Article
Automation & Control Systems
Ze-yin Dong, Zi-hang Yu
Summary: Baseline drift problem commonly exists in spectral analysis due to detector nonlinearities and temperature variations. To address this, we propose a novel and high-precise algorithm called baseline estimation using morphological and iterative local extremum (MILE). Our algorithm first obtains all local extrema of the measured spectrum and estimates a coarse baseline using the PCHIP interpolation method. It then iteratively updates the baseline by finding local extrema of the coarse baseline and their adjacent data. By subtracting the estimated baseline, the spectrum is corrected with high precision.
CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS
(2023)
Article
Computer Science, Information Systems
Hao Shi, Ruixia Liu, Changfang Chen, Minglei Shu, Yinglong Wang
Summary: The article introduces a sparse optimization method for ECG signal processing, which combines low-pass filter technique and takes into account the group sparse characteristics of ECG signals for denoising and baseline estimation. Through data comparison and analysis, the proposed method shows lower root mean square error and higher signal-to-noise ratio improvement compared to other methods.
Article
Engineering, Electrical & Electronic
Yi Liao, Weiguo Huang, Changqing Shen, Zhongkui Zhu, Jianping Xuan, Lingfeng Mao
Summary: This paper introduces a multivariate non-convex logarithm penalty based on generalized infimal convolution smoothing for vibration signal denoising and compound fault diagnosis in gearboxes. By ensuring the global minimum of the overall cost function and deriving a convexity condition for the non-convex penalty, an optimal sparse solution can be calculated using a convex algorithm. The proposed method effectively induces sparsity and enhances estimation accuracy of signal components compared to classical convex L1 norm and newly developed non-convex generalized minimax-concave penalties.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2021)
Article
Chemistry, Analytical
Martial Tazifor Tchantcho, Egon Zimmermann, Johan Alexander Huisman, Markus Dick, Achim Mester, Stefan van Waasen, Giovanni Betta
Summary: The article presents a temperature drift correction method for electromagnetic induction (EMI) systems using two low-pass filters to account for delayed thermal variations. Experimental results show that this method effectively reduces the influence of temperature drift and improves measurement accuracy.
Article
Engineering, Electrical & Electronic
Yuqiang Li, Xinjie Wang, Huijing Yu, Wenli Du
Summary: In this article, a pattern-coupled learning framework considering the local coupling property is proposed for pure spectrum fitting and baseline correction. By characterizing the local sparse structure corresponding to characteristic peaks with wavelength coupling, the proposed method achieves accurate baseline estimation without prior knowledge of spectral structure. The proposed method shows improved performance compared to state-of-the-art methods in three measured NIR datasets, with reduced RMSE and increased R-2 values.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2023)
Article
Engineering, Mechanical
Ziwei Zhang, Weiguo Huang, Yi Liao, Zeshu Song, Juanjuan Shi, Xingxing Jiang, Changqing Shen, Zhongkui Zhu
Summary: The study introduces a new non-convex penalty, the generalized logarithm (G-log) penalty, to enhance sparsity and reduce noise disturbance, thereby improving the accuracy of bearing fault diagnosis.
MECHANICAL SYSTEMS AND SIGNAL PROCESSING
(2022)
Article
Computer Science, Artificial Intelligence
Zhiqiang Zhang, Qingyu Yang
Summary: The article introduces an intelligent fault diagnosis method based on reconstruction sparse filtering (RSF), which extracts diverse features by constraining the basis vectors, enabling precise description of the health conditions of rotating machinery and achieving significant performance improvement.
APPLIED SOFT COMPUTING
(2022)
Article
Chemistry, Analytical
Zengshun Chen, Jun Fu, Yanjian Peng, Tuanhai Chen, LiKai Zhang, Chenfeng Yuan
Summary: The paper proposed a deep neural network model based on empirical mode decomposition to solve baseline drift in displacement data, achieving real displacement time history by removing the drifting trend. In shaking table tests, the proposed EMD-DNN method showed better baseline correction effect compared to traditional methods.
Article
Engineering, Electrical & Electronic
Zhibin Zhao, Shibin Wang, David Wong, Chuang Sun, Ruqiang Yan, Xuefeng Chen
Summary: This paper proposes a robust enhanced trend filtering algorithm called RobustETF, which can extract trends in the presence of various types of non-Gaussian noise or outliers. The algorithm utilizes an extended EM algorithm to solve the resulting non-convex optimization problem.
Article
Chemistry, Multidisciplinary
Bin Zhang, Liuliu Wang, Shuang Li, Futai Xie, Lideng Wei
Summary: This study introduces a multi-look compressive sensing method and a multi-look compressive sensing method based on separable approximate sparse reconstruction, which enhance the separation ability of ground scatterers and the stability of height estimation in InSAR by increasing the dimension of observation vectors, reducing noise level, and improving the sparse reconstruction capability of compressive sensing.
APPLIED SCIENCES-BASEL
(2022)
Article
Engineering, Geological
Zhizhou He, Yanjie Zhang, Haishen Wang, Peng Pan
Summary: This paper proposes a time-domain baseline correction method for acceleration time history based on target residual displacement and convex optimization. The method aims to accurately reproduce the displacement time history from the baseline corrected acceleration time history records. Compared to traditional methods, this proposed method can automatically determine the occurrences, timings, and amplitudes of baseline drifts without manual intervention. Numerical simulations and shaking table tests validate the effectiveness and feasibility of the method in reproducing displacement time histories with high accuracy.
SOIL DYNAMICS AND EARTHQUAKE ENGINEERING
(2023)
Article
Chemistry, Analytical
Xiaoshan Li, Xiaojun Tang, Bin Wang, Youshui Lu, Houqing Chen
Summary: Baseline drift is an important issue in spectral analysis, and most existing methods for baseline correction are not effective in high noise, complex baselines, and overlapping peaks situations. To address these challenges, an adaptive extended Gaussian peak derivative reweighted penalised least squares (agdPLS) method was proposed, which achieved more accurate baseline estimation. Experimental results showed that agdPLS outperformed other methods and was computationally efficient, making it effective for baseline correction in spectra with high noise, complex baselines, and overlapping peaks.
ANALYTICAL METHODS
(2023)
Article
Spectroscopy
Chen Su-yi, Li Hao-ran, Dai Ji-sheng
Summary: This paper introduces a baseline correction method based on sparse Bayesian learning, and proposes to adaptively learn the block-sparse structure by introducing a coupling pattern model, thereby improving the performance of baseline correction. Simulation and experimental results verify the superiority of this method.
SPECTROSCOPY AND SPECTRAL ANALYSIS
(2022)
Article
Computer Science, Artificial Intelligence
Zhenni Li, Chao Wan, Benying Tan, Zuyuan Yang, Shengli Xie
Summary: This paper presents an efficient DC-based algorithm for handling dictionary learning with nonconvex penalty, by decomposing the problem into subproblems related to single-vector factors and using alternating optimization and DC technology to address the nonconvexity. The proposed algorithm outperforms state-of-the-art algorithms in numerical experiments with synthetic and real-world data, showcasing better performance in handling different sparsity-including constraints.
Article
Geochemistry & Geophysics
Zahra Sadeghi, Alireza Goudarzi, Parvaneh Pakmanesh, Sadegh Moghaddam
Summary: In this research, sparsity is employed for the first time in seismic data processing to correct residual statics. The BEADS method, based on modeling the sparsity of the series of seismogram peaks and its derivatives, estimates seismogram data and residual static correction more accurately, and is validated by comparison with traditional methods.
JOURNAL OF SEISMIC EXPLORATION
(2022)
Article
Engineering, Electrical & Electronic
Yun Kong, Tianyang Wang, Fulei Chu, Zhipeng Feng, Ivan Selesnick
Summary: The novel DDL-SC framework proposed for data-driven machinery fault diagnosis efficiently learns a discriminative dictionary and a linear classifier, enhancing recognition accuracy. By introducing discriminative sparse code error and connecting a binary hard thresholding operator with classifier predictions, DDL-SC outperforms state-of-the-art methods in recognition performance for machinery fault diagnosis.
IEEE SENSORS JOURNAL
(2021)
Article
Engineering, Electrical & Electronic
Arthur Marmin, Marc Castella, Jean-Christophe Pesquet, Laurent Duval
Summary: This method proposes a new approach to reconstruct sparse signals degraded by nonlinear distortion, using nonconvex minimization. Unlike previous works, this method aims for global solutions and relies on Lasserre relaxation of polynomial optimization. Additionally, it includes the case of piecewise rational functions for handling a wider range of nonconvex relaxations.
Article
Microbiology
Remi Hocq, Surabhi Jagtap, Magali Boutard, Andrew C. Tolonen, Laurent Duval, Aurelie Pirayre, Nicolas Lopes Ferreira, Francois Wasels
Summary: This study used Capp-Switch sequencing to determine transcription start site (TSS) positions in three model solventogenic clostridia genomes. The distributions of these sites were compared, revealing both similarities and differences between the strains. The study also found that promoter structure is generally poorly conserved between C. acetobutylicum and C. beijerinckii.
MICROBIOLOGY SPECTRUM
(2022)
Article
Biochemical Research Methods
Surabhi Jagtap, Abdulkadir Celikkanat, Aurelie Pirayre, Frederique Bidard, Laurent Duval, Fragkiskos D. Malliaros
Summary: This study proposes a novel random walk-based matrix factorization method called BraneMF for learning node representation in a multilayer network and its application to omics data integration. The applicability of learned features for essential multi-omics inference tasks is demonstrated using PPI networks of Saccharomyces cerevisiae, and BraneMF outperforms baseline methods in various downstream tasks.
Article
Biochemical Research Methods
Surabhi Jagtap, Aurelie Pirayre, Frederique Bidard, Laurent Duval, Fragkiskos D. Malliaros
Summary: BRANEnet is a novel multi-omics integration framework proposed in this study, aiming to efficiently analyze the regulatory aspects of multilayered processes in organisms. By applying to transcriptomics and metabolomics data of Saccharomyces cerevisiae, BRANEnet demonstrates superior performance in tasks such as transcription factor prediction, integrated omics network inference, and module identification.
BMC BIOINFORMATICS
(2022)
Article
Automation & Control Systems
Louna Alsouki, Laurent Duval, Clement Marteau, Rami El Haddad, Francois Wahl
Summary: Relating variables X to response y is important in chemometrics. Qualitative interpretation can enhance quantitative prediction by identifying influential features. Projections (e.g. PLS) and variable selections (e.g. lasso) are used for dimension reduction in high-dimensional problems. Dual-sPLS, a variant of PLS1, provides a balance between accurate prediction and efficient interpretation through penalizations inspired by classical regression methods and the dual norm notion. It performs favorably compared to similar regression methods on simulated and real chemical data.
CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS
(2023)
Article
Engineering, Electrical & Electronic
Paul Zheng, Emilie Chouzenoux, Laurent Duval
Summary: In this paper, we propose a novel method called PENDANTSS for joint trend removal and blind deconvolution of sparse peak-like signals. By combining a parsimonious prior with the assumption that smooth trend and noise can be separated to some extent by low-pass filtering, our method achieves both convergence and efficiency through the combination of generalized quasi-norm ratio sparse penalties and the BEADS ternary-assisted source separation algorithm. The proposed method outperforms comparable methods when applied to typically peaked analytical chemistry signals. Reproducible code is available.
IEEE SIGNAL PROCESSING LETTERS
(2023)
Article
Engineering, Electrical & Electronic
Arman Kheirati Roonizi, Ivan W. Selesnick
Summary: This paper proposes a Kalman filter framework that combines conventional LTI filtering and TV denoising for signal denoising. The approach treats the desired signal as a mixture of two distinct components and formulates an iterative Kalman filter/smoother approach for signal estimation.
IEEE TRANSACTIONS ON SIGNAL PROCESSING
(2022)
Article
Computer Science, Artificial Intelligence
Mojtaba Lashgari, Hossein Rabbani, Gerlind Plonka, Ivan Selesnick
Summary: This paper presents a new approach for reconstructing disconnected digital lines based on a constrained regularization model to ensure connectivity in the image plane. The proposed technique improves intersection detection and has potential for fast binary image inpainting.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2022)
Article
Engineering, Electrical & Electronic
Abdullah H. Al-Shabili, Ivan Selesnick
Summary: This paper interprets the inner workings of shallow CNNs in the task of positive sparse signal denoising, identifies common structures among trained CNNs, and shows that the learned CNN denoisers can be interpreted as a nonlinear locally-adaptive thresholding procedure. It also introduces constrained CNN denoisers that demonstrate no loss in performance despite having fewer trainable parameters.
IEEE SIGNAL PROCESSING LETTERS
(2022)
Proceedings Paper
Acoustics
Mouna Gharbi, Emilie Chouzenoux, Jean-Christophe Pesquet, Laurent Duval
Summary: In this work, a combination of traditional iterative MM algorithms and deep learning is proposed to achieve a fast and accurate sparse spectroscopy signal restoration method. By unfolding MM algorithms onto deep network architectures, the restoration of a large dataset of realistic mass spectrometry data is demonstrated, overcoming the limitations of traditional methods.
29TH EUROPEAN SIGNAL PROCESSING CONFERENCE (EUSIPCO 2021)
(2021)
Proceedings Paper
Acoustics
Surabhi Jagtap, Abdulkadir Celikkanat, Aurelic Piravre, Frederiuue Bidard, Laurent Duval, Fragkiskos D. Malliaros
Summary: The advancement of omics technologies has led to the generation of vast and high-dimensional omics data, which can be studied using network representation and embedding learning methods to understand the relationships between biological entities. This study introduces a methodology based on exponential family distributions to model the interactions among biological entities, specifically for inferring gene regulatory networks.
29TH EUROPEAN SIGNAL PROCESSING CONFERENCE (EUSIPCO 2021)
(2021)
Proceedings Paper
Acoustics
Elie Leroy, Arthur Marmin, Marc Castella, Laurent Duval
Summary: This paper proposes a new nonlinear activation function that depends on both current and past inputs through a hysteresis effect. The study shows that choosing nonlinearity in the class of rational functions for weight identification is equivalent to solving a rational optimization problem in neural networks. Simulations demonstrate that hysteresis nonlinear activation functions cannot be approximated by traditional ones, illustrating the effectiveness of the weight identification method proposed in this study.
2021 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP 2021)
(2021)
Article
Engineering, Electrical & Electronic
Abdullah H. Al-Shabili, Yining Feng, Ivan Selesnick
Summary: The sharpening sparse regularizers (SSR) framework is proposed as a middle ground to design non-separable non-convex penalties that induce sparsity more effectively than convex penalties, without sacrificing cost function convexity. By exploiting the data fidelity relative strong convexity, the overall problem convexity is preserved.
IEEE OPEN JOURNAL OF SIGNAL PROCESSING
(2021)
Article
Engineering, Electrical & Electronic
Seisuke Kyochi, Shunsuke Ono, Ivan Selesnick
Summary: This paper introduces an epigraphical relaxation (ERx) technique for non-proximable mixed norm minimization, providing a method to handle a wide range of non-proximable mixed norms in optimization without changing the minimizer of the original problem. Additionally, novel regularizers based on ERx are developed and shown to be effective through experiments.
IEEE TRANSACTIONS ON SIGNAL PROCESSING
(2021)
Article
Automation & Control Systems
Haifei Peng, Jian Long, Cheng Huang, Shibo Wei, Zhencheng Ye
Summary: This paper proposes a novel multi-modal hybrid modeling strategy (GMVAE-STA) that can effectively extract deep multi-modal representations and complex spatial and temporal relationships, and applies it to industrial process prediction.
CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS
(2024)
