Component Analysis Based Feature Extraction Research Articles

Probabilistic Principle Component Analysis based Feature Extraction of Embedded System Applications with Deep Neural Network based Implementation in FPGA

The study of hardware and software systems is of major are very important advent in new devices for communication and progress in system of security. In fast pace mobile and embedded devices application in every day’s life leads some new emerging area for research in data mining field. In this we have some technologies which have demand and error free using the principle of component of PPCA. For Embedded system the applications of PCA is basically applied initially for the lessen the having different qualities especially being to simple of the data. PPCA which have the updated version of PCA which is surveyed by similarity measure. In this work, experiments are extensively carried out, using a FPGA based light weight cryptographic data set having benchmark set to check and illustrate the viability, competence, litheness which are reconfigurable embedded system which are having data mining . Which have FPGA are reconfigurable for the computing architectures for hardware and in neural network. FPGA using the multilayer Cascaded for neural network which are forward in nature (CFFNN) and Deep Neural Network also called as DNN with a huge neuron is still a thought-provoking task. This shortcoming leads to elect the FPGA capacity for a particular application we have used the method of implementation which has two neural network have been implemented and compared , namely, CFFNN and DNN. It can be shown that for reconfigurable embedded system, PPCA based data mining and Machine learning based realization can give more speed up less iteration and more space savings when we have compared it with the static conventional version.

A Novel Approach for Detection of Moving Objects in Complex Scenes Using Fuzzy Colour Difference Histogram

Background subtraction schematic is widely used for motion detection. For effective automation of this process, a robust algorithm with high accuracy is needed. One of the major challenges of such algorithms is the identification of objects from an environment with composite elements that may be a dynamic background, frames with a camouflaged background and foreground pixels, and consecutive frames with varying illumination. The existing system uses a multi-color space histogram superposition principle having the biggest challenge of choosing appropriate color components in suitable proportion. Overcoming this challenge, a novel approach, MODITBS, processed in a differential domain, is proposed. A fuzzified color difference histogram-based background modeling is done to significantly deal with complex background scenes followed by principal component analysis-based feature extraction. The foreground objects detected are enhanced using a Kalman filter. The results show that MODITBS attains an accuracy of 95.16% in comparison to the existing system having an accuracy of 91.25%.

Robust Regressive Feature Extraction Based Relevance Vector Margin Boosting For Aerial Images Scene Classification

Aerial image scene classification is a key problem to be resolved in image processing. Many research works have been designed for carried outing scene classification. But, accuracy of existing scene classification was lower. In order to overcome such limitation, a Robust Regressive Feature Extraction Based Relevance Vector Margin Boosting Scene Classification (RRFERVMBSC) Technique is proposed. The RRFE-RVMBSC technique is designed for improving the classification performance of aerial images with minimal time. The RRFERVMBSC technique comprises two main processes namely feature extraction and classification. Initially, RRFE-RVMBSC technique gets number of aerial images as input. After taking input, Robust Regressive Independent Component Analysis Based Feature Extraction process is performed in order to extract the features i.e. shape, color, texture and size from aerial image. After completing feature extraction process, RRFERVMBSC technique carried outs Ensembled Relevance Vector Margin Boosting Classification (ERVMBC) where all the input aerial images are classified into multiple classes with higher accuracy. The RRFE-RVMBSC technique constructs a strong classifier by reducing the training error of weak RVM classifier for effectual aerial images scene categorization. The RRFERVMBSC technique accomplishes simulation work using parameters such as feature extraction time classification accuracy and false positive rate with respect to number of aerial images.

Unsupervised Machine Learning-Based Detection of Covert Data Integrity Assault in Smart Grid Networks Utilizing Isolation Forest

Being one of the most multifaceted cyber-physical systems, smart grids (SGs) are arguably more prone to cyber-threats. A covert data integrity assault (CDIA) on a communications network may be lethal to the reliability and safety of SG operations. They are intelligently designed to sidestep the traditional bad data detector in power control centers, and this type of assault can compromise the integrity of the data, causing a false estimation of the state that further severely distresses the entire power system operation. In this paper, we propose an unsupervised machine learning-based scheme to detect CDIAs in SG communications networks utilizing non-labeled data. The proposed scheme employs a state-of-the-art algorithm, called isolation forest, and detects CDIAs based on the hypothesis that the assault has the shortest average path length in a constructed random forest. To tackle the dimensionality issue from the growth in power systems, we use a principal component analysis-based feature extraction technique. The evaluation of the proposed scheme is carried out through standard IEEE 14-bus, 39-bus, 57-bus, and 118-bus systems. The simulation results show that the proposed scheme is proficient at handling non-labeled historical measurement datasets and results in a significant improvement in attack detection accuracy.

Comparative Analysis of Illumination Normalizations on Principal Component Analysis Based Feature Extraction for Face Recognition

Principle Component Analysis (PCA) is an appearance-based technique for extraction of feature extraction that is commonly used in computer vision and image processing. This technique suffers from illumination variations, thus knowing which illumination control method to be used in PCA-based face recognition system is very important. This paper applies three selected normalization techniques; Discrete Cosine Transform (DCT), Adaptive Histogram Equalization (AHE) and Contrast Limited Adaptive Histogram Equalization (CLAHE) to normalize face images. PCA was further used to extract features from the normalized face images. Euclidean distance was used to classify extracted features. The best recognition accuracy of 91.84% was obtained in DCT for ORL Database, while the best accuracy of 76% was achieved in DCT for FERET Database. The highest FAR of 0.9 was achieved in DCT for ORL Database, while the highest FAR of 0.5 was obtained in DCT and AHE for FERET Database. The highest FRR of 0.2821 was achieved in CLAHE for ORL Database, while 0.3000 was obtained in AHE for FERET Database. It was concluded that illumination control approaches have predominant effect on PCA–based facial recognition system. Keywords— Adaptive Histogram Equalization, Contrast Adaptive Histogram Equalization, Discrete Cosine Transform Illumination Normalization, Principal Component Analysis

Automatic spike sorting for high-density microelectrode arrays.

High-density microelectrode arrays can be used to record extracellular action potentials from hundreds to thousands of neurons simultaneously. Efficient spike sorters must be developed to cope with such large data volumes. Most existing spike sorting methods for single electrodes or small multielectrodes, however, suffer from the "curse of dimensionality" and cannot be directly applied to recordings with hundreds of electrodes. This holds particularly true for the standard reference spike sorting algorithm, principal component analysis-based feature extraction, followed by k-means or expectation maximization clustering, against which most spike sorters are evaluated. We present a spike sorting algorithm that circumvents the dimensionality problem by sorting local groups of electrodes independently with classical spike sorting approaches. It is scalable to any number of recording electrodes and well suited for parallel computing. The combination of data prewhitening before the principal component analysis-based extraction and a parameter-free clustering algorithm obviated the need for parameter adjustments. We evaluated its performance using surrogate data in which we systematically varied spike amplitudes and spike rates and that were generated by inserting template spikes into the voltage traces of real recordings. In a direct comparison, our algorithm could compete with existing state-of-the-art spike sorters in terms of sensitivity and precision, while parameter adjustment or manual cluster curation was not required. NEW & NOTEWORTHY We present an automatic spike sorting algorithm that combines three strategies to scale classical spike sorting techniques for high-density microelectrode arrays: 1) splitting the recording electrodes into small groups and sorting them independently; 2) clustering a subset of spikes and classifying the rest to limit computation time; and 3) prewhitening the spike waveforms to enable the use of parameter-free clustering. Finally, we combined these strategies into an automatic spike sorter that is competitive with state-of-the-art spike sorters.

Tensorial Independent Component Analysis-Based Feature Extraction for Polarimetric SAR Data Classification

For polarimetric synthetic aperture radar (PolSAR) data, various polarimetric signatures can be obtained by target decomposition techniques, which are of great help for characterizing the land cover. It is straightforward to combine these polarimetric features together and formulate them as a third-order polarimetric feature tensor. However, how to make full use of the abundant information provided by these polarimetric features remains a challenge. A feasible solution is applying feature extraction (FE) techniques on the high-dimensional polarimetric manifold to obtain a lower dimensional intrinsic feature set. Common FE methods, such as principal component analysis (PCA), independent component analysis (ICA), etc., use matrix linear algebra and require rearranging the original tensor into a matrix. This leads to the loss of the spatial information of the PolSAR data. In this paper, to jointly take advantage of the spatial and feature information, a novel FE scheme incorporating ICA with the tensor decomposition techniques is proposed. After applying the proposed FE method on the third-order polarimetric feature tensor, each PolSAR image pixel is represented by a low-dimensional intrinsic feature vector. Furthermore, these feature vectors are fed to the $k$ -nearest neighbor (KNN) classifier and support-vector-machine classifier for supervised classification. Simulated data, together with two measured data sets, i.e., Flevoland of Airborne Synthetic Aperture Radar (AIRSAR) and Quebec City of Uninhabited Aerial Vehicle Synthetic Aperture Radar (UAVSAR), are utilized to evaluate the performance of the proposed method. For comparison purpose, several classical and advanced FE methods, such as PCA, ICA, Laplacian eigenmaps, and $\hbox{LRTA}_{\rm dr}-(K_{1},K_{2},p)$ , are also applied. The experimental results demonstrate the superiority of the proposed FE method in three folds: 1) The extracted features by the proposed method are more discriminative, characterized by the high separability in the scatterplots; 2) the classification accuracy is improved as much as approximately 7% compared with the complex Wishart classifier; and 3) the proposed method is computational efficient and has fast convergence.

Hilbert-Huang transform-based video analysis for detecting colon polyps using composite similarity measure

The third leading cause of deaths from cancer is colorectal cancer (10% men and 9.2% in women). The only prevention is to detect and remove the colon polyps during optical colonoscopy. Hilbert-Huang transformation is an adaptive data driven technique. Nucleus of HHT is empirical mode decomposition. In this work, separate sets of training and testing samples are opted. This paper proposes bi-dimensional statistical empirical mode decomposition-based colon polyp detection strategy, wherein a composite similarity measure has been used. Few samples are randomly chosen for training database, remaining samples make the testing database. The proposed method is implemented on sequences of sample images from an optical colonoscopy video database provided by American College of Gastroenterology. Principal component analysis-based feature extraction is used in this work, as it reduces the dimensions efficiently from the main object. The obtained results demonstrate the achieved improvement in the recognition rates, in comparison with other procedures.

Principal Component Analysis Based Feature Extraction Approach to Identify Circulating microRNA Biomarkers

The discovery and characterization of blood-based disease biomarkers are clinically important because blood collection is easy and involves relatively little stress for the patient. However, blood generally reflects not only targeted diseases, but also the whole body status of patients. Thus, the selection of biomarkers may be difficult. In this study, we considered miRNAs as biomarker candidates for several reasons. First, since miRNAs were discovered relatively recently, they have not yet been tested extensively. Second, since the number of miRNAs is relatively limited, selection is expected to be easy. Third, since they are known to play critical roles in a wide range of biological processes, their expression may be disease specific. We applied a newly proposed method to select combinations of miRNAs that discriminate between healthy controls and each of 14 diseases that include 5 cancers. A new feature selection method is based on principal component analysis. Namely this method does not require knowledge of whether each sample was derived from a disease patient or a healthy control. Using this method, we found that hsa-miR-425, hsa-miR-15b, hsa-miR-185, hsa-miR-92a, hsa-miR-140-3p, hsa-miR-320a, hsa-miR-486-5p, hsa-miR-16, hsa-miR-191, hsa-miR-106b, hsa-miR-19b, and hsa-miR-30d were potential biomarkers; combinations of 10 of these miRNAs allowed us to discriminate each disease included in this study from healthy controls. These 12 miRNAs are significantly up- or downregulated in most cancers and other diseases, albeit in a cancer- or disease-specific combinatory manner. Therefore, these 12 miRNAs were also previously reported to be cancer- and disease-related miRNAs. Many disease-specific KEGG pathways were also significantly enriched by target genes of up−/downregulated miRNAs within several combinations of 10 miRNAs among these 12 miRNAs. We also selected miRNAs that could discriminate one disease from another or from healthy controls. These miRNAs were found to be largely overlapped with miRNAs that discriminate each disease from healthy controls.

Principal Component Analysis Based Feature Extraction, Morphological Edge Detection and Localization for Fast Iris Recognition

This study involves the Iris Localization based on morphological or set theory which is well in shape detection. Principal Component Analysis (P CA) is used for preprocessing, in which the removal of redundant and unwanted data is done. Applications such as Median Filtering and Adaptive thresholding are used for handling the variations i n lighting and noise. Features are extracted using Wavelet Packet Transform (WPT). Finally matching is performed using KNN. The proposed method is better than the previous method and is proved by th e results of different parameters. The testing of t he proposed algorithm was done using CASIA iris database (V1.0) and (V3.0).

Independent Component Analysis–Based Feature Extraction Technique for Defect Classification Applied for Pulsed Eddy Current NDE

With the development of nondestructive detection, the emerging testing techniques provide new challenges to signal analysis and interpretation approach applied to the inspection evaluation. Some researchers have developed the methods that focus on feature analysis of detected signals. This article presents a new feature analysis by the Independent Component Analysis (ICA) approach to evaluate the defects tested by the pulsed eddy current (PEC) technique. ICA is a high-order statistics technique used to separate multi-unknown sources, which has been successfully applied to facial image identification and separation of the components of 1D signal. In this article, the ICA approach is utilized to project the response signals of various defects into the independent components (ICs) feature subspace by signal representation model. Dependent on the selected ICs, each defect is represented by different projected coefficients, which are proposed to discriminate and classify the defects that belong to three categories. The improved ICA model is proposed to improve the classification of two similar categories of single defects: metal loss and subsurface defects. The evaluation using the series of experimental data has validated the classification of single defects and the defects with lift-off effect by our ICA approach. The comparison with Principal component analysis (PCA)–based approach further verified the better performance of the ICA-based model.

Fault detection in centrifugal pumping systems using neural networks

Fault detection and diagnosis of technical plants is of great importance for the safe operation and long life. An early detection of faults may help to avoid product deterioration, performance degradation, damage to the machinery and damage to human operators. This paper presents the design and development of Artificial Neural Network (ANN)-based model for the fault detection in centrifugal pumping system. The network is developed to detect a total of 20 faults. The training and testing data required to develop the neural network model were generated at different operating conditions by running the pumping system and by creating various faults in real time in a laboratory experimental model. A principal component analysis-based feature extraction method is proposed to reduce the dimension of the input features. The performance of the trained network is found to be satisfactory for the real-time fault diagnosis.

Multiple sensors on pulsed eddy-current detection for 3-D subsurface crack assessment

This paper proposes the use of multiple sensors in pulsed eddy-current detection for three-dimensional (3-D) subsurface flaw imaging. A normalization technique has been proposed to eliminate the characteristic variation among the Hall devices used in the probe and lift off effects. A principal component analysis-based feature extraction that provides orthogonal information for multiple sensor fusion has been introduced and investigated. Using the features of multiple projection coefficients, 3-D surface flaws can be measured and reconstructed. The experimental tests have illustrated that the proposed method has delivered more defect information than the conventional peak value and time for pulsed eddy-current sensors.