Non-negative Matrix Factorization Features Research Articles

Segmentation of Infant Brain Using Nonnegative Matrix Factorization

This study develops an atlas-based automated framework for segmenting infants’ brains from magnetic resonance imaging (MRI). For the accurate segmentation of different structures of an infant’s brain at the isointense age (6–12 months), our framework integrates features of diffusion tensor imaging (DTI) (e.g., the fractional anisotropy (FA)). A brain diffusion tensor (DT) image and its region map are considered samples of a Markov–Gibbs random field (MGRF) that jointly models visual appearance, shape, and spatial homogeneity of a goal structure. The visual appearance is modeled with an empirical distribution of the probability of the DTI features, fused by their nonnegative matrix factorization (NMF) and allocation to data clusters. Projecting an initial high-dimensional feature space onto a low-dimensional space of the significant fused features with the NMF allows for better separation of the goal structure and its background. The cluster centers in the latter space are determined at the training stage by the K-means clustering. In order to adapt to large infant brain inhomogeneities and segment the brain images more accurately, appearance descriptors of both the first-order and second-order are taken into account in the fused NMF feature space. Additionally, a second-order MGRF model is used to describe the appearance based on the voxel intensities and their pairwise spatial dependencies. An adaptive shape prior that is spatially variant is constructed from a training set of co-aligned images, forming an atlas database. Moreover, the spatial homogeneity of the shape is described with a spatially uniform 3D MGRF of the second-order for region labels. In vivo experiments on nine infant datasets showed promising results in terms of the accuracy, which was computed using three metrics: the 95-percentile modified Hausdorff distance (MHD), the Dice similarity coefficient (DSC), and the absolute volume difference (AVD). Both the quantitative and visual assessments confirm that integrating the proposed NMF-fused DTI feature and intensity MGRF models of visual appearance, the adaptive shape prior, and the shape homogeneity MGRF model is promising in segmenting the infant brain DTI.

Combining adaptive sparse NMF feature extraction and soft mask to optimize DNN for speech enhancement

In masking-based deep neural network (DNN) speech enhancement, the time–frequency masking value cannot be estimated accurately because the potential structure information of speech is ignored. In this paper, a speech enhancement method is proposed by combining adaptive sparse non-negative matrix factorization (NMF) feature extraction and soft mask to optimize DNN, using the advantages of the sparse matrix in catching the protruding structure of speech and combining with optimized masking-based prediction. First, considering the dominance of speech and noise interference in different noisy speech signals, this paper proposes a new method for estimating soft mask value, and the initial soft mask value is estimated by using speech cochleagram and noise cochleagram. Then, speech cochleagram and noise cochleagram are learned separately by the sparse NMF (SNMF) to obtain a joint dictionary. The noisy speech is sparsely represented on the joint dictionary, and the adaptive adjustment factor related to the changes of speech and noise dictionary is added to obtain the sparse coefficient. The sparse coefficient is used as the input of the DNN model, and the initial soft mask value is used as the learning label to estimate the final soft mask value. Finally, the estimated soft mask value is combined with the noisy speech cochleagram to obtain enhanced speech. Compared with other methods, the results show that 1.6039 dB increases the average signal-to-noise ratio (SNR) of the proposed method, the average perceptual evaluation of speech quality (PESQ) is increased by 0.1994, and the average short-time objective intelligibility (STOI) is improved by 0.0271, which fully illustrate the superiority of the proposed algorithm.

FER based on the improved convex nonnegative matrix factorization feature

Facial expression recognition is an important research issue in the pattern recognition field. In this paper, we intend to present an accurate facial expression recognition (FER) system, which employs an improved convex non-negative matrix factorization (ICNMF) method based on a novel objective function and smaller iterative step sizes for feature extraction. Since negative values appearing in the facial expression feature will weaken the features and reduce the recognition rate, the nonnegative matrix factorization (NMF) methods are adopted to guarantee the non-negativity of the extracted feature value to improve the recognition rate. To enhance the performance of NMF method for FER, the ICNMF approach based on a novel convergent objective function and smaller iterative step sizes is proposed, and the FER rate can be improved effectively. In the FER system, the face region is detected firstly, and is enhanced by histogram specification, secondly the ICNMF approach is adopted to extract features and then the feature coefficient matrix is achieved. Finally, the SVM classifier is applied to recognize the extracted features. To validate the effectiveness of FER system, four public available datasets of MultiPIE, CK+, FER2013 and SFEW are tested and then high recognition rates can be achieved based on ICNMF method. In addition, the proposed ICNMF approach is compared with the methods of multi-layer NMF, sparse non-negative matrix factorization (SNMF), the traditional convex non-negative matrix factorization (CNMF), deep belief networks (DBN) and stacked auto-encoder (SAE), and results of experiments show that the proposed ICNMF approach is significantly effective contrasting to the other five expression extraction methods.

CLASSIFICATION OF INTRAVASCULAR ULTRASOUND IMAGES BASED ON NON-NEGATIVE MATRIX FACTORIZATION FEATURES AND MAXIMUM LIKELIHOOD CLASSIFIER

The amount of plaque in coronary arteries in any particular point is identified by the IntraVascular UltraSound (IVUS) images. The classification of IVUS images is very important to diagnose various coronary artery diseases. In this study, the classification of IVUS images based on Non-negative Matrix Factorization (NMF) technique and Maximum Likelihood Classifier (MLC) is presented. Initially, the IVUS images are given to frost filter to remove speckle noise as the imaging technique uses ultrasound waves. Then, NMF technique is employed to extract the features and stored in database. Then MLC is used for classification of IVUS images for both normal and abnormal categories. The IVUS Image Classification (IIC) system obtains 98% classification accuracy by using NMF features and MLC classification.

Deep Learning of EEG Time–Frequency Representations for Identifying Eye States

A new Convolutional Neural Network (CNN) architecture to classify nonstationary biomedical signals using their time–frequency representations is proposed. The present method uses the spectrogram of the biomedical signals as an input to CNN, in addition Non-negative matrix factorization (NMF) dictionary elements are used as an additional feature to improve the performance of the CNN model. Considering a number of applications involving eye state classification, such as in Parkinson’s disease detection, analysis of eye fatigue in 3D TVs, driver’s drowsiness detection, infant sleep-waking state identification, and classification of bipolar mood disorder and attention deficit hyperactivity, the proposed method was applied to Electroencephalography (EEG) data for classification of eye state. First, the spectrogram of EEG signal is obtained and used as an image input to CNN, simultaneously, the NMF feature is also fed to CNN. Further, both features are combined in fully connected layer of CNN architecture. The proposed method is compared with other existing methods for eye state detection and shows good classification accuracy with 96.16%. The prediction rate for the proposed method is 134 observations/second, which is suitable for brain–computer interface applications.

DeepEye: Link prediction in dynamic networks based on non-negative matrix factorization

A Non-negative Matrix Factorization (NMF)-based method is proposed to solve the link prediction problem in dynamic graphs. The method learns latent features from the temporal and topological structure of a dynamic network and can obtain higher prediction results. We present novel iterative rules to construct matrix factors that carry important network features and prove the convergence and correctness of these algorithms. Finally, we demonstrate how latent NMF features can express network dynamics efficiently rather than by static representation, thereby yielding better performance. The amalgamation of time and structural information makes the method achieve prediction results that are more accurate. Empirical results on real-world networks show that the proposed algorithm can achieve higher accuracy prediction results in dynamic networks in comparison to other algorithms.

Adaptive Optimal Kernel Time–Frequency Representation Technique for Partial Discharge Ultra-high-frequency Signals Classification

An intelligent system for automatic partial discharge pattern recognition is proposed using adaptive optimal kernel time-frequency representation and a fuzzy k-nearest neighbor classifier. The adaptive optimal kernel technique is employed to acquire the joint time-frequency information for partial discharge signals, which is characterized by the adaptive optimal kernel amplitude matrix. A new feature extraction algorithm, i.e., non-negative matrix factorization aided principal component analysis, is proposed to solve the difficulties of principal component analysis for feature extraction of partial discharge adaptive optimal kernel amplitude matrices due to the high dimensionality. Using an ultra-high frequency detector, 600 partial discharge signals sampled from 4 categories of typical artificial defect models in the laboratory are employed for testing. It is shown that the maximum classification accuracy of 94.33% is obtained considering different non-negative matrix factorization parameter r and various non-negative matrix factorization iterations T. Also, the classification performance of the non-negative matrix factorization–principal component analysis features is superior to that of principal component analysis features extracted from original partial discharge signals, two-dimensional non-negative matrix factorization features and phase-resolved partial discharge statistical operators. The proposed technique can be used for partial discharge pattern recognition based on ultra-high-frequency detection arrangements.

Discriminating Variable Star Candidates in Large Image Databases from the HiTS Survey Using NMF

Abstract New instruments and technologies are allowing the acquisition of large amounts of data from astronomical surveys. Nowadays there is a pressing need for autonomous methods to discriminate the interesting astronomical objects in the vast sky. The High Cadence Transient Survey (HiTS) project is an astronomical survey that is trying to find a rare transient event that occurs during the first instants of a supernova. In this paper we propose an autonomous method to discriminate stellar variability from the HiTS database, that uses a feature extraction scheme based on Non-negative matrix factorization (NMF). Using NMF, dictionaries of image prototypes that represent the data in a compact way are obtained. The projections of the dataset into these dictionaries are fed into a random forest classifier. NMF is compared with other feature extraction schemes, on a subset of 500,000 transient candidates from the HiTS survey. With NMF a better class separability at feature level is obtained which enhances the classification accuracy significantly. Using the NMF features less than 4% of the true stellar transients are lost, at a manageable false positive rate of 0.1%.

Local discriminant non-negative matrix factorization feature extraction for hyperspectral image classification

Non-negative matrix factorization (NMF) ignores both the local geometric structure of and the discriminative information contained in a data set. A manifold geometry-based NMF dimension reduction method called local discriminant NMF (LDNMF) is proposed in this paper. LDNMF preserves not only the non-negativity but also the local geometric structure and discriminative information of the data. The local geometric and discriminant structure of the data manifold can be characterized by a within-class graph and a between-class graph. An efficient multiplicative updating procedure is produced, and its global convergence is guaranteed theoretically. Experimental results on two hyperspectral image data sets show that the proposed LDNMF is a powerful and promising tool for extracting hyperspectral image features.

Application of non-negative matrix factorization to multispectral FLIM data analysis

Existing methods of interpreting fluorescence lifetime imaging microscopy (FLIM) images are based on comparing the intensity and lifetime values at each pixel with those of known fluorophores. This method becomes unwieldy and subjective in many practical applications where there are several fluorescing species contributing to the bulk fluorescence signal, and even more so in the case of multispectral FLIM. Non-negative matrix factorization (NMF) is a multivariate data analysis technique aimed at extracting non-negative signatures of pure components and their non-negative abundances from an additive mixture of those components. In this paper, we present the application of NMF to multispectral time-domain FLIM data to obtain a new set of FLIM features (relative abundance of constituent fluorophores). These features are more intuitive and easier to interpret than the standard fluorescence intensity and lifetime values. The proposed approach, unlike several FLIM data analysis methods, is not limited by the number of constituent fluorescing species or their possibly complex decay dynamics. Moreover, the new set of FLIM features can be obtained by processing raw multispectral FLIM intensity data, thereby rendering time deconvolution unnecessary and resulting in lesser computational time and relaxed SNR requirements. The performance of the NMF method was validated on simulated and experimental multispectral time-domain FLIM data. The NMF features were also compared against the standard intensity and lifetime features, in terms of their ability to discriminate between different types of atherosclerotic plaques.

도심 영상에서의 비음수행렬분해를 이용한 차량 인식

차량인식은 차량 후보영역 검출단계와 검출된 후보 영역에서 특징을 기반으로 차량을 검증하는 차량 검증단계로 나누어진다. 선형 변환 방식의 특징은 차원 감소 효과와 통계적인 특징을 지니게 되어, 이동이나 회전에 강인한 특성을 갖는다. 선형 변환 방식 중 비음수행렬분해(Non-negative Matrix Factorization, NMF)는 부분 기반 표현 방식으로 차량의 국소적인 특징을 기저벡터로 사용하여 희소성을 갖는 특징을 추출할 수 있기 때문에 도심영상에서 발생하는 차폐 영역에 따른 인식률 저하를 방지할 수 있다. 본 논문에서는 차량 인식에 적합한 NMF 특징 추출 방법을 제안하고, 인식률을 검증하였다. 또한 희소성 제약 조건을 이용하여 기저 벡터에 희소성을 가지는 SNMF(Sparse NMF)와 LVQ2(Learning Vector Quantization) 신경 회로망을 결합하여 차폐 영역에 대한 차량 인식 효율을 기존의 NMF를 이용한 방법과 비교하였다. NMF를 이용하는 특징은 차량이 혼재되어 차폐 영역이 빈번히 발생하는 도심에서도 강건한 특징임을 보였다. The vehicle recognition consists of two steps; the vehicle region detection step and the vehicle identification step based on the feature extracted from the detected region. Features using linear transformations have the effect of dimension reduction as well as represent statistical characteristics, and show the robustness in translation and rotation of objects. Among the linear transformations, the NMF(Non-negative Matrix Factorization) is one of part-based representation. Therefore, we can extract NMF features with sparsity and improve the vehicle recognition rate by the representation of local features of a car as a basis vector. In this paper, we propose a feature extraction using NMF suitable for the vehicle recognition, and verify the recognition rate with it. Also, we compared the vehicle recognition rate for the occluded area using the SNMF(sparse NMF) which has basis vectors with constraint and LVQ2 neural network. We showed that the feature through the proposed NMF is robust in the urban scene where occlusions are frequently occur.

Recognition of Nonprototypical Emotions in Reverberated and Noisy Speech by Nonnegative Matrix Factorization

We present a comprehensive study on the effect of reverberation and background noise on the recognition of nonprototypical emotions from speech. We carry out our evaluation on a single, well-defined task based on the FAU Aibo Emotion Corpus consisting of spontaneous children's speech, which was used in the INTERSPEECH 2009 Emotion Challenge, the first of its kind. Based on the challenge task, and relying on well-proven methodologies from the speech recognition domain, we derive test scenarios with realistic noise and reverberation conditions, including matched as well as mismatched condition training. As feature extraction based on supervised Nonnegative Matrix Factorization (NMF) has been proposed in automatic speech recognition for enhanced robustness, we introduce and evaluate different kinds of NMF-based features for emotion recognition. We conclude that NMF features can significantly contribute to the robustness of state-of-the-art emotion recognition engines in practical application scenarios where different noise and reverberation conditions have to be faced.

FONT CLASSIFICATION USING NMF WITH HIERARCHICAL CLUSTERING

The current paper proposes a font classification method for document images that uses non-negative matrix factorization (NMF), that is able to learn part-based representations of objects. The basic idea of the proposed method is based on the fact that the characteristics of each font are derived from parts of individual characters in each font rather than holistic textures. Spatial localities, parts composing of font images, are automatically extracted using NMF and, then, used as features representing each font. Using hierarchical clustering algorithm, these feature sets are generalized for font classification, resulting in the prototype templates construction. In both the prototype construction and font classification, earth mover's distance (EMD) is used as the distance metric, which is more suitable for the NMF feature space than Cosine or Euclidean distance. In the experimental results, the distribution of features and the appropriateness of the features specifying each font are investigated, and the results are compared with a related algorithm: principal component analysis (PCA). The proposed method is expected to improve the performance of optical character recognition (OCR), document indexing and retrieval systems, when such systems adopt a font classifier as a preprocessor.