Research on speech enhancement based on nonnegative matrix factorization and improved genetic algorithm

Imaging genetics is an approach that explores the underlying mechanisms of brain disorders such as Alzheimer's disease (AD) by analyzing the correlation between neuroimaging and genetic data. Traditional non-negative matrix factorization (NMF) algorithms are based on linear assumptions, which limits the potential of nonlinear feature extraction among multi-omics data. This study proposes a novel joint-connectivity-based deep semi-supervised non-negative matrix factorization (JCB-DSNMF) model to overcome this limitation and incorporate prior knowledge from both within and between different modalities of data. The model effectively integrates physiological constraints such as connectivity to identify regions of interest (ROI), risk genes, and risk SNP loci associated with AD patients. JCB-DSNMF outperformed other NMF-based algorithms, such as JDSNMF and NMF, in identifying and predicting biologically relevant biomarkers closely related to AD from essential modules. The accuracy of the selected features was further validated by constructing a diagnostic model with high classification accuracy, achieving an AUC value of 0.8621 on the test set. In particular, the brain region Putamen_L and the gene RALGAPB achieved AUC values of 0.903 and 0.924, respectively, highlighting the importance of these features in early AD diagnosis.

Graph clustering is a fundamental task in network analysis, aimed at uncovering meaningful groups of nodes based on structural and attribute-based similarities. Traditional Nonnegative Matrix Factorization (NMF) methods have shown promise in clustering tasks by providing low-dimensional representations of data. However, most existing NMF-based approaches are highly sensitive to noise and outliers, leading to suboptimal performance in real-world scenarios. Additionally, these methods often struggle to capture the underlying nonlinear structures of complex networks, which can significantly impact clustering accuracy. To address these limitations, this paper introduces Robust Self-Supervised Symmetric NMF (R3SNMF) to improve graph clustering. The proposed algorithm leverages a robust principal component model to handle noise and outliers effectively. By incorporating a self-supervised learning mechanism, R3SNMF iteratively refines the clustering process, enhancing the quality of the learned representations and increasing resilience to data imperfections. The symmetric factorization ensures the preservation of network structures, while the self-supervised approach allows the model to adaptively improve its clustering performance over successive iterations. In addition, R3SNMF integrates a graph-boosting method to improve how relationships within the network are represented. Extensive experimental evaluations on various real-world graph datasets demonstrate that R3SNMF outperforms state-of-the-art clustering methods in terms of both accuracy and robustness.

Traditional Nonnegative Matrix Factorization (NMF) is a linear and unsupervised algorithm. This would limit the classification power of NMF for the complicated data. To overcome the above limitations of NMF, this paper proposes a novel supervised and nonlinear NMF algorithm based on kernel theory and discriminant analysis. We incorporate the class label information into the decomposition of NMF in the Reproducing Kernel Hilbert Space (RKHS). A new iterative algorithm for NMF is derived and the objective function is non-increasing under the update rules. The proposed method is evaluated on the ORL and Yale face databases. The experimental results demonstrate the the proposed method is superior to the state of-the-art algorithms.

Discovering community structures is an important step to understanding the structure and dynamics of real-world networks in social science, biology and technology. In this paper, we develop a deep stochastic model based on non-negative matrix factorization to identify communities, in which there are two sets of parameters. One is the community membership matrix, of which the elements in a row correspond to the probabilities of the given node belongs to each of the given number of communities in our model, another is the community-community connection matrix, of which the element in the i-th row and j-th column represents the probability of there being an edge between a randomly chosen node from the i-th community and a randomly chosen node from the j-th community. The parameters can be evaluated by an efficient updating rule, and its convergence can be guaranteed. The community-community connection matrix in our model is more precise than the community-community connection matrix in traditional non-negative matrix factorization methods. Furthermore, the method called symmetric nonnegative matrix factorization, is a special case of our model. Finally, based on the experiments on both synthetic and real-world networks data, it can be demonstrated that our algorithm is highly effective in detecting communities.

The mismatch between the training conditions and the test conditions severely degrades the performance of speaker verification. Aiming at solving this problem, this paper presents a method which uses a combination of improved nonnegative matrix factorization (IMNMF) and mel frequency cepstral coefficients (MFCC) with feature warping for improving identity-vector (i-vector) speaker verification performance in noisy environment. Unlike the traditional nonnegative matrix factorization (NMF), IMNMF uses extra free basis vectors to capture the features which are not included in training data, and linear constraints on dictionary atoms. Feature warping is used to remove channel noises. Therefore, the proposed method can reduce distortion of reconstructed speech while enhancing the recovered speech quality. The performance of i-vector speaker verification is evaluated by using the short utterance database and the NOISEX-92 database. The experiment results indicate that the score level fusion of feature-warped IMNMF-MFCC and feature-warped MFCC is superior to the baseline system at the equal error rate under the majority of signal-to-noise ratios (SNRs).

Hyperspectral unmixing (HU) is a research hotspot of hyperspectral remote sensing technology. As a classical HU method, the nonnegative matrix factorization (NMF) unmixing method can decompose an observed hyperspectral data matrix into the product of two nonnegative matrices, i.e., endmember and abundance matrices. Because the objective function of NMF is the traditional least-squares function, NMF is sensitive to noise. In order to improve the robustness of NMF, this paper proposes a maximum likelihood estimation (MLE) based NMF model (MLENMF) for unmixing of hyperspectral images (HSIs), which substitutes the least-squares objective function in traditional NMF by a robust MLE-based loss function. Experimental results on a simulated and two widely used real hyperspectral data sets demonstrate the superiority of our MLENMF over existing NMF methods.

Incremental semi-supervised graph learning NMF with block-diagonal

Semi-supervised graph regularized nonnegative matrix factorization with local coordinate for image representation

Non-negative matrix factorization (NMF) has been successfully applied to speech enhancement in non-stationary noisy environments. Recently proposed online semi-supervised NMF algorithms are of particular interest as they carry the two nice properties (online and semi-supervised) of classical speech enhancement approaches. These algorithms, however, have only been evaluated using noisy mixtures shorter than 30 seconds. In this paper we find that these algorithms work well when it is run for less than 1 minute, but degradation of the enhanced speech signal starts to appear after 2 minutes. We analyze that the reason is due to the inappropriate dictionary update rule, which gradually loses its ability in updating the speech dictionary. We then propose a simple rotational reset strategy to solve the problem: Instead of continuously updating the entire speech dictionary, we periodically and rotationally select elements and reset their values to random numbers. Experiments show that this strategy successfully solves the degradation problem and the improved algorithm outperforms classical speech enhancement algorithms significantly even when they are run for 10 minutes.

This paper proposes a speech enhancement method for non-stationary Gaussian noise based on regularized non-negative matrix factorization (NMF). The magnitudes of speech and noise are implemented by a model based in iterative posterior NMF which are applied using prior distributions in transform domain. This is used since the sample distributions of the above are well suited to Weibull and Rayleigh densities well. For the accomplishment in time-varying noise environments, both the speech and noise bases of NMF are adapted simultaneously. With the usage of estimated speech presence probability, this paper proposes to adaptively estimate the statistics of these distributions. The method in this paper gives the best results for perceptual evaluation of speech quality (PESQ) and the signal-to-distortion ratio (SDR).

Abstract. The Pearl River Delta (PRD) of China, which has a population of more than 58 million people, is one of the largest agglomerations of cities in the world and had severe PM2.5 pollution at the beginning of this century. Due to the implementation of strong pollution control in recent decades, PM2.5 in the PRD has continuously decreased to relatively lower levels in China. To comprehensively understand the current PM2.5 sources in the PRD to support future air pollution control strategies in similar regions, we performed regional-scale PM2.5 field observations coupled with a state-of-the-art source apportionment model at six sites in four seasons in 2015. The regional annual average PM2.5 concentration based on the 4-month sampling was determined to be 37 µg m−3, which is still more than 3 times the WHO standard, with organic matter (36.9 %) and SO42- (23.6 %) as the most abundant species. A novel multilinear engine (ME-2) model was first applied to a comprehensive PM2.5 chemical dataset to perform source apportionment with predetermined constraints, producing more environmentally meaningful results compared to those obtained using traditional positive matrix factorization (PMF) modeling. The regional annual average PM2.5 source structure in the PRD was retrieved to be secondary sulfate (21 %), vehicle emissions (14 %), industrial emissions (13 %), secondary nitrate (11 %), biomass burning (11 %), secondary organic aerosol (SOA, 7 %), coal burning (6 %), fugitive dust (5 %), ship emissions (3 %) and aged sea salt (2 %). Analyzing the spatial distribution of PM2.5 sources under different weather conditions clearly identified the central PRD area as the key emission area for SO2, NOx, coal burning, biomass burning, industrial emissions and vehicle emissions. It was further estimated that under the polluted northerly air flow in winter, local emissions in the central PRD area accounted for approximately 45 % of the total PM2.5, with secondary nitrate and biomass burning being most abundant; in contrast, the regional transport from outside the PRD accounted for more than half of PM2.5, with secondary sulfate representing the most abundant transported species.

이 논문은 고차원의 데이터를 저 차원으로 줄이는 방법 중 하나인 특징추출에 대한 방법들의 특성을 비교한다. 비교대상 방법은 전통적인 PCA(Principal Component Analysis)방법과 시각피질의 특성을 보인다고 알려진 ICA(Independent Component Analysis), 국소기반인식을 구현한 NMF(Non-negative Matrix Factorization), 그리고 이의 성능을 개선한 sNMF(Sparse NMF)로 정하였다. 추출된 특징들의 특성을 시각적으로 확인하기 위하여 필기체 숫자 영상을 대상으로 특징추출을 수행하였으며, 인식기에 적용한 효과의 확인을 위하여 추출된 특징을 다층퍼셉트론에 학습시켜보았다. 각 방법의 특성을 비교한 결과는 응용하고자 하는 문제에서 어떤 특징을 추출하기 원하느냐에 따라 특징추출 방법을 선정할 때 유용할 것이다. In this paper, feature extraction methods, which is one field of reducing dimensions of high-dimensional data, are empirically investigated. We selected the traditional PCA(Principal Component Analysis), ICA(Independent Component Analysis), NMF(Non-negative Matrix Factorization), and sNMF(Sparse NMF) for comparisons. ICA has a similar feature with the simple cell of V1. NMF implemented a "parts-based representation in the brain" and sNMF is a improved version of NMF. In order to visually investigate the extracted features, handwritten digits are handled. Also, the extracted features are used to train multi-layer perceptrons for recognition test. The characteristic of each feature extraction method will be useful when applying feature extraction methods to many real-world problems.

Reducing highly non-stationary transient noise, such as keyboard typing noise, remains a challenging problem for many singlechannel speech enhancement algorithms. This paper proposes two approaches based on nonnegative matrix factorization (NMF) and probabilistic latent component analysis for transient noise reduction using a pre-trained transient noise dictionary and a universal speaker-independent speech dictionary. In addition, we develop an NMF-based speech enhancement scheme to simultaneously reduce transient and non-transient background noise, in which a low-dimensional dictionary is learnt from the noisy observations to model the background noise. We exploit the temporal dependencies of speech and background noise to design and apply informative priors via a probabilistic framework, while ignoring the temporal dynamics of the transient noise. Experimental results show that the proposed algorithms can improve the perceptual evaluation of speech quality (PESQ) up to 1.2 MOS for the keyboard typing noise.

본 논문은 시간 연속성을 갖는 비음수 행렬 분해(Nonnegative Matrix Factorization, NMF)를 이용하여 잡음에 열화된 음성 신호의 음질을 개선하는 문제를 다룬다. 음성과 잡음 신호는 포아송 분포로 모델되며, NMF의 기본 벡터와 이득 벡터는 감마 분포로 모델된다. 이득 벡터의 시간 연속성은 음질 개선에 중요한 영향을 미치는 것으로 알려져 있다. 본 논문에서 시간의 연속성은 이득 벡터를 감마-마르코프 연쇄(Gamma-Markov chain, GMC) 사전 분포로 모델함으로써 이루어진다. 실험 결과는 제안된 알고리즘이 잡음 신호의 시간 연속성을 효과적으로 모델하는 것을 보여준다. In this paper, speech enhancement using nonnegative matrix factorization with temporal continuity has been addressed. Speech and noise signals are modeled as Possion distributions, and basis vectors and gain vectors of NMF are modeled as Gamma distributions. Temporal continuity of the gain vector is known to be critical to the quality of enhanced speech signals. In this paper, temporal continiuty is implemented by adopting Gamma-Markov chain priors for noise gain vectors during the separation phase. Simulation results show that the Gamma-Markov chain models temporal continuity of noise signals and track changes in noise effectively.

This study proposes a bi-directional recurrent neural network (Bi-RNN) post-processing method for speech enhancement (SE) at low signal-to noise ratios (SNR). Current speech enhancement solutions performed badly under low SNR situations. Loizou and Kim proposed a solution to reduce speech distortion errors in time-frequency (T-F) domain but it requires the knowledge of ground truth. As ground truth is unknown in real-life applications, the current study proposes to use a Bi-RNN to implement Loizou and Kim’s solution as a post-processing method for SE engines. Our solutions do not require prior knowledge of ground truth. The effectiveness of the proposed method is investigated with a spectral subtraction (SS) SE engine, a non-negative matrix factorization (NMF) SE engine, and a deep neural network ideal ratio mask (DNN-IRM) SE engine, under matched/mis-matched noise and different SNR conditions. Experimental results demonstrate that the proposed post-processing method effectively improved both perceptual evaluation of speech quality (PESQ) and short-time objective intelligibility (STOI) for all of these SE engines, especially at low SNR conditions.