Article
Automation & Control Systems
Ping Ma, Yongkai Chen, Xinlian Zhang, Xin Xing, Jingyi Ma, Michael W. Mahoney
Summary: In this article, we develop an asymptotic analysis to derive the distribution of RandNLA sampling estimators for the least-squares problem. We show that the sampling estimator is asymptotically normally distributed under mild regularity conditions and is asymptotically unbiased in both full sample approximation and model parameter inference settings. Based on our asymptotic analysis, we identify optimal sampling probabilities using two criteria and propose several new optimal sampling probability distributions. Our theoretical and empirical results provide insights on the role of leverage in the sampling process and demonstrate improvements over existing methods.
JOURNAL OF MACHINE LEARNING RESEARCH
(2022)
Article
Engineering, Electrical & Electronic
Swagata Nandi, Debasis Kundu
Summary: This paper investigates the problem of estimating parameters in a multichannel sinusoidal model. Two estimation methods, the minimization of the sum of residual sum of squares and the use of more efficient generalized least squares estimators, are proposed and compared through simulation experiments. The results show that the variances of the generalized least squares estimators reach the Cramer-Rao lower bound, and the computational complexity does not significantly increase with the number of channels.
CIRCUITS SYSTEMS AND SIGNAL PROCESSING
(2022)
Article
Multidisciplinary Sciences
Lianjie Duan
Summary: This paper uses China's trade and production data from 2000 to 2006 to estimate the export cutoff productivity using the non-parametric ROC method. The results show variations in export cutoff productivity among domestic and foreign enterprises, different industries, and regions. The study also reveals a changing trend of the export threshold in a horizontal S-shape during the sample period and an inverted U-shape after China's accession to WTO in 2001.
Article
Mathematics, Applied
Lingling He, Xiaoqin Li, Yan Shen, Qiuyue Wu
Summary: In this paper, we investigate the partially linear regression model based on asymptotically almost negatively associated (AANA) random variables. Convergence results for the parametric least squares estimator and nonparametric weighted estimator are obtained under weak conditions. The results extend the corresponding ones for negatively associated (NA) errors to AANA errors and the selection of design points and weight functions is discussed. Simulation experiments are conducted to demonstrate the performance of the obtained results.
JOURNAL OF MATHEMATICAL INEQUALITIES
(2023)
Article
Physics, Multidisciplinary
Tianyuan Guan, Mohammed Khorshed Alam, Marepalli Bhaskara Rao
Summary: This article addresses the issue of determining the sample size for a given level and power in a simple linear regression model. The standard approach deals with planned experiments, where both the predictor X and the response variable Y are observed. The statistic used relies on the least squares' estimator, but it is problematic because it is based on the conditional distribution given the data on X. For unplanned experiments where X and Y are sampled simultaneously, we don't have data on X yet. To solve this, the article determines the exact unconditional distribution of the test statistic in unplanned cases.
Article
Engineering, Electrical & Electronic
Foad Fereidoony, Ali Jishi, Maziar Hedayati, Yuanxun Ethan Wang, Sridhar Kowdley
Summary: This paper proposes the Magnitude-Delay Least Mean Squares Equalization (MD-LMSE) for super-resolution time delay estimation, offering high accuracy, super resolution, and low computational time. The experimental results show that the approach can improve the range resolution of a system by 95% with low error.
IEEE SENSORS JOURNAL
(2021)
Article
Mathematics
Guoping Zeng, Sha Tao
Summary: This paper investigates the effects of linear transformations in logistic regression. A generalized linear transformation for multiple variables is proposed using matrix multiplication. The study shows that an invertible linear transformation has no effects on predictions, multicollinearity, pseudo-complete separation, and complete separation.
Article
Green & Sustainable Science & Technology
Jinshan Yu, Zhongyuan Zheng, Yamin Li, Haohui Wang, Ying Hao, Xiaoxia Liang, Jianzheng Gao
Summary: This study develops a real-time active noise control system and evaluates different ANC methods for low-frequency noise reduction. The results show that the ANC system is effective in attenuating substation noise.
Article
Engineering, Biomedical
Jose Henrique Ferreira de Souza, Tiago Zanotelli, Leonardo Bonato Felix, Felipe Antunes
Summary: Auditory Steady-State Responses (ASSR) are evoked potentials used for estimating hearing thresholds. This study proposes an alternative technique to the discrete Fourier transform (DFT) called the least squares method with phase compensation. Results showed a small calibration error in the dataset and demonstrated the performance degradation of the magnitude-squared coherence (MSC) when using either very small or very large epoch lengths in real data. The proposed method allows for analysis with varying epoch lengths and frequencies, which was not possible with DFT.
BIOMEDICAL SIGNAL PROCESSING AND CONTROL
(2023)
Article
Geosciences, Multidisciplinary
Hongkai Shi, Xiufeng He, Yihao Wu, Ole Baltazar Andersen, Per Knudsen, Yanxiong Liu, Zhetao Zhang
Summary: The study proposes a least squares-based approach to model mean dynamic topography signal, which performs better in recovering current signal compared to traditional isotropic filtering methods. Error analysis indicates errors are primarily concentrated near coastal regions.
FRONTIERS IN EARTH SCIENCE
(2022)
Article
Acoustics
Sridhar Chintala, Jaisingh Thangaraj, Damodar Reddy Edla
Summary: A novel adaptive algorithm, based on a new step size, is proposed to eliminate ocular artifacts from recorded raw EEG signals. By using second and fourth-order power optimization algorithms, reference signals are processed and subtracted to obtain true EEG signals.
Article
Mathematics
Xiao Ke, Sirao Wang, Min Zhou, Huajun Ye
Summary: This paper presents new methods for parameter estimation of gamma distribution using representative points. The first part discusses the theoretical existence and uniqueness of gamma mean squared error representative points (MSE-RPs). By comparing three types of representative points, the second part demonstrates that gamma MSE-RPs perform well in parameter estimation and simulation. The last part proposes a new Harrel-Davis sample standardization technique. Simulation studies show that standardized samples can improve estimation performance or generate MSE-RPs. Additionally, a real data analysis indicates that the proposed technique efficiently estimates gamma parameters.
Article
Statistics & Probability
Zahra Zafar, Muhammad Aslam
Summary: This article extensively discusses the issue of heteroscedasticity and its negative impact on linear regression model estimation. It introduces a relatively new approach called the least squares ratio method to estimate a linear regression model in the presence of heteroscedasticity and outliers. An adaptive version of this technique is proposed, taking advantage of existing adaptive estimators to address the issue of unknown heteroscedasticity. Monte Carlo simulation is used to evaluate the performance of the proposed estimator under varying degrees of heteroscedasticity and number of outliers.
COMMUNICATIONS IN STATISTICS-SIMULATION AND COMPUTATION
(2023)
Article
Computer Science, Software Engineering
Abdul Majid, Shakeel Ahmad, Muhammad Aslam, Muhammad Kashif
Summary: This article proposes a robust version of the Kibria-Lukman estimator (KLE) to address the problems of multicollinearity and outliers in regression models. The performance of the proposed method is evaluated through Monte Carlo simulation and real-life data, demonstrating its superiority.
CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE
(2023)
Article
Automation & Control Systems
Ridvan Demir
Summary: This study proposes stator currents-based model reference adaptive system (MRAS) estimators for speed estimation of the PMSM. Adaptive filtering algorithms, namely the least mean squares (LMS) and least mean Kurtosis (LMK) algorithms, are used in the adaptation mechanism. The proposed estimators directly estimate the rotor speed of the PMSM by considering the error between the measured stator currents (reference model) and the stator currents at the output of the adaptive model. Simulation results demonstrate that the proposed estimators outperform MRAS using a traditional PI controller and eliminate the need for a fixed-gain PI controller often used in MRAS structures.
INTERNATIONAL JOURNAL OF CONTROL AUTOMATION AND SYSTEMS
(2023)