization rate of equipment has been improved. It is shown that the FPMDE can optimize the conditions of the production of sodium gluconate effectively and efficiently.

      PubDate: 2018-04-15T09:57:29Z
       
  • Duality based interpretation of uniqueness in the trilinear decompositions
    • Abstract: Publication date: 15 June 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 177
      Author(s): Mahdiyeh Ghaffari, Hamid Abdollahi
      Application of trilinearity constraint in curve resolution of three-way data sets can play an important role in the triadic decompositions. Matrix Augmented-MCR-ALS is a two-way analysis method while its incorporation with the trilinearity constraint can provide a triadic decomposition like PARAFAC model. Trilinearity is a very strong constraint which can lead unique decomposition under mild conditions. Additionally, duality concept represents a relation between column and row spaces of bilinear data matrices. Thus, in case of a unique solution or uniqueness condition, the duality concept is a general and powerful approach for visualization. Based on the duality concept, it is necessary to define a particular hyper-plane in the dual abstract space in order to have a unique profile. It is discussed and visualized that the trilinearity constraint with the idea of parallel proportional profiles can define a particular hyper-plane in the space causing/leading to uniqueness in the dual space. Several simulated and real data sets were exemplified to show the ability of the duality concept to study and visualize the uniqueness due to the application of trilinearity constraint.
      Graphical abstract image

      PubDate: 2018-04-15T09:57:29Z
       
  • k-nearest neighbor normalized error for visualization and reconstruction
           – A new measure for data visualization performance
    • Abstract: Publication date: 15 May 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 176
      Author(s): Hiromasa Kaneko
      It can be very difficult to automatically determine the hyperparameter values of nonlinear data visualization methods. In this study, a new measure called the k-nearest neighbor normalized error for visualization and reconstruction (k3n-error) is developed to compare the visualization performance and automatically optimize the hyperparameters of nonlinear visualization methods using only unsupervised data. For a given sample, the k3n-error approach is based on the standardized errors between the Euclidean distances to neighboring samples before and after projection onto the latent space. Case studies are conducted using two numerical simulation datasets and four quantitative structure-activity/property relationship datasets. The results confirm that, for each nonlinear visualization method, samples can be mapped to the two-dimensional space while maintaining their proximity relationship from the original space by selecting the hyperparameters using the proposed k3n-error.

      PubDate: 2018-04-15T09:57:29Z
       
  • libPLS: An integrated library for partial least squares regression and
           linear discriminant analysis
    • Abstract: Publication date: 15 May 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 176
      Author(s): Hong-Dong Li, Qing-Song Xu, Yi-Zeng Liang
      Partial least squares (PLS) have gained wide applications especially in chemometrics, metabolomics/metabonomics as well as bioinformatics. Here, we present libPLS, a library that integrates not only basic PLS modeling algorithms but also advanced and/or recently developed methods on model assessment, outlier detection, and variable selection. This package is featured in a set of Model Population Analysis (MPA)-type approaches that have not been integrated into a single package yet and thus functionally complement existing toolboxes. libPLS provides an integrated platform for developing PLS regression and/or linear discriminant analysis (PLS-LDA) models. It is written in MATLAB and freely available at www.libpls.net.

      PubDate: 2018-04-15T09:57:29Z
       
  • Robust ridge regression based on self-paced learning for multivariate
           calibration
    • Abstract: Publication date: 15 May 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 176
      Author(s): Jiangtao Peng, Long Tian
      In this paper, we propose a robust ridge regression model based on self-paced learning (RR-SPL) for the high-dimensional spectroscopic data. The proposed RR-SPL model consists of a weighted least-squares loss term on all training samples, a self-paced regularizer on sample weights, and a smoothness penalty on the model parameter. Designating an explicit form of the self-paced regularizer, the weights that indicate the importance of training samples can be automatically optimized in an augmented ridge regression framework. By increasing the model age, more and more training samples from easy to hard are added into the training set to learn a mature model. As a result, the RR-SPL model can weaken the effect of outliers and obtain an accurate spectra-concentrate relation. Experimental results on simulated data sets and four real near-infrared (NIR) spectra data sets show the effectiveness of the proposed RR-SPL method in a wide range of specific prediction tasks with or without outliers.

      PubDate: 2018-04-15T09:57:29Z
       
  • A new data representation based on relative measurements and fingerprint
           patterns for the development of QSAR regression models
    • Abstract: Publication date: 15 May 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 176
      Author(s): Irene Luque Ruiz, Miguel Ángel Gómez Nieto
      Relative distance matrixes represent measurements of the structural characteristics of the molecules, having into account a reference pattern common to the whole data set considered in the development of QSAR regression models. These matrixes store relationships between the data set molecules, measuring the transformation cost between pairs of molecules and a pattern from the common fragments to the entire data set. These measurements are quite related with the activity value changes and, therefore, its use allows the building of robust QSAR regression models. In this paper, we describe the building of relative distance matrixes for the representation of two data sets with clearly different characteristics and previously used as benchmark. Applying Support Vector machine algorithms, several training models and external validation were carried out randomly selecting both sets. The results obtained with correlation coefficient greater than 0.9, low values of error and values of slope and bias close to the ideality have shown the goodness of the presented proposal, clearly improving the results obtained in the literature.

      PubDate: 2018-04-15T09:57:29Z
       
  • Study of the effect of the presence of silver nanoparticles on migration
           of bisphenol A from polycarbonate glasses into food simulants
    • Abstract: Publication date: 15 May 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 176
      Author(s): C. Reguera, S. Sanllorente, A. Herrero, L.A. Sarabia, M.C. Ortiz
      The impact that the presence of nanoparticles in food can have on the migration from food contact materials (FCMs) of substances, which occurrence in foodstuffs is regulated, is posed in this paper through a case-study. Migration of bisphenol A (BPA) from polycarbonate glasses into aqueous food simulant B (3% acetic acid, w/v) and simulant D1 (50% ethanol, v/v), both in the absence and presence of silver nanoparticles is tested. The analysis of the amount of BPA released into the food simulants is conducted by comparing population results instead of using the classical location and scatter estimates. β-content tolerance intervals are used to model the statistical distribution of BPA migrated from the polycarbonate glasses. Experimental measurements are performed by HPLC-FLD, and partial least squares regression models are then fitted to determine the concentration of BPA. The analytical procedure fulfils the trueness property. The capability of detection of the method is between 1.7 and 2.3 μg L−1 when the probabilities of false positive and false negative are fixed at 0.05. Using β-content tolerance intervals, in 90% of the specimens of a population of polycarbonate glasses, the amount of BPA migrated into simulant B in the presence of AgNPs is 13.34 μg L−1, at least twice the quantity that migrated in the absence of them.

      PubDate: 2018-04-15T09:57:29Z
       
  • Wavelet based classification of MALDI-IMS-MS spectra of serum N-Linked
           glycans from normal controls and patients diagnosed with Barrett's
           esophagus, high grade dysplasia, and esophageal adenocarcinoma
    • Abstract: Publication date: 15 May 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 176
      Author(s): B.K. Lavine, C.G. White, T. Ding, M.M. Gaye, D.E. Clemmer
      Profiling of complex biological samples (e.g., serum) using mass spectrometry continues to be an active area of research with a large and growing literature. Pattern recognition techniques can be effective methods for the analysis of complex data sets generated in these types of studies. Currently, we are investigating the discrimination of disease phenotypes associated with esophageal adenocarcinoma by analysis of single N-linked glycans using matrix assisted laser desorption ionization-ion mobility spectrometry-mass spectrometry (MALDI-IMS-MS). The glycans were extracted from sera of healthy (normal) controls (NC) and patients diagnosed with Barrett's Esophagus (BE), high grade dysplasia (HGD), and esophageal adenocarcinoma (EAC). MALDI-IMS-MS spectral images were collected in duplicate for these 58 serum samples: BE (14 individuals), HGD (7 individuals), EAC (20 individuals) and NC (17 individuals). Ion mobility distributions of N-linked glycans that possessed sufficient signal to noise in all 116 spectra were extracted from the images by box selection across a specific drift bin and m/z range corresponding to a single linked N-glycan ion. A composite ion mobility distribution profile was obtained for each image by sequentially splicing together the mobility distributions of each N-linked glycan across an arbitrary drift bin axis. Wavelet preprocessing of the composite ion mobility distribution profiles was performed using the discrete wavelet transform, which was coupled to a genetic algorithm for variable selection to identify a subset of wavelet coefficients within the data set that optimized the separation of the four classes (BE, HGD, EAC, and NC) in a plot of the two largest principal components of the wavelet transformed data. A discriminant developed from the wavelet coefficients identified by the pattern recognition GA correctly classified all ion mobility distribution profiles in the training set (45 individuals and 87 distribution profiles) and 23 of 26 blinds (13 individuals and 26 distribution profiles) in the prediction set. The proposed MALDI-IMS-MS and pattern recognition methodology has the potential to exploit molecules in serum samples that can serve as the basis of a potential method for cancer prescreening.

      PubDate: 2018-04-15T09:57:29Z
       
  • Gaussian process regression coupled with MPT-AES for quantitative
           determination of multiple elements in ginseng
    • Abstract: Publication date: 15 May 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 176
      Author(s): Yangwei Ying, Wei Jin, Yuwei Yan, Ying Mu, Qinhan Jin
      A comprehensive understanding of spectral information is the most reliable approach to identify ginseng characteristics and quality. More data is produced due to the use of a high resolution spectrometer, which in turn requires high-efficiency computational capability. To well address such problems, chemometrics plays a significant part and shows satisfactory results. In this paper, a Gaussian process regression (GPR) is proposed to couple microwave plasma torch-atomic emission spectrometry (MPT-AES) for quantitative determination of multiple elements in ginseng. GPR as a probabilistic method is capable of dealing with massive data and illustrating the outputs with probabilistic meanings. With high-dimension input variables involved, the principal component analysis (PCA) is proposed to reduce computational amount and improve variable representativeness. The results show that GPR is superior to the conventional approach compared with support vector regression (SVR) by evaluation indexes. All the work is performed based on real ginseng spectral data. Taking into account the computational amount caused by large datasets of scales, three approximation methods (subset of datasets (SD), subset of regressors (SR) and projected process (PP)) are applied to address this problem, which in a way shrink the training set with a certain similar result. It is concluded that the Gaussian process performs well in spectral analysis and has the potential for further practical applications.

      PubDate: 2018-04-15T09:57:29Z
       
  • An intelligent non-optimality self-recovery method based on reinforcement
           learning with small data in big data era
    • Abstract: Publication date: 15 May 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 176
      Author(s): Yan Qin, Chunhui Zhao, Furong Gao
      Batch processes have attracted extensive attention as a crucial manufacturing way in modern industries. Although they are well equipped with control devices, batch processes may operate at a non-optimal status because of process disturbances, equipment aging, feedstock variations, etc. As a result, the quality indices or economic benefits may be undesirable using the pre-defined normal operation conditions. And this phenomenon is called non-optimality here. Therefore, it is indispensable to timely remedy the process to its optimal status without accurate models or amounts of data. To solve this problem, this study proposes an intelligent non-optimality self-recovery method based on reinforcement learning. First, the causal variables that lead to the non-optimality are identified by developing a status-degraded Fisher discriminant analysis with consideration of sparsity. Second, on the basis of self-learning mechanism, an intelligent self-recovery method is proposed using the reinforcement learning to automatically adjust the set-points of the causal controlled variables. The self-recovery action is taken iteratively through the Actor-Critic structure with two neural networks. In this way, effective actions are taken to remedy the process to its expected status which only require small data. Finally, the efficacy of the proposed method is illustrated by both numerical case and a typical batch-type manufacturing process, i.e., the injection molding process.

      PubDate: 2018-04-15T09:57:29Z
       
  • Experimental design and solid state fermentation: A holistic approach to
           improve cultural medium for the production of fungal secondary metabolites
           
    • Abstract: Publication date: 15 May 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 176
      Author(s): Quentin Carboué, Magalie Claeys-Bruno, Isabelle Bombarda, Michelle Sergent, Jérôme Jolain, Sevastianos Roussos
      Solid-state fermentation (SSF) is a cultural method that holds tremendous potentials for the production of numerous microbial value-added compounds in various industries. As for every other process, experimental designs can provide tools to improve the product yields, diminish the production time and thus eventually decreasing the cost of the whole process. However, SSF, because of its solid nature, implies some constraints which consequently require specific tools to efficiently overcome them. The aim of this study was the improvement of the production of antioxidant naphtho-gamma-pyrones produced by Aspergillus niger G131 cultivated using SSF. Two experimental designs were presented, a combined design, taking into account two different types of variables to determine a proper solid medium, and a screening design with mixed-level factors to find solutes with significant positive effects on the output.

      PubDate: 2018-04-15T09:57:29Z
       
  • Multi-class classification method using twin support vector machines with
           multi-information for steel surface defects
    • Abstract: Publication date: 15 May 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 176
      Author(s): Maoxiang Chu, Xiaoping Liu, Rongfen Gong, Liming Liu
      Focus on the efficiency and accuracy of multi-class classification for steel surface defects, we propose a new twin support vector machines with multi-information (MTSVMs). The new MTSVMs model is based on binary twin support vector machines. It infuses three kinds of information: boundary samples information, representative samples information and feature weight information. Boundary samples information describes the distribution of samples in boundary region for defect dataset. Representative samples information provides important samples in global and local distribution. They make MTSVMs classifier have perfect execution efficiency and anti-noise performance. Feature weight information excavates strongly relevant features, which improves the accuracy of classifier. For six types of steel surface defect, the MTSVMs model is extended as a multi-class classifier. Experimental results show that our proposed multi-information algorithms have satisfactory performance. Moreover, the final comparative experiments prove that our MTSVMs model has perfect performance in efficiency and accuracy, especially for corrupted defect dataset.

      PubDate: 2018-04-15T09:57:29Z
       
  • New estimation methods for the Grubbs model
    • Abstract: Publication date: 15 May 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 176
      Author(s): Lorena Cáceres Tomaya, Mário de Castro
      Generalized fiducial inference for the precision of a measuring instrument without available replications on the observations is our main interest. In this paper, we study two new estimation procedures for the precision parameters, the product variability and the mean of the quantity of interest under the Grubbs model considering the two-instrument case. One method is based on a fiducial generalized pivotal quantity and the other one is built on the method of the generalized fiducial distribution. The behavior of the point and interval estimators is assessed numerically through Monte Carlo simulation studies. Comparisons with two existing frequentist approaches and five existing Bayesian approaches are reported. Finally, a data set is analyzed using the proposed methodology.

      PubDate: 2018-04-15T09:57:29Z
       
  • Functional extensions of Mandel's h and k statistics for outlier detection
           in interlaboratory studies
    • Abstract: Publication date: 15 May 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 176
      Author(s): Miguel Flores, Javier Tarrío-Saavedra, Rubén Fernández-Casal, Salvador Naya
      Functional data analysis (FDA) alternatives, based on the classical Mandel h and k statistics, are proposed to identify the laboratories that supply inconsistent results in interlaboratory studies (ILS). ILS is the procedure performed by a number of laboratories to test the precision of an analytical method, to measure the proficiency of laboratories in implementing an analytical procedure, to certify reference materials, and to evaluate a new experimental standard. The use of outlier tests, such as h and k Mandel statistics proposed by the ASTM E691, is crucial to assess these aims, estimating inter- and intra-laboratory data position and variability from a univariate point of view. Considering that experimental results obtained in analytical sciences are often functional, the use of FDA techniques can prevent the loss of important data information. The FDA approaches of h and k statistics are presented and point-wise obtained to deal with functional experimental data. Both functional statistics are estimated for each laboratory, their functional critical limits are obtained by bootstrap resampling, and new FDA versions of h and k graphics are presented. Real and synthetic thermogravimetric data are utilized to assess the good performance of the proposed FDA h and k statistics and their advantages with respect to the univariate approach.

      PubDate: 2018-04-15T09:57:29Z
       
  • Bayesian estimation of the analyte concentrations using the sensor
           responses and the design optimization of a sensor system
    • Abstract: Publication date: 15 May 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 176
      Author(s): David Han, Kevin Johnson
      Using an array of sensors with well calibrated but different tuning curves, it is possible to appreciate a wide range of stimuli. In this work, we first revisit the statistical estimation of the stimuli concentrations given the responses of a sensor array, discussed in Sanchez-Montanes & Pearce [18]. Since it is not a typical regression problem, the Bayesian concept is adopted to develop an estimation method by elucidating the dynamic and uncertain nature of the environment-dependent stimuli with a proper choice of the probability distribution. Other studies confirm that the proposed method can demonstrate a superior performance in terms of accuracy and precision when compared to the popular frequentist methods in addition to the theoretical soundness it enjoys as a statistical estimation problem. Under the proposed framework, the design optimization of an artificial sensory system is also formulated using the expected Bayes risk as an objective function to minimize. The same approach may be equally applied to any sensory system in order to optimize its performance within a population of sensors. Finally, illustrative examples are provided to describe how the proposed method can be applied for the optimal configuration of a sensory system for a given sensing task.

      PubDate: 2018-04-15T09:57:29Z
       
  • A novel multivariate calibration method based on variable adaptive
           boosting partial least squares algorithm
    • Abstract: Publication date: 15 May 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 176
      Author(s): Pao Li, Guorong Du, Yanjun Ma, Jun Zhou, Liwen Jiang
      With the aid of chemometric methods, near-infrared (NIR) spectroscopy is widely applied to the analysis of complex samples. Boosting partial least squares (BPLS) is one of the important approaches and has many applications to improve the predictive stability and accuracy of the NIR models. However, accurate calculation results are difficult to obtain due to the redundant variables that contribute more collinearity and noise than relevant information to models. In this work, an algorithm named as variable adaptive boosting partial least squares (VABPLS) was proposed to get higher robustness models and enhance the prediction ability. The theory and calculation of VABPLS are just similar with BPLS algorithm, but a variable adaptive strategy based on adaptive reweighted sampling (ARS) theory is fused into the algorithm to improve the accuracy, instead of simply adding. Simultaneous weighting of samples and variables in the boosting series is found to be more effective than the single weighting. The performance of VABPLS is tested with the NIR spectral datasets of corn and tobacco leaf samples. Results show that the advantages of VABPLS are many, such as enhanced accuracy, less parameters and easy to implement, compared with the traditional approaches.
      Graphical abstract image

      PubDate: 2018-04-15T09:57:29Z
       
  • Partial least squares fusing unsupervised learning
    • Abstract: Publication date: 15 April 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 175
      Author(s): Jae Keun Yoo
      In this paper, partial least squares to fuse unsupervised learning, called fused clustered least squares (FCLS), is proposed. As an unsupervised method, the K-means clustering algorithm is adopted, and it clusters either the original predictors or its principal components. This unsupervised learning procedure has a function to discover unknown structures of the predictors, and this information is utilized in their further reduction. Within each cluster, the covariance of the response and the predictors is computed and successively projected onto the covariance matrix of the predictors. This is called clustered least squares. Then we fuse all clustered least squares from the various numbers of clusters. The FCLS is basically implemented by combining supervised and unsupervised statistical methods, and it overcomes the deficits that the ordinary least squares, including its popular variation of partial least squares, have in practice. Numerical studies support the theory, and its application to near infrared spectroscopy data confirms the potential advantage of FCLS in practice.

      PubDate: 2018-04-15T09:57:29Z
       
  • Comparing multiple statistical methods for inverse prediction in nuclear
           forensics applications
    • Abstract: Publication date: 15 April 2018
      Source:Chemometrics and Intelligent Laboratory Systems, Volume 175
      Author(s): John R. Lewis, Adah Zhang, Christine M. Anderson-Cook
      Forensic science seeks to predict source characteristics using measured observables. Statistically, this objective can be thought of as an inverse problem where interest is in the unknown source characteristics or factors (X) of some underlying causal model producing the observables or responses ( Y = g ( X ) + e r r o r ). This paper reviews several statistical methods for use in inverse problems and demonstrates that comparing results from multiple methods can be used to assess predictive capability. Motivation for assessing inverse predictions comes from the desired application to historical and future experiments involving nuclear material production for forensics research in which inverse predictions, along with an assessment of predictive capability, are desired. Four methods are reviewed in this article. Two are forward modeling methods and two are direct inverse modeling methods. Forward modeling involves building a forward casual model of the responses (Y) as a function of the source characteristics (X) using content knowledge and data ideally obtained from a well-designed experiment. The model is then inverted to produce estimates of X given a new set of responses. Direct inverse modeling involves building prediction models of the source characteristics ( X ) as a function of the responses (Y) – subverting estimation of any underlying causal relationship. Through use of simulations and a data set from an actual plutonium production experiment, it is shown that agreement of predictions across methods is an indication of strong predictive capability, whereas disagreement indicates the current data are not conducive to making good predictions.

      PubDate: 2018-04-15T09:57:29Z
       
  • A Modified Moving Window dynamic PCA with Fuzzy Logic Filter and
           application to fault detection
    • Abstract: Publication date: Available online 13 April 2018
      Source:Chemometrics and Intelligent Laboratory Systems
      Author(s): Mustapha Ammiche, Abdelmalek Kouadri, Abderazak Bensmail
      Principal Component Analysis (PCA) model is constructed from measured data and used to monitor new testing samples. In fact, the statistical independency assumption between observations is true only for long sampling intervals. Nowadays, industrial systems are sophisticated and fast for which this assumption becomes no longer valid and the current observation becomes highly dependent on the past observations. In another hand, Dynamic PCA (DPCA) is a PCA extension to deal with the aforementioned problem, but monitoring process using this method with fixed control limits showed a high False Alarms Rate (FAR), high Missed Detection Rate (MDR) and long Detection Time Delay (DTD). In this paper, a Modified Moving Window DPCA (MMW-DPCA) with Fuzzy Logic Filter (FLF) is proposed to address the above issue. The developed monitoring scheme continually updates control limits throughout an obtained DPCA-based model. The adaptive thresholds are established by moving a fixed size window over the data. The dynamic behavior of the data is handled by DPCA, whereas the sensitivity enhancement and the FAR reduction are handled by the developed adaptive thresholds for which the FLF is employed to ensure robustness to outliers and noise without affecting the fault detection performances. The proposed technique has been tested on Tennessee Eastman Process (TEP). It has been compared to other well-known fault detection methods. The obtained results demonstrate that the MMW-DPCA with FLF detects different types of faults with high accuracy and in a short time delay. The experimental application of the MMW-DPCA with FLF has been carried out on cement rotary kiln. The obtained results illustrate that the proposed method has successfully detected a real fault.

      PubDate: 2018-04-15T09:57:29Z
       
  • A variable selection method for multiclass classification problems using
           two-class ROC analysis
    • Abstract: Publication date: Available online 12 April 2018
      Source:Chemometrics and Intelligent Laboratory Systems
      Author(s): Miguel de Figueiredo, Christophe B.Y. Cordella, Delphine Jouan-Rimbaud Bouveresse, Xavier Archer, Jean-Marc Bégué, Douglas N. Rutledge
      Modern procedures in analytical chemistry generate enormous amounts of data, which must be processed and interpreted. The treatment of such high-dimensional datasets often necessitates the prior selection of a reduced number of variables in order to extract knowledge about the system under study and to maximize the predictability of the models built. Therefore, this article describes a variable selection method for multiclass classification problems using two-class ROC analysis and its associated area under the ROC curve as a variable selection criterion. The variable selection method has been successfully applied to two datasets. For comparison purposes, two other variable selection methods, ReliefF and mRMR, were used and double cross-validation PLS-DA was applied using: (1) all variables and (2) the variables selected using the three methods. It has been demonstrated that correct variable selection can substantially reduce the dimensionality of the datasets, while maximizing the predictability of the models.

      PubDate: 2018-04-15T09:57:29Z
       
  • OCPMDM: Online computation platform for materials data mining
    • Abstract: Publication date: Available online 11 April 2018
      Source:Chemometrics and Intelligent Laboratory Systems
      Author(s): Qing Zhang, Dongping Chang, Xiuyun Zhai, Wencong Lu
      With the rapid development of the Materials Genome Initiative (MGI), scientists and engineers are confronted with the need to conduct sophisticated data analytics in modeling the behaviors of materials. Nowadays, it is inconvenient for material researchers to carry out materials data mining work without an efficient platform for materials machine learning. So, it is meaningful to develop an online platform for material researchers in urgent need of using machine learning techniques by themselves. The typical case study is given to demonstrate the applications of the online computation platform for material data mining (OCPMDM) in our lab: The quantitative structure property relationship (QSPR) model for rapid prediction of Curie temperature of perovskite material can be applied to screen out perovskite candidates with higher Curie temperature than those of training dataset collected from references, efficiently. Material data mining tasks can be implemented via the OCPMDM, which provides powerful tools for material researchers in machine learning-assisted materials design and optimization. The URL of OCPMDM is http://materialdata.shu.edu.cn.
      Graphical abstract image

      PubDate: 2018-04-15T09:57:29Z
       
  • Support vector machine with truncated pinball loss and its application in
           pattern recognition
    • Abstract: Publication date: Available online 10 April 2018
      Source:Chemometrics and Intelligent Laboratory Systems
      Author(s): Liming Yang, Hongwei Dong
      Support vector machine(SVM) with pinball loss(PINSVM) has been recently proposed and shown its advantages in pattern recognition. In this paper, we present a robust bounded loss function (called L t -loss) that truncates pinball loss function. Then a novel robust SVM formulation with L t -loss(called TPINSVM) is proposed to enhance noise robustness. Moreover, we demonstrate that the proposed TPINSVM satisfies Bayes rule and it has a certain sparseness. However, the non-convexity of the proposed TPINSVM makes it difficult to optimize. We develop a continuous optimization method, DC(difference of convex functions) programming method, to solve the proposed TPINSVM. The resulting DC optimization algorithm converges finitely. Furthermore, the proposed TPINSVM is directly applied to recognize the purity of hybrid maize seeds using near-infrared spectral data. Experiments show that the proposed method achieves better performance than the traditional methods in most spectral regions. Meanwhile we simulate the proposed TPINSVM in benchmark datasets in different situations. In noiseless setting, the proposed TPINSVM either improves or shows no significant difference in generalization compared to the traditional approaches. While in noise situations, TPINSVM improves generalization in most cases.

      PubDate: 2018-04-15T09:57:29Z
       
  • Comparison of different image analysis algorithms on MRI to predict
           physico-chemical and sensory attributes of loin
    • Abstract: Publication date: Available online 9 April 2018
      Source:Chemometrics and Intelligent Laboratory Systems
      Author(s): Daniel Caballero, Andrés Caro, Anders B. Dahl, Bjarne K. ErsbØll, José Manuel Amigo, Trinidad Pérez-Palacios, Teresa Antequera
      Computer vision algorithms on MRI have been presented as an alternative to destructive methods to determine the quality traits of meat products. Since, MRI is non-destructive, non-ionizing and innocuous methods. The use of fractals to analyze MRI could be another possibility for this purpose. In this paper, a new fractal algorithm is developed, to obtain features from MRI based on fractal characteristics. This algorithm is called OPFTA (One Point Fractal Texture Algorithm). Three fractal algorithms (Classical Fractal Algorithm –CFA-, Fractal Texture Algorithm –FTA- and OPFTA) and three classical texture algorithms (Grey level co-occurrence matrix –GLCM-, Grey level run length matrix –GLRLM- and Neighbouring grey level dependence matrix –NGLDM-) were tested in this study. The results obtained by means of these computer vision algorithms were correlated to the results obtained by means of physico-chemical and sensory analysis. CFA reached low relationship for the quality parameters of loins, the remaining algorithms achieved correlation coefficients higher than 0.5 noting OPFTA that reached the highest correlation coefficients in all cases except for the L* coordinate color that GLCM obtained the highest correlation coefficient. These high correlation coefficients confirm the new algorithm as an alternative to the other computer vision approaches in order to compute the physico-chemical and sensory parameters of meat products in a non-destructive and efficient way.

      PubDate: 2018-04-15T09:57:29Z
       
  • In-situ particle segmentation approach based on average background
           modeling and graph-cut for the monitoring of l-glutamic acid
           crystallization
    • Abstract: Publication date: Available online 9 April 2018
      Source:Chemometrics and Intelligent Laboratory Systems
      Author(s): Zhi M. Lu, Fan C. Zhu, Xue Y. Gao, Bing C. Chen, Tao Liu, Zhen G. Gao
      The crystal morphology often represents a critically important property for the reason that different morphological characteristics such as crystal shape and size distribution will directly affect crystal growth and the quality of final products. However, minor changes in reaction conditions can have a significant impact on crystal morphology. As a consequence, crystallization processes demand an on-line technique for real-time monitoring of crystal morphology. Image-based monitoring methods show a great potential for real-time monitoring, and performing particle segmentation accurately becomes a key issue in the particle image analysis. To avoid the influences of droplets and to eliminate the particle shadow of particle images obtained by invasive imaging system, an effective approach is proposed for particle segmentation based on combing the background difference method and the graph-cut based local threshold method. Through building the background model and performing subtraction between particle image and background model, droplets could be eliminated effectively. Then the local threshold method is performed further to eliminate the influences of particle shadow. Several particle images are used to demonstrate the efficiency and accuracy of the proposed method. In order to further evaluate the real-time performance of the proposed method, comparison experiments between the proposed method and other advanced or classic algorithms are presented. Experimental results show that the proposed method is efficient and has an impressive real-time performance.

      PubDate: 2018-04-15T09:57:29Z
       
  • Detecting and classifying minor bruised potato based on hyperspectral
           imaging
    • Abstract: Publication date: Available online 4 April 2018
      Source:Chemometrics and Intelligent Laboratory Systems
      Author(s): Dandan Ye, Laijun Sun, Wenyi Tan, Wenkai Che, Mingcan Yang
      Potato with minor bruise is difficult to be detected in the process of damage identification and is perishable in storage, thus leading to a serious problem of food safety and economic issue. In view of the complexity of bruise detection for potatoes, a nondestructive detection method, based on hyperspectral imaging technique, was proposed in this study. All samples, including healthy potatoes and bruised potatoes belonging to 3 different levels, were taken as experiment objects. First of all, the background in every hyperspectral image was removed by masking aiming at acquiring the average spectra of each potato. Then, Savitzky-Golay smoothing, first derivative, second derivative, standard normal variate and its combinatorial methods were applied to pre-process spectral data, respectively, and the grid search algorithm was applied to optimize modeling parameters. Confirm that the standard normal variate pre-processing technique reinforced the model performance at utmost, and the identification accuracy of bruised samples reached 90.63%. In addition, given the interference of redundant information, the optimized simulated annealing algorithm based on correlation coefficient algorithm was applied to reduce the dimension of the spectral data, which promoted the identification accuracy of bruised samples to 96.88%. Furthermore, the bruise levels of samples were classified using above methods, and the higher classification accuracy of 100% was achieved. The results indicated that potato with minor bruise could be identified accurately and effectively and the bruise levels could be classified by the hyperspectral imaging technique, which provided an idea for on-line and non-destructive testing of potato.

      PubDate: 2018-04-15T09:57:29Z
       
  • Birnbaum-Saunders spatial regression models: Diagnostics and application
           to chemical data
    • Abstract: Publication date: Available online 29 March 2018
      Source:Chemometrics and Intelligent Laboratory Systems
      Author(s): Fabiana Garcia-Papani, Víctor Leiva, Miguel A. Uribe-Opazo, Robert G. Aykroyd
      Geostatistical modelling is widely used to describe data with spatial dependence structure. Such modelling often assumes a Gaussian distribution, an assumption which is frequently violated due to the asymmetric nature of variables in diverse applications. The Birnbaum-Saunders distribution is asymmetrical and has several appealing properties, including theoretical arguments for describing chemical data. This work examines a Birnbaum-Saunders spatial regression model and derives global and local diagnostic methods to assess the influence of atypical observations on the maximum likelihood estimates of its parameters. Modelling and diagnostic methods are then applied to experimental data describing the spatial distribution of magnesium and calcium in the soil in the Parana state of Brazil. This application shows the importance of such a diagnostic analysis in spatial modelling with chemical data.

      PubDate: 2018-04-15T09:57:29Z
       
  • An online sequential multiple hidden layers extreme learning machine
           method with forgetting mechanism
    • Authors: Dong Xiao; Beijing Li; Shengyong Zhang
      Abstract: Publication date: Available online 7 February 2018
      Source:Chemometrics and Intelligent Laboratory Systems
      Author(s): Dong Xiao, Beijing Li, Shengyong Zhang
      In many practical applications, training data are presented one-by-one or chuck-by-chuck and also have the property of timeliness very frequently. The ensemble of an online sequential extreme learning machine (EOS-ELM) can learn data one-by-one or chunk-by-chunk with fixed or varying chunk size. The online sequential extreme learning machine with forgetting mechanism (FOS-ELM) can learn data with the property of timeliness. In many practical applications, such as stock forecasting or weather forecasting, the training accuracy can be improved by discarding the outdated data and reducing the influence on later training processes. Since the real-time variations of data are accompanied by a series of unavoidable noise signals, to make the training output closer to the actual output, an online sequential multiple hidden layers extreme learning machine with forgetting mechanism (FOS-MELM) is proposed in this paper. The proposed FOS-MELM can retain the advantages of FOS-ELM, eliminate the influence of unavoidable noise and improve the prediction accuracy. In this work, experiments have been completed on chemical (styrene) data. The experimental results show that FOS-MELM has high accuracy, better stability and better short-term prediction than FOS-ELM.

      PubDate: 2018-02-26T08:05:50Z
      DOI: 10.1016/j.chemolab.2018.01.014
       
 
 
JournalTOCs
School of Mathematical and Computer Sciences
Heriot-Watt University
Edinburgh, EH14 4AS, UK
Email: journaltocs@hw.ac.uk
Tel: +00 44 (0)131 4513762
Fax: +00 44 (0)131 4513327
 
Home (Search)
Subjects A-Z
Publishers A-Z
Customise
APIs
Your IP address: 54.166.172.33
 
About JournalTOCs
API
Help
News (blog, publications)
JournalTOCs on Twitter   JournalTOCs on Facebook

JournalTOCs © 2009-