• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.13 no.12
• /
• pp.6043-6062
• /
• 2019

Low-rank matrix decomposition has shown its capability in many applications such as image in-painting, de-noising, background reconstruction and defect detection etc. In this paper, we consider the texture background of rail track images and the sparse foreground of the defects to construct a low-rank matrix decomposition model with block sparsity for defect inspection of rail tracks, which jointly minimizes the nuclear norm and the 2-1 norm. Similar to ADM, an alternative method is proposed in this study to solve the optimization problem. After image decomposition, the defect areas in the resulting low-rank image will form dark stripes that horizontally cross the entire image, indicating the preciselocations of the defects. Finally, a two-stage defect extraction method is proposed to locate the defect areas. The experimental results of the two datasets show that our algorithm achieved better performance compared with other methods.

• The Journal of the Acoustical Society of Korea
• /
• v.21 no.5
• /
• pp.494-502
• /
• 2002

In this paper, the Low Rank Approximation (LRA) method to suppress the interference of signals from temporal fluctuations is applied. The reverberation signals and temporally fluctuating signals are separated from the measured data using the Ink. The Singular value decomposition (SVD) method is applied to extract the low rank and the temporally stable reverberation was extracted using the LRA. The reverberation suppression is performed on the LRA residual value obtained by removing the approximate reverberation signals. In overall, the method can be applied to the suppression of reververation in active sonar system as well as to the modeling of reverberation.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.46 no.5
• /
• pp.15-24
• /
• 2009

Compressed sensing addresses the recovery of a sparse vector from its few linear measurements. Recently, the success for the vector case has been extended to the matrix case. Compressed sensing of low-rank matrices solves the ill-posed inverse problem with fie low-rank prior. The problem can be formulated as either the rank minimization or the low-rank approximation. In this paper, we survey recently proposed efficient algorithms to solve these two formulations.

• ETRI Journal
• /
• v.36 no.1
• /
• pp.167-170
• /
• 2014

In this letter, we propose an unsupervised framework for speech noise reduction based on the recent development of low-rank and sparse matrix decomposition. The proposed framework directly separates the speech signal from noisy speech by decomposing the noisy speech spectrogram into three submatrices: the noise structure matrix, the clean speech structure matrix, and the residual noise matrix. Evaluations on the Noisex-92 dataset show that the proposed method achieves a signal-to-distortion ratio approximately 2.48 dB and 3.23 dB higher than that of the robust principal component analysis method and the non-negative matrix factorization method, respectively, when the input SNR is -5 dB.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.11 no.9
• /
• pp.4510-4526
• /
• 2017

Most available methods of facial gender recognition work well under a constrained situation, but the performances of these methods have decreased significantly when they are implemented under unconstrained environments. In this paper, a method via low-rank and collaborative representation is proposed for facial gender recognition in the wild. Firstly, the low-rank decomposition is applied to the face image to minimize the negative effect caused by various corruptions and dynamical illuminations in an unconstrained environment. And, we employ the collaborative representation to be as the classifier, which using the much weaker $l_2-norm$ sparsity constraint to achieve similar classification results but with significantly lower complexity. The proposed method combines the low-rank and collaborative representation to an organic whole to solve the task of facial gender recognition under unconstrained environments. Extensive experiments on three benchmarks including AR, CAS-PERL and YouTube are conducted to show the effectiveness of the proposed method. Compared with several state-of-the-art algorithms, our method has overwhelming superiority in the aspects of accuracy and robustness.

• ETRI Journal
• /
• v.39 no.1
• /
• pp.21-29
• /
• 2017

This paper addresses the problem of unsupervised speech separation based on robust non-negative matrix factorization (RNMF) with ${\beta}$-divergence, when neither speech nor noise training data is available beforehand. We propose a robust version of non-negative matrix factorization, inspired by the recently developed sparse and low-rank decomposition, in which the data matrix is decomposed into the sum of a low-rank matrix and a sparse matrix. Efficient multiplicative update rules to minimize the ${\beta}$-divergence-based cost function are derived. A convolutional extension of the proposed algorithm is also proposed, which considers the time dependency of the non-negative noise bases. Experimental speech separation results show that the proposed convolutional RNMF successfully separates the repeating time-varying spectral structures from the magnitude spectrum of the mixture, and does so without any prior training.

• Journal of Broadcast Engineering
• /
• v.26 no.2
• /
• pp.125-131
• /
• 2021

In recent years, Convolutional Neural Networks (CNNs) have achieved outstanding performance in the fields of computer vision such as image classification, object detection, visual quality enhancement, etc. However, as huge amount of computation and memory are required in CNN models, there is a limitation in the application of CNN to low-power environments such as mobile or IoT devices. Therefore, the need for neural network compression to reduce the model size while keeping the task performance as much as possible has been emerging. In this paper, we propose a method to compress CNN models by combining matrix decomposition methods of LR (Low-Rank) approximation and CP (Canonical Polyadic) decomposition. Unlike conventional methods that apply one matrix decomposition method to CNN models, we selectively apply two decomposition methods depending on the layer types of CNN to enhance the compression performance. To evaluate the performance of the proposed method, we use the models for image classification such as VGG-16, RestNet50 and MobileNetV2 models. The experimental results show that the proposed method gives improved classification performance at the same range of 1.5 to 12.1 times compression ratio than the existing method that applies only the LR approximation.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.6
• /
• pp.2634-2648
• /
• 2020

In this paper, an adaptive sparse singular value decomposition (ASSVD) algorithm is proposed to estimate the signal matrix when only one data matrix is observed and there is high dimensional white noise, in which we assume that the signal matrix is low-rank and has sparse singular vectors, i.e. it is a simultaneously low-rank and sparse matrix. It is a structured matrix since the non-zero entries are confined on some small blocks. The proposed algorithm estimates the singular values and vectors separable by exploring the structure of singular vectors, in which the recent developments in Random Matrix Theory known as anisotropic Marchenko-Pastur law are used. And then we prove that when the signal is strong in the sense that the signal to noise ratio is above some threshold, our estimator is consistent and outperforms over many state-of-the-art algorithms. Moreover, our estimator is adaptive to the data set and does not require the variance of the noise to be known or estimated. Numerical simulations indicate that ASSVD still works well when the signal matrix is not very sparse.

• Clean Technology
• /
• v.22 no.4
• /
• pp.299-307
• /
• 2016

These days, coal is one of the most important energy resources used for transportation, industry, and electricity. There are two types of coal: high-rank and low-rank. Low-rank coal has a low calorific value and contains large amounts of useless moisture. The quality of low-rank coal can be increased by simple drying technology and it needs to be stabilized by hydrocarbons (e.g. palm acid oil, PAO) to prevent spontaneous combustion and moisture re-adsorption. Spontaneous combustion becomes a major problem during coal mining, storage, and transportation. It can involve the loss of life, property, and economic value; reduce the quality of the coal; and increase greenhouse gas emissions. Besides spontaneous combustion, moisture re-adsorption also leads to a decrease in quality of the coal due to its lower heating value. In this work, PAO was used for additive to stabilize the upgraded coal. The objectives of the experiments were to determine the stabilization characteristic of coal by analyzing the behavior of upgraded coal by drying and PAO addition regarding crossing-point temperature of coal, the moisture behavior of briquette coal, and thermal decomposition behavior of coal.

• ETRI Journal
• /
• v.40 no.4
• /
• pp.421-434
• /
• 2018

The application of deep neural networks (DNNs) to connect the world with cyber physical systems (CPSs) has attracted much attention. However, DNNs require a large amount of memory and computational cost, which hinders their use in the relatively low-end smart devices that are widely used in CPSs. In this paper, we aim to determine whether DNNs can be efficiently deployed and operated in low-end smart devices. To do this, we develop a method to reduce the memory requirement of DNNs and increase the inference speed, while maintaining the performance (for example, accuracy) close to the original level. The parameters of DNNs are decomposed using a hybrid of canonical polyadic-singular value decomposition, approximated using a tensor power method, and fine-tuned by performing iterative one-shot hybrid fine-tuning to recover from a decreased accuracy. In this study, we evaluate our method on frequently used networks. We also present results from extensive experiments on the effects of several fine-tuning methods, the importance of iterative fine-tuning, and decomposition techniques. We demonstrate the effectiveness of the proposed method by deploying compressed networks in smartphones.