Article
Computer Science, Interdisciplinary Applications
Bo Lin, Qianxiao Li, Weiqing Ren
Summary: This paper proposes a data-driven method for inferring dynamics and learning the invariant distribution from trajectory data without prior knowledge of the equations. The method combines maximum likelihood estimation and a decomposition of the force field suggested by the Fokker-Planck equation. It is demonstrated with four numerical examples.
JOURNAL OF COMPUTATIONAL PHYSICS
(2023)
Article
Engineering, Electrical & Electronic
Julien Lesouple, Barbara Pilastre, Yoann Alunann, Jean-Yves Tourneret
Summary: This study introduces a new expectation maximization (EM) algorithm for solving the problem of circle, sphere, and hypersphere fitting, by incorporating random latent vectors that follow a priori independent von Mises-Fisher distributions defined on the hypersphere. The statistical model leads to a complete data likelihood whose expected value has a Von Mises-Fisher distribution, allowing the inference problem to be addressed with a simple EM algorithm. Evaluation of the resulting hypersphere fitting algorithm is conducted for circle and sphere fitting to assess its performance.
IEEE SIGNAL PROCESSING LETTERS
(2021)
Article
Acoustics
He Wang, Ting Zhang, Lei Cheng, Hangfang Zhao
Summary: This paper proposes a 2D active target localization method based on expectation-maximization-vertical-time-record (EMVTR) approach, which can achieve high-resolution localization effectively in snapshot-deficient scenarios by reconstructing the covariance matrix. The proposed approach reduces the temporal correlation of echoes through short-time Fourier transform, achieving robust beam-time localization in mild reverberation. Multi-frequency EMVTR is derived from the single-frequency case to improve weak echo localization.
JOURNAL OF THE ACOUSTICAL SOCIETY OF AMERICA
(2023)
Article
Computer Science, Interdisciplinary Applications
Pratyush Kumar, James B. Rawlings
Summary: This paper proposes a novel model-free Q-learning approach to estimate linear feedback controllers from noisy process data. The approach is modified to handle unknown noise covariances and is applied to estimate feedback controllers for linear systems with both process and measurement noise. A model-based approach is also presented for comparison.
COMPUTERS & CHEMICAL ENGINEERING
(2023)
Article
Automation & Control Systems
Mohsen Kompany-Zareh, Bahram Dalvand, Peter D. Wentzell, Mahsa Dadashi, Mohammad Taghi Baharifard
Summary: This study applies multivariate chemometric methods, including PCA, MLFA, and MCR-ALS, to analyze a data set and estimate the measurement error structure without replicates. A smartphone-based diffuse reflectance spectrophotometer (smartDRS) is developed to acquire spectral information from synthesized pigment samples. The effects of different factors on the color of synthesized CdS pigment are investigated, and MLFA-MCR-ALS is used to confirm the contribution of different forms of CdS. The experimental design results show significant factors affecting the CdS color, and MLFA analysis reveals heteroscedastic noise in the reflectance data.
CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS
(2023)
Article
Chemistry, Analytical
Jamile Mohammad Jafari, Roma Tauler, Hamid Abdollahi
Summary: The paper introduces a new method - Balanced Scaling (BS) method, combined with Multivariate Curve Resolution Alternating Least Squares (BSMCR-ALS) method, for analyzing data sets with heteroscedastic noise, showing good performance especially in environmental data analysis. Comparisons with other methods revealed that BS-MCR-ALS and MLPCA-MCR-ALS solutions were very similar in performance.
MICROCHEMICAL JOURNAL
(2021)
Article
Biochemical Research Methods
Ariel A. Hippen, Matias M. Falco, Lukas M. Weber, Erdogan Pekcan Erkan, Kaiyang Zhang, Jennifer Anne Doherty, Anna Vaharautio, Casey S. Greene, Stephanie C. Hicks
Summary: The miQC package was developed to predict low-quality cells in a given scRNA-seq dataset by jointly modeling the proportion of reads mapping to mitochondrial DNA (mtDNA) genes and the number of detected genes using mixture models in a probabilistic framework. The QC metric easily adapts to different types of single-cell datasets to remove low-quality cells while preserving high-quality cells for downstream analyses.
PLOS COMPUTATIONAL BIOLOGY
(2021)
Article
Physics, Fluids & Plasmas
Philipp J. Schneider, Thomas A. Weber
Summary: This paper proposes a method for estimating process parameters from time-censored data, which generates and corrects synthetic sample paths to match observed bin counts, and approximates the stochastic characteristics of the observed process iteratively.
Article
Computer Science, Artificial Intelligence
Jian-wei Liu, Zheng-ping Ren, Run-kun Lu, Xiong-lin Luo
Summary: This paper introduces a Gaussian Mixture Discriminant Analysis (GMDA) approach to learning from datasets with noisy labels, utilizing flipping probability and class probability. Experimental results demonstrate the superior performance of the proposed method over other state-of-the-art methods.
NEURAL COMPUTING & APPLICATIONS
(2021)
Article
Quantum Science & Technology
Tomoki Tanaka, Yohichi Suzuki, Shumpei Uno, Rudy Raymond, Tamiya Onodera, Naoki Yamamoto
Summary: This paper discusses a quantum algorithm based on maximum likelihood estimation for estimating the amplitude of a given quantum state in quantum devices. The research found the issue of estimation error saturation in the presence of noise, and proposed a noise model and basic hardware component requirements.
QUANTUM INFORMATION PROCESSING
(2021)
Article
Chemistry, Analytical
Mohamed Marey, Ahmed Sedik, Hala Mostafa
Summary: This study proposes a novel maximum likelihood recognizer for distinguishing SFBC OFDM waveforms in the presence of transmission defects. The theoretical findings show that IQDs from the transmitter and recipient can be combined with channel paths to generate effective channel paths. Simulation results demonstrate that the suggested strategy achieves a higher recognition accuracy compared to typical approaches in the literature.
Article
Automation & Control Systems
Amr Alanwar, Anne Koch, Frank Allgoewer, Karl Henrik Johansson
Summary: This paper discusses how to compute reachable sets directly from noisy data without a given system model. Several reachability algorithms are presented for different types of systems generating the data. For linear systems, an algorithm based on matrix zonotopes is proposed, which computes over-approximated reachable sets. Constrained matrix zonotopes are introduced to provide less conservative reachable sets at the cost of increased computational expenses and incorporate prior knowledge about the unknown system model. The approach is also extended to polynomial and nonlinear systems with theoretical guarantees of proper over-approximation.
IEEE TRANSACTIONS ON AUTOMATIC CONTROL
(2023)
Article
Computer Science, Information Systems
K. N. R. Surya Vara Prasad, Vijay K. Bhargava
Summary: This paper focuses on localization using received signal strength, addressing unknown transmit power and log-distance pathloss exponent. It proposes maximum-likelihood estimation, two-step linear least squares estimation, and a maximum-a-posteriori estimator for joint estimation of source location and PLE, demonstrating improved localization accuracy.
IEEE WIRELESS COMMUNICATIONS LETTERS
(2021)
Article
Geochemistry & Geophysics
Shinya Sato, Tada-Nori Goto, Takafumi Kasaya, Hiroshi Ichihara
Summary: The study introduces the magnetotelluric (MT) method and its applications in visualizing resistivity structures and monitoring resistivity changes. The Frequency-domain independent component analysis (FDICA) method mentioned in the article can effectively reduce the impact of cultural noise on MT data analysis and has high accuracy in MT data processing.
Article
Mathematics, Applied
Bin Dong, Zuowei Shen, Jianbin Yang
Summary: This paper explores the approximation of functions from noisy and nonsmooth observed data, with a focus on sparse noise removal schemes. Theoretical analysis is presented, highlighting the importance of sparsity-based denoising for effective approximation. A new approximation scheme is proposed for large datasets to significantly reduce noise level and ensure asymptotic convergence.
SIAM JOURNAL ON NUMERICAL ANALYSIS
(2021)