Article
Engineering, Biomedical
Yuanyuan Zhang, Jianghui Xin
Summary: The blind separation algorithm faces the issue of separation stability in the presence of multiple convoluted signals. By enhancing joint diagonalization with cluster analysis, optimal distance matrix, and optimal window, the stability and accuracy of signal separation have been effectively improved. Introducing full frequency divergence as the objective function for permutation ambiguity in the convolution separating process has resolved failures in convoluted blind source separation. The combination of new joint diagonalization and convolution separation forms a systematic approach that, when applied to floor signals on an actual bench, successfully captures the influence of excitation sources on frequency, making the system algorithm a valuable reference for mechanical vibration analysis.
JOURNAL OF VIBROENGINEERING
(2021)
Article
Computer Science, Artificial Intelligence
Yuan Xie, Kan Xie, Shengli Xie
Summary: In this paper, a novel framework is proposed to solve the underdetermined blind source separation of speech mixtures problem using a compressed sensing model. The method includes noise reduction pretreatment, blind identification for accurate mixing matrix estimation, and simultaneous updating of codewords and coefficients for dictionary selection. The approach reduces computational complexity and demonstrates superiority in experimental results.
INTERNATIONAL JOURNAL OF MACHINE LEARNING AND CYBERNETICS
(2021)
Article
Acoustics
Jiawei Jian, Li Wang, Zhong-Rong Lu
Summary: This paper proposes a novel method for underdetermined operational modal analysis (OMA) using blind source separation (BSS). The method efficiently transforms the underdetermined problem into determined or overdetermined ones and recovers modal responses and mode shapes using an improved second-order blind identification (SOBI) technique. The effectiveness of the proposed strategy is verified through numerical examples, an experimental frame case, and a field test of a pedestrian bridge.
JOURNAL OF SOUND AND VIBRATION
(2023)
Article
Acoustics
Nobutaka Ito, Rintaro Ikeshita, Hiroshi Sawada, Tomohiro Nakatani
Summary: This paper introduces two methods for blind source separation (BSS): one using the multichannel Wiener filter (MWF) applicable to underdetermined cases, and one using a joint diagonalization approach. The study shows that the latter is more efficient than the former and suitable for large data or limited computational resources.
IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING
(2021)
Article
Physics, Multidisciplinary
Jindong Wang, Xin Chen, Haiyang Zhao, Yanyang Li, Zujian Liu
Summary: This paper proposes a two-stage clustering method that combines hierarchical clustering and K-means to enhance the reliability of mixing matrix estimation, aiming to solve the problems in the K-means algorithm. By clustering the observed signals and eliminating outliers using cosine distance, the improved K-means is used to estimate the mixing matrix and recover the source signals successfully.
Article
Computer Science, Information Systems
Zhanyu Zhu, Xingjie Chen, Zhaomin Lv
Summary: This study proposes a two-stage single-source point screening method that combines the cosine angle algorithm and the L1-norm optimization algorithm for estimating the mixing matrix and achieving blind source separation. Experimental results demonstrate that this method can obtain more accurate and robust mixing matrix estimation, leading to better separation of the source signals.
Article
Computer Science, Information Systems
Mengdie Niu, Ye Zhang
Summary: In this paper, a novel autoencoder network architecture with clustering mechanism is proposed for underdetermined blind speech source separation. Experimental results demonstrate that the proposed method outperforms baseline algorithms.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Computer Science, Information Systems
Zhongqiang Luo, Ruiming Guo, Chengjie Li
Summary: This paper proposes an independent vector analysis (WA) detection receiver for blindly deconvolving the convolutive mixtures of digitally modulated signals for wireless communications. The method jointly carries out separation work for different frequency bin data fusion, and solves the random permutation problem of separation signals by exploiting the dependencies of frequency bins. Simulation results and analysis demonstrate the effectiveness of the proposed detection method.
Article
Chemistry, Analytical
Norsalina Hassan, Dzati Athiar Ramli
Summary: Blind source separation (BSS) is a method to recover source signals without knowing the mixing process or source signals. Sparse component analysis (SCA) is a commonly used solution for underdetermined BSS, which includes mixing matrix estimation and source recovery estimation. Adaptive time-frequency thresholding (ATFT) is introduced to improve the accuracy of the mixing matrix estimation, while least squares methods are used for source recovery estimation.
Article
Engineering, Electrical & Electronic
Jiayi Guo, Sen Liu, Kun Yu, Xiaoman Chen, Yunpeng Liu, Fangcheng Lu
Summary: An UHV shunt reactor acoustic signal separation method based on masking beamforming and undetermined blind source separation was proposed. This method can separate acoustic components with different spatial characteristics and frequency domain characteristics from the original mixed acoustic signal. Laboratory and substation tests were conducted to verify the feasibility of the method, and the results showed a power spectrum similarity improvement from 0.617 to 0.924.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2023)
Article
Computer Science, Information Systems
Mahbanou Zohrevandi, Saeed Setayeshi, Azam Rabiee, Midia Reshadi
Summary: This paper proposes a method for separating underdetermined convolutive blind speech in a multi-speaker environment based on mask prediction in the time-frequency domain. Clustering with a weighting function is used to consider parts of masks with potentially only one active source, and sparse filters are utilized in the time-domain to improve signal quality. Performance evaluation shows that the proposed method is more accurate than conventional algorithms.
MULTIMEDIA TOOLS AND APPLICATIONS
(2021)
Article
Computer Science, Information Systems
Weihong Fu, Xiaowei Bai, Fan Shi, Chunhua Zhou, Yongyuan Liu
Summary: The paper introduces an original algorithm for underdetermined mixing matrix estimation, which effectively estimates the mixing matrix and source signal number, improving sparsity and accuracy. By utilizing methods such as transform matrix and element sorting, successful source point detection and signal number estimation have been achieved.
Article
Environmental Sciences
Wenyi Li, Gang Liu, Xiaowei Cui, Mingquan Lu
Summary: This study proposed an innovative integrated algorithm to improve RTK performance in GNSS-challenged environments by utilizing RTK to register LiDAR features and adopting parallel filters in the ambiguity-position-joint domain. This method successfully weakened the effects of low satellite availability, cycle slips, and multipath, thus enhancing the RTK fix rate and stability. Theoretical analyses, simulation experiments, and a road test all demonstrated the effectiveness of the proposed method in improving RTK performance and ensuring the global positioning precision of the integrated system.
Article
Multidisciplinary Sciences
Qingyi Wang, Yiqiong Zhang, Shuai Yin, Yuduo Wang, Genping Wu
Summary: The proposed method for underdetermined blind source separation (UBSS) includes three main steps: screening single source points using principal component analysis, estimating the mixing matrix using a combination of OPTICS and an improved potential function, and recovering source signals using an improved subspace projection method. The method is independent of input parameters, offers high accuracy and robustness, and performs well in noisy environments.
Article
Engineering, Electrical & Electronic
Hong Zhong, Yang Ding, Yahui Qian, Liangmo Wang, Baogang Wen
Summary: This paper proposes a novel nonlinear underdetermined blind source separation (UBSS) solution for bearing fault diagnosis. It utilizes source number estimation and improved sparse component analysis (SCA) to deal with the problem of nonlinear mixture of vibration signals. The proposed approach includes ensemble empirical mode decomposition (EEMD), correlation coefficient (CC), and adaptive threshold singular value decomposition (ATSVD) for source number estimation, and short-time Fourier transform (STFT) for transforming observed signals into the time-frequency domain. The results from simulations and experiments demonstrate that the proposed UBSS solution can accurately estimate the source number and effectively separate the signals.
Article
Mathematics, Applied
Mikael Sorensen, Lieven De Lathauwer, Nicholaos D. Sidiropoulos
Summary: The Canonical Polyadic Decomposition (CPD) has been extended to the more general case of bilinear factorizations subject to monomial equality constraints in this paper. A deterministic uniqueness condition is proposed and the problem is simplified by reducing the bilinear factorization problem into a CPD problem, solvable via Matrix EigenValue Decomposition (EVD). The discussed EVD-based algorithms are guaranteed to return the exact bilinear factorization under certain conditions.
LINEAR ALGEBRA AND ITS APPLICATIONS
(2021)
Article
Engineering, Electrical & Electronic
Fengyu Zhou, Ahmed S. Zamzam, Steven H. Low, Nicholas D. Sidiropoulos
Summary: Simulation results show that semi-definite relaxations of AC optimal power flow on unbalanced distribution networks with wye and delta connections may lead to inexact solutions due to the non-uniqueness of relaxation solutions and numerical errors. The proposed algorithms in this study are able to recover exact optimal solutions with numerical precision for such networks.
IEEE TRANSACTIONS ON SMART GRID
(2021)
Article
Computer Science, Artificial Intelligence
Bo Yang, Ahmed S. Zamzam, Nicholas D. Sidiropoulos
Summary: In this study, a parallel tensor factorization algorithm called ParaSketch is proposed to handle large tensors. The algorithm compresses the large tensor into multiple small tensors and decomposes and combines them in parallel to reconstruct the latent factors. By utilizing sketching matrices for compression, the proposed method significantly reduces compression complexity and maintains latent identifiability under certain conditions. Numerical experiments confirm the effectiveness of the algorithm.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2022)
Article
Engineering, Electrical & Electronic
Ruiyuan Wu, Wing-Kin Ma, Yuening Li, Anthony Man-Cho So, Nicholas D. Sidiropoulos
Summary: This study presents PRISM, a probabilistic simplex component analysis approach to identifying the vertices of a data-circumscribing simplex from data. PRISM uses a simple probabilistic model and carries out inference by maximum likelihood. It has strong connections with simplex volume minimization and shows potential in combating noise.
IEEE TRANSACTIONS ON SIGNAL PROCESSING
(2022)
Article
Engineering, Electrical & Electronic
Mohamed Salah Ibrahim, Ahmed S. Zamzam, Aritra Konar, Nicholas D. Sidiropoulos
Summary: This paper proposes a method based on generalized canonical correlation analysis (GCCA) to improve the uplink quality of service for users located around the boundaries between cells. By leveraging selective BS cooperation, the method can recover the cell-edge user signal subspace even at low SNR, and extract the cell-edge user signals from the resulting mixture.
IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS
(2021)
Article
Computer Science, Artificial Intelligence
Aritra Konar, Nicholas D. Sidiropoulos
Summary: This article proposes a new perspective for designing an approximation algorithm for graph matching, and its effectiveness is demonstrated through experiments.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2022)
Article
Engineering, Electrical & Electronic
Mohamed Salah Ibrahim, Ahmed Hussain, Nicholas D. Sidiropoulos
Summary: This paper proposes a new pilot-free TDD frame structure that allows designing highly effective multiuser decoders and precoders in an unsupervised manner. The key idea is to use a pre-assigned permutation code to repeat and permute each user's uplink data, and then utilize canonical correlation analysis (CCA) to generate high quality CCA-based beamformers. The paper also presents a pilotless synchronization framework that leverages CCA to recover timing and frequency offsets. Experimental results demonstrate the effectiveness of the proposed approach in multiuser and multicell networks, even under common hardware imperfections.
IEEE TRANSACTIONS ON COMMUNICATIONS
(2022)
Article
Engineering, Electrical & Electronic
Mohamed Salah Ibrahim, Paris A. Karakasis, Nicholas D. Sidiropoulos
Summary: This paper proposes a practical underlay scheme for reliable secondary communication in low-power short-range scenarios. By exploiting the repetition structure and using canonical correlation analysis, the secondary signal can be recovered efficiently. Experimental results demonstrate the effectiveness of the approach in detecting and recovering the secondary signal within a certain signal to interference plus noise ratio (SINR) range.
IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS
(2022)
Article
Engineering, Electrical & Electronic
Mikael Sorensen, Nicholas D. Sidiropoulos, Ananthram Swami
Summary: A new SBMF model is proposed for community detection, which can handle overlapping communities effectively, and its effectiveness is demonstrated through experiments.
IEEE TRANSACTIONS ON SIGNAL PROCESSING
(2022)
Article
Computer Science, Artificial Intelligence
Paris A. Karakasis, Athanasios P. Liavas, Nicholas D. Sidiropoulos, Panagiotis G. Simos, Efrosini Papadaki
Summary: Functional magnetic resonance imaging (fMRI) is widely used for studying the human brain. This study proposes a new fMRI data generating model that takes into account both task-related and resting-state components. Experimental tests show that our method can accurately estimate temporal and spatial components even at low Signal to Noise Ratio (SNR), and outperforms standard procedures based on General Linear Models (GLMs).
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2022)
Article
Engineering, Electrical & Electronic
Magda Amiridi, Nikos Kargas, Nicholas D. Sidiropoulos
Summary: This paper proposes a novel approach based on tensor factorization for non-parametric density estimation in high-dimensional multivariate data analysis. By using a tensor model of the characteristic function, the density can be accurately estimated and the curse of dimensionality can be overcome.
IEEE TRANSACTIONS ON SIGNAL PROCESSING
(2022)
Article
Engineering, Electrical & Electronic
Magda Amiridi, Nikos Kargas, Nicholas D. Sidiropoulos
Summary: This paper proposes a framework that combines dimensionality reduction with non-parametric density estimation. The proposed model captures the underlying distribution of the input data by designing a nonlinear dimensionality reducing auto-encoder. The model achieves promising results on various tasks and datasets.
IEEE TRANSACTIONS ON SIGNAL PROCESSING
(2022)
Article
Engineering, Electrical & Electronic
Magda Amiridi, Nikos Kargas, Nicholas D. Sidiropoulos
Summary: This paper introduces a method of feature selection through low-rank tensor modeling to mitigate complexity and maximize classification performance. By learning the "principal components" of the joint distribution to avoid the curse of dimensionality, and by using a greedy algorithm to tackle the feature selection problem.
IEEE TRANSACTIONS ON SIGNAL PROCESSING
(2021)
Article
Engineering, Electrical & Electronic
Mikael Sorensen, Charilaos Kanatsoulis, Nicholas D. Sidiropoulos
Summary: Generalized Canonical Correlation Analysis (GCCA) is an important tool used in data mining, machine learning, and artificial intelligence to find common random variables across multiple feature representations. This paper offers a fresh algebraic perspective on GCCA and proposes a novel algorithm based on subspace intersection. Experimental results demonstrate the effectiveness of this approach in handling large GCCA tasks.
IEEE TRANSACTIONS ON SIGNAL PROCESSING
(2021)
Article
Engineering, Electrical & Electronic
Faisal M. Almutairi, Charilaos I. Kanatsoulis, Nicholas D. Sidiropoulos
Summary: This paper discusses the aggregation and disaggregation of multidimensional data, introducing a method called Prema to reconstruct finer-scale data from multiple coarse views.
IEEE JOURNAL OF SELECTED TOPICS IN SIGNAL PROCESSING
(2021)