Component Extraction Algorithm Research Articles

A modified hysteresis current controller with DFCEA for current harmonic mitigation using PV-SHAPF

ABSTRACT A Modified Hysteresis Current Controller (MHCC) with Dual Fundamental Component Extraction Algorithm (DFCEA) is presented in this paper. Its purpose is to improve the current harmonics mitigation performance of a three-level T-type inverter-based Shunt Active Power Filter (SHAPF) while operating under non-sinusoidal voltage conditions. The DFCEA is responsible for isolating the fundamental voltage and current components that are necessary for synchronizing each phase and producing the reference current. The existing dual fundamental component extraction technique is updated in order to create the optimized reference currents that are required by the proposed PV-linked T-Type inverter-based Shunt Active Power Filter (PV-SHAPF). This was done in order to meet the requirements of the PV-SHAPF. It is also anticipated that it is excellent at extracting reference currents even under situations of voltage that are not optimum. To maintain the dc-link capacitor voltage at 600 V, a control algorithm of the dc-link capacitor known as the ‘DC link voltage controller’ is used. Both control algorithms are tested in a three-phase T-Type inverter-based PV-SHAPF by simulating them in a MATLAB/Simulink environment. The performance of the MHCC-DFCEA that was designed to mitigate current harmonics was proven by the findings that were gathered.

Smart prototype for an electronic color sensor device for visual simultaneous detection of macrofuran based on a coated paper strip

Nowadays, in the clinical, pharmaceutical, and environmental sectors, the development of facile and sensitive analytical methods and/or innovative devices for the follow-up and detection of antibiotics and pharmaceutical formulations, in general, are urgently needed and still challenging. This work declared three vital applications for broad-spectrum nitrofurantoin (macrofuran) antibiotic detection and quantification: A colorimetric method, a coated paper strip-based nano-lanthanum complex prototype and fabrication of smart electronic color sensor device-based coated paper strips. The colorimetric method showed a significant response upon increasing the concentration of the nitrofurantoin in a range between (1.0–100.0 ng/mL) via a visual color change from orange-yellow to red colors degree with detection and quantification limits of 0.175 and 0.53 ng/mL, respectively, whereas the nano-lanthanum complex coated paper strip prototype showed qualitative on-site sensing for nitrofurantoin via naked eye color changes which can be detected anywhere. Moreover, a smart prototype for detecting macrofuran in the means of paper color change in the RGB color component extraction algorithm and the grayscale projection value processing algorithm was fabricated. The change in RGB color on the coated paper strip was detected using an electronic color sensor device. The developed colorimetric method, coated paper strip, and the electronic color sensor device prototype exhibited fast, simple, costless, and selective towards macrofuran over the competing analyzed. As well as, showed good applicability in the different real samples spiked with different concentrations of macrofuran.Graphical abstract

Debris Flow Prediction Based on the Fast Multiple Principal Component Extraction and Optimized Broad Learning

In the current research of debris flow geological disaster prediction, determining reasonable disaster-inducing factors and ensuring the accuracy and rapidity of the prediction model are considered vital issues, and also, essential foundations for disaster early warning and disaster prevention and mitigation. Aiming at the problems of low prediction accuracy and long prediction time in the current debris flow research, firstly, six debris flow impact factors were selected relying on the fast multiple principal component extraction (FMPCE) algorithm, including rainfall, slope gradient, gully bed gradient, relative height difference, soil moisture content and pore water pressure. Next, based on the broad learning (BL) algorithm, the debris flow prediction model based on FMPCE and the optimized BL is established with the input of debris flow-inducing factors and the output of debris flow probability. Then the model is optimized using matrix stochastic approximate singular value decomposition (SVD), and the debris flow disaster prediction model, based on SVDBL, is constructed. The prediction results of the optimized model are compared with those of the gradient descent optimized the BP neural network model(GD-BP), Support Vector Machines model(SVM) based on grid search and BL model. The results show that the accuracy of SVDBL is 7.5% higher than that of GD-BP, 3% higher than that of SVM and 0.5% higher than that of BL. The RMSE sum of SVDBL was 0.05870, 0.0478 and 0.0227 less than that of GD-BPSVM and BL, respectively; the MAPE sum of SVDBL was 1.95%, 1.66% and 0.49% less than that of GD-BPSVM and BL; the AUC values of SVDBL were 12.75%, 7.64% and 2.79% higher than those of the above three models, respectively. In addition, the input dataset is expanded to compare the training time of each model. The simulation results show that the prediction accuracy of this model is the highest and the training time is the shortest after the dataset is expanded. This study shows that the BL can be used for debris flow prediction, and can also provide references for disaster early warning and prevention.

Vision-assisted BIM reconstruction from 3D LiDAR point clouds for MEP scenes

Mechanical, electrical and plumbing (MEP) system provides various services and creates comfortable environments to residents in cities. To enhance the management efficiency of the highly complex MEP system, as-built building information modeling (BIM) is increasingly adopted in the world. Currently, as-built BIMs are mostly drawn manually by modelers in BIM modeling software referring to point clouds or on-site photos, which is time consuming and labor intensive. This study presents a novel fused BIM reconstruction approach for MEP scenes. The proposed approach makes the best of the rich semantic information provided by images and accurate geometry information provided by 3D LiDAR point clouds. Firstly, a state-of-the-art deep learning model focusing on semantic segmentation is fine-tuned for the MEP dataset, and then RGB images collected with depth camera are segmented with the well-trained model. Secondly, taking the segmented images and the corresponding depth images as input, a semantic-rich 3D map is generated. Thirdly, an instance-aware component extraction algorithm in LiDAR point clouds given approximate object distribution in 3D space is developed. In the component extraction algorithm, a label transfer technique is proposed to firstly determine the rough locations of targeting objects in LiDAR point clouds. Then, accurate component locations are determined for three types of components including irregular shaped components, regular shaped components, and secondary components attached to walls. Finally, the BIM model is reconstructed based on component extraction results. To validate the proposed technique, experiments were conducted in four MEP rooms in a water treatment plant in Hong Kong. It is demonstrated that the proposed technique is more accurate and more efficient with wider range of applications compared to previous BIM reconstruction methods.

Equivalent characteristic impedance based hybrid-HVDC transmission line fault location

Aiming at the transmission line fault in Hybrid-HVDC system, this paper divides the fault evolution process into different stages according to the difference of transition resistance at the fault point and the conduction state change of power electronic devices, emphatically analyzes the conversion conditions between different stages, and proposes an adaptive criterion for judging the moment when the fault evolves into AC feed-in stage. Based on the characteristic component of electrical quantities at AC feed-in stage, this paper defines the equivalent characteristic impedance at any point of transmission line, derives the functional relationship between the fault distance and the equivalent characteristic impedance at inverter side DC bus, and locates the fault point according to this relationship. In order to extract the characteristic component of electrical quantities, a characteristic component extraction algorithm is studied, it can effectively eliminate the influence of constant DC component and decaying aperiodic component. This paper builds the Hybrid-HVDC system simulation model and carries out simulation experiments on PSCAD/EMTDC, and compares the proposed fault location principle with the existing methods. The simulation results verify the effectiveness of the adaptive criterion for judging the AC feed-in stage starting moment and the proposed fault location principle.

A smartphone-based rapid quantitative detection platform for lateral flow strip of human chorionic gonadotropin with optimized image algorithm

Colloidal gold immunochromatographic test strip has been widely used as a rapid, simple and low-cost correct detection technology. However, its detection is often qualitative or semi-quantitative, which limits its clinical application to some extent. Herein, a portable test strip quantitative detection device based on smartphone to detect human chorionic gonadotropin (HCG) is developed. In experiment, a colloidal gold HCG detection strip based on antigen antibody immune response is constructed, and the quantitative results of three different image processing methods on the same strip detection are compared, including the threshold processing algorithm based on location information, the RGB color component extraction algorithm and the grayscale projection value processing algorithm, the results show that the last algorithm can realize the best recognition of the region of interest of strip. The mobile phone application software (App) based on this design shows that the detection limit of constructed colloidal gold HCG strip is 3 ng/mL with a linear range of 6–300 ng/mL. The detection result of real urine sample is consistent with the spiked concentration (R2 = 0.988), indicating that the concentration of HCG can be accurately measured in urine with this method, presenting the potential for instant diagnosis.

Writer Identification of Lampung Handwritten Documents Based on Selected Characters

Writer identification is a sub-field in handwriting recognition which its objective is to determine the identity of the writer based on handwriting input. The goal is usually for forensic purposes such as finding the perpetrators of crimes that leave traces of evidence in the form of written messages. In addition, writer identification can also be used to determine the identity of a historical actor if he or she leaves a valuable written artefact. The object of this research is the traditional character of the Lampung region which is so-called Had Lampung by the local community. The traditional character of Lampung consists of 20 main characters and 12 diacritics. Based on selected characters, the writer will be recognized using the Principal Component Analysis (PCA) feature. PCA is one linear feature extraction method of an object in pattern recognition. The PCA algorithm consists of several stages, namely the calculation of the average dataset, the subtraction of the vector dataset with averages, the calculation of covariance, the calculation of eigenvectors and eigenvalues, eigenvector reduction, and the projection of the dataset against reduced eigenvector space. PCA in this paper is used as a feature in image recognition. The dataset utilized in this study is the Lampung Dataset which is a handwritten character recognition (HWCR) dataset. Lampung Dataset consists of 82 Lampung handwritten documents. All Lampung character images in the dataset were extracted from these documents using the connected component extraction algorithm and eventually generated 32,140 images. Furthermore, these images are converted into grayscale images. In this research, as many as 12,500 grayscale images of Lampung handwriting characters were chosen to represent 82 different writers. This data is employed as training and testing data on the proposed method. The highest accuracy of the identification of the writer using this PCA feature is 82.92%, while the lowest accuracy is 28.29%.

A novel iterative learning control method and control system design for active magnetic bearing rotor imbalance of primary helium circulator in high-temperature gas-cooled reactor

As one of the key technologies of high-temperature gas-cooled reactor, primary helium circulator–equipped active magnetic bearing provides driving force for primary helium cooling system. However, repetitive periodic vibration produced by rotor imbalance may introduce risks to primary helium circulator (even for high-temperature gas-cooled reactors). First, this article analyzes a periodic component extraction algorithm which is widely used in active magnetic bearing rotor unbalance control methods and points out the problem that the periodic component extraction algorithm occupies numerous computing resources which cannot satisfy the real-time request of active magnetic bearing control system. Then, a novel iterative learning control algorithm based on the iteration before last iteration of system information (iterative learning control-2) and a plug-in parallel control mechanism based on the existing control system are put forward, meanwhile, an integrated independent distributed active magnetic bearing control system is designed to solve the problem. Finally, both the simulation and experiment are carried out, respectively. The corresponding results show that the control method and control system proposed in this article have significant suppression effect on the repetitive periodic vibration of the active magnetic bearing system without degrading the real-time requirement and can provide important technical support for the safe and stable operation of the primary helium circulator in high-temperature gas-cooled reactor.

Adaptive Recognition of Motion Decomposition Image Based on Adaptive Principal Component Extraction Algorithm

Moving target detection and recognition methods are the foundation and key of modern intelligent video recognition systems. It combines advanced technologies in many fields such as image processing, pattern recognition, and artificial intelligence, and is a research hotspot in computer vision technology. Therefore, it is of great significance to study moving target detection and recognition algorithms. The non-local image extraction algorithm proposed in this paper uses an adaptive clustering method to perform fine cluster analysis on non-local image blocks with different feature types. Through the step-by-step principal component approximation method, we carefully find the features in each class. This progressive principal component approximation implements singular value hard threshold processing based on the Marchenko-Pastur (MP) theorem to select the main part of the feature, and uses special soft thresholds in the principal component transform domain to further improve the extraction performance. The Lower Bound-Based Within-Class Maximum Division (LBWCMD) is proposed, and this method is used as a preprocessing step of robust principal component analysis in moving target detection. This article applies LBWCMD to the video frame set based on the position information of the moving target, and the obtained frame subset meets the signal requirements of Robust Principle Component Analysis (RPCA) to the greatest extent. On this basis, we add frames with smaller motion amplitudes to each frame subset to increase the proportion of background pixels in each subset. Frame set division and low-rank decomposition realize the detection of moving targets under a unified framework. The detection rate of the proposed method is higher than that of the current popular methods in sports video data sets, and the detection accuracy is improved compared with the original RPCA method.

Research on the Switching Arc Loss of on-Load Tap Changer

The on-load tap changer (OLTC) switching arc will affect the life of the contact, so it is necessary to test its working condition. Based on DQ coordinate transformation, fundamental wave component extraction, and wavelet algorithm, proposed an OLTC arc detection method based on energy conservation theory in this paper. Used the fundamental wave component extraction algorithm extracted the transformer excitation parameters, and established the OLTC equivalent circuit by combining the Habedank arc model. Compared the differences between the follow effect and frequency calculation of simulated and measured waveforms, which were caused by the five orthogonal transformation algorithms, including Transport-Delay, Hilbert Transformation, Inverse Park Transformation, SOGI and Improved SOGI. The DQ coordinate transformation based on the Transport-Delay orthogonal converter is determined to calculate the active power of each part of the circuit. According to the principle of energy conservation, obtained the data of arcing loss under nine simulated conditions. Based on the simulation circuit, built the OLTC prototype and dSPACE hardware in loop simulation platform, and measured three load conditions. The sampling signal was filtered by the wavelet algorithm. Considering the measurement and calculation errors, the simulation and measured results show that the actual arcing power has a consistent change with the calculated value between different working conditions, and the maximum difference between simulation results and measured results is thirty times and ten times. It can distinguish the difference of arcing energy of diverter switch under different working conditions and has the measurability. Meanwhile, it verifies the effectiveness of this study, hoping to provide a theoretical basis for developing OLTC on-line contact detection system.

Variational Bayesian Subgroup Adaptive Sparse Component Extraction for Diagnostic Imaging System

A novel unsupervised sparse component extraction algorithm is proposed for detecting micro defects while employing a thermography imaging system. The proposed approach is developed using the variational Bayesian framework. This enables a fully automated determination of the model parameters and bypasses the need for human intervention in manually selecting the appropriate image contrast frames. An internal subsparse grouping mechanism and adaptive fine-tuning strategy have been built to control the sparsity of the solution. The proposed algorithm is computationally affordable and yields a high-accuracy objective performance. Experimental tests on both artificial and natural defects have been conducted to verify the efficacy of the proposed method.

Operation of Three-Level Inverter-Based Shunt Active Power Filter Under Nonideal Grid Voltage Conditions With Dual Fundamental Component Extraction

In this paper, a new reference current generation method is proposed for effective harmonics mitigation and reactive power compensation of three-level neutral-point diode clamped inverter-based shunt active power filter (SAPF) under nonideal grid voltage conditions. The proposed method is named as dual fundamental component extraction algorithm. In operation, the proposed algorithm extracts at the same time, the desired fundamental current and voltage components for generating reference current and synchronization phases, respectively. As a result, the proposed algorithm is able to generate reference current that ensures in phase operation of SAPF with the operating power system, without depending on any phase-locked loop elements. Besides, the proposed algorithm employs self-tuning filter (STF) for accurate computation of the fundamental components. Design concept and effectiveness of the proposed algorithm are thoroughly studied and evaluated in MATLAB–Simulink. Additionally, a laboratory prototype utilizing TMS320F28335 digital signal processor is built to validate its feasibility. Encouraging findings obtained from both simulation and experimental works demonstrate effectiveness of the proposed algorithm under both ideal and nonideal grid voltage conditions.

A Generalized Minor Component Extraction Algorithm and Its Analysis

Generalized minor component analysis (GMCA) is of great use in modern signal processing. The GMCA algorithms can be simplified to extract the minor generalized eigenvector of the autocorrelation input matrices pencil. In contrast to batching methods, the Hebbian-rule-based algorithm can extract the minor generalized eigenvector online. Few Hebbian-rule-based GMCA algorithms have been reported in the literature, and most of them are not self-stabilizing. Thus, a novel algorithm for GMCA, which is advantageous in terms of good convergence speed, self-stabilizing property, and multiple generalized minor component extraction in sequence, is proposed in this paper. A theoretical analysis verifies these properties via matrix theory and the deterministic discrete-time method. Numerical simulations are conducted to further demonstrate the advantages of the proposed algorithm.

A recursive least square algorithm for online kernel principal component extraction

The online extraction of kernel principal components has gained increased attention, and several algorithms proposed recently explore kernelized versions of the generalized Hebbian algorithm (GHA) [1], a well-known principal component analysis (PCA) extraction rule. Consequently, the convergence speed of such algorithms and the accuracy of the extracted components are highly dependent on a proper choice of the learning rate, a problem dependent factor. This paper proposes a new online fixed-point kernel principal component extraction algorithm, exploring the minimization of a recursive least-square error function, conjugated with an approximated deflation transform using component estimates obtained by the algorithm, implicitly applied upon data. The proposed technique automatically builds a concise dictionary to expand kernel components, involves simple recursive equations to dynamically define a specific learning rate to each component under extraction, and has a linear computational complexity regarding dictionary size. As compared to state-of-art kernel principal component extraction algorithms, results show improved convergence speed and accuracy of the components produced by the proposed method in five open-access databases.

Abstract 15: Metabolomic profiling of cell death in human lung cancer cells by a novel digitoxin analog

Abstract Background: Digitoxin, a cardiac glycoside had shown considerable promise as an anti-cancer therapeutic. However, doses of digitoxin required to inhibit cancer cell proliferation lead to cardiotoxic side effects. We have previously shown that a novel analog of digitoxin, MonoD, activates apoptosis in human lung cancer cells at much lower concentrations as compared to digitoxin. We postulate that treatment with MonoD leads to distinct metabolic signatures on lung cancer cells as compared to digitoxin, which will help delineate the pathways that dictate MonoD action. Experimental Approach: NCI-H460 cells from identical passages were used for metabolite analysis using an optimized version of the methanol/water (80:20) protocol for metabolite extraction. The supernatant containing metabolites were suspended in 0.1% formic acid spiked with an internal standard for Q-Exactive analysis. The data was analyzed using SIEVETM 2.0 software with the component extraction algorithm that has been specifically designed for optimal data analysis in untargeted metabolomics experiments. Metabolites that were found to be significantly different between treatment groups were detected via Chemspider/HMDB database search. Results: We observed several metabolites along the prostaglandin synthesis pathway which were downregulated in MonoD treated lung cancer cells. Prostaglandins play a key role in the generation of the inflammatory response. Chronic inflammation has been shown to be a causative factor in a variety of cancers. Cyclooxygenase-2 (COX-2) derived prostaglandin has been previously shown to promote tumor growth by binding its receptors and activating signaling pathways which control cell proliferation, migration, apoptosis, and angiogenesis. Using the metabolite profiling data, we posited that COX-2, a key enzyme involved in prostaglandin synthesis, in addition to NFκB, a key transcription factor of proinflammatory signaling pathway are downregulated with MonoD treatment. Immunoblotting confirmed the downregulation of COX-2 and NFκB in MonoD-treated cells. Conclusion: Using metabolite profiling, we have elucidated an important mode of action for MonoD in lung cancer cells, and shown that MonoD exerts its anti-tumorigenic properties by downregulating key factors involved in inflammatory response. Citation Format: Yogesh Kulkarni, Neelam Azad, Vivek Kaushik, Juan Sebastian Yakisich, Rajkumar Venkatadri, Clayton Wright, Yon Rojanasakul, George O’Doherty, Anand Krishnan V Iyer. Metabolomic profiling of cell death in human lung cancer cells by a novel digitoxin analog. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 15.

Dual Function of Unified Adaptive Linear Neurons Based Fundamental Component Extraction Algorithm for Shunt Active Power Filter Operation

Unified Adaptive Linear Neurons (ADALINEs) based fundamental component extraction algorithm is presented for better operation of a Shunt Active Power Filter (SAPF). The proposed algorithm is developed for dual functionality. The first role is to generate a reference current for the SAPF. Whilst, the second function is to coordinate the phase of the generated reference current with the phase of any operating power system. Consequently, the operation of the SAPF will be synchronized with the operating power system. Hence, it replaces the necessity of using conventional synchronization algorithms such as Phase-Locked Loops (PLLs) and Zero-Crossing Detectors (ZCDs). In this work, the proposed algorithm is executed using an integration of two ADALINEs based fundamental current and voltage extraction algorithms. Also, it implements the modified on-line Widrow-Hoff (W-H) weight updating algorithm for high speed extraction process. The effectiveness of the proposed algorithm has been verified in both simulation and experimental works.

오디오 컨텐츠를 위한 비음수 행렬 분해 기법 기반의 실시간 단일채널 배경 잡음 추출 기법

본 논문에서는 비음수 행렬 분해 (NMF) 기법을 이용하여 단일 채널에서 배경음 성분을 추출하는 알고리즘에 대해 서술한다. 이러한 배경음 성분 추출은 오디오 업믹싱 시스템을 고려하여 개발되었으며, 기존의 연구를 통하여 분리된 배경음 신호가 업믹싱 시스템에 적용될 경우 공간감을 향상시킬 수 있다는 사실이 이미 확인된 바 있다. 다만 기존의 기법은 음향 신호를 모두 축적하여 일괄적으로 처리해야 한다는 단점이 있어, 스트리밍 시스템이나 디지털 시그널 프로세서 (DSP) 등을 이용한 시스템에서 사용되기 어렵다. 본 논문에서는 이를 해소하기 위하여 실시간 비음수 행렬 분해 기법을 이용한 배경음 추출 시스템을 고안하여 실험하였다. 실험에서 처리된 음원을 스펙트럼 평활도를 이용하여 분석한 결과, 고안된 배경음 추출 시스템이 기존의 일괄 추출 시스템과 유사한 정도로 배경음 성분을 추출했음을 확인할 수 있었다. In this paper, monaural ambient component extraction algorithm based on nonnegative matrix factorization (NMF) is described. The ambience component extraction algorithm in this paper is developed for audio upmixing system; Recent researches have shown that they can enhance listener envelopment if the extracted ambient signal is applied into the multichannel audio upmixing system. However, the conventional method stores all of the audio signal and processes all at once, so it cannot be applied to streaming system and digital signal processor (DSP) system. In this paper, the ambient component extraction algorithm based on on-line nonnegative matrix factorization is developed and evaluated to solve the problem. As a result of analysis of the processed signal with spectral flatness measures in the experiment, it was shown that the developed system can extract the ambient signal similarly with the conventional batch process system.

Noisy Component Extraction (NoiCE)

<para xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> To achieve efficient blind source extraction (BSE) from noisy mixtures, we propose a noisy component extraction (NoiCE) algorithm that combines standard BSE and a cascaded nonlinear adaptive estimator. There are no assumptions of statistical independence, and also as a byproduct of BSE after deflation, we may also obtain asymptotic identification of the <emphasis emphasistype="italic">a priori</emphasis>unknown observation noise sources. By yielding an asymptotically efficient estimator in the presence of an unknown observation noise, the proposed algorithm may also be viewed as a robust approach to NoiCE. Simulations on both synthetic and real-world data confirm the validity of the proposed algorithm in noisy mixing environments. </para>

Applications of Non-linear Component Extraction to Spectrogram Representations of Auditory Data.

Event Abstract Back to Event Applications of Non-linear Component Extraction to Spectrogram Representations of Auditory Data. Jorg Bornschein1* and Jorg Lucke1 1 Frankfurt Institute for Advanced Studies, Germany The state-of-the-art in component extraction for many types of data is based on variants of models such as principle component analysis (PCA), independent component analysis (ICA), sparse coding (SC), factor analysis (FA), or non-negative matrix factorization (NMF). These models are linear in the sense that they assume the data to consist of linear super-positions of hidden causes, i.e., these models try to explain the data with linear super-positions of generative fields. This assumption becomes obvious in the generative interpretation of these models [1].For many types of data, the assumption of linear component super-positions represents a good approximation. An example is the super-position of air-pressure waveforms. In contrast, we here study auditory data represented in the frequency domain. We consider data similar to those processed by the human audio system just after the cochlea. Such data is closely aligned with the log-power-spectrogram representations of auditory signals. It is long known that the super-position of data components in these data is non-linear and well approximated by a point-wise maximum of the individual spectrograms [2].For component extraction from auditory spectrogram data we therefore investigate learning algorithms based on a class of generative models that assume a non-linear superposition of data components. The component extraction algorithm of Maximal Causes Analysis (MCA; [3]) assumes a maximum combination where other algorithms use the sum. Training such non-linear models is, in general, computationally expensive but can be made feasible using approximation schemes based on Expectation Maximization (EM). Here we apply an EM approximation scheme that is based on the pre-selection of the most probable causes for every data-point. The approximation results in approximate maximum likelihood solutions, reduces the computational complexity significantly while at the same time allowing for an efficient and parallelized implementation running on clustered compute nodes. To evaluate the applicability of non-linear component extraction to auditory spectrogram data, we generated training data by randomly choosing and linearly mixing waveforms from a set of 10 different phonemes (sampled at 8000Hz). We then applied an MCA algorithm based on EM and pre-selection. The algorithm was presented only the log-spectrograms of the mixed signals. Assuming Gaussian noise the algorithm was able to extract the log-spectrograms of the individual phonemes. We obtained similar results for different forms of phoneme mixtures including mixtures of always three randomly chosen phonemes.

Global convergence of an adaptive minor component extraction algorithm

The convergence of neural networks minor component analysis (MCA) learning algorithms is crucial for practical applications. In this paper, we will analyze the global convergence of an adaptive minor component extraction algorithm via a corresponding deterministic discrete time (DDT) system. It is shown that if the learning rate satisfies certain conditions, almost all the trajectories of the DDT system are bounded and converge to minor component of the autocorrelation matrix of input data. Simulations are carried out to illustrate the results achieved.