Local Principal Component Analysis Research Articles

Application of a novel local and automatic PCA algorithm for diffraction pattern denoising in TEM-ASTAR analysis in microelectronics

This paper introduces a novel denoising method for TEM-ASTAR™ Diffraction Pattern (DP) datasets, termed LAT–PCA (Local Automatic Thresholding – Principal Component Analysis). This approach enhances the established PCA algorithm by partitioning the 4D dataset (a 2D map of 2D DPs) into localized windows. Within these windows, PCA identifies a basis where the physical signal predominantly resides in the higher-order principal components. By thresholding lower-order components, the method effectively reduces noise while retaining the essential features of the DPs. The locality of the approach, focusing on small windows, enhances computational efficiency and aligns with the localized nature of the crystallographic grain signals in ASTAR. The automatic aspect of the method employs a theoretical pure noise distribution, i.e. a Marchenko-Pastur Distribution, to set a threshold, beyond which the components are mostly noise.The LAT–PCA method offers significant reductions in acquisition and post-processing times. With denoised data, selecting the correct parameters for accurate phase maps and grain orientations becomes more straightforward, facilitating robust quantitative grain analysis. Experiments performed on a silicon-germanium-carbon sample validate the method's efficacy. The sample was analyzed with varying acquisition times to produce a high signal-to-noise ratio reference dataset and a lower ratio test dataset. The LAT–PCA algorithm's denoising results on the test dataset were benchmarked against the reference, demonstrating considerable improvements and reliability.In summary, LAT–PCA is an effective, automatic solution for denoising TEM DP datasets. Its adaptability to different noise levels and local processing capability makes it a valuable tool for enhancing dataset quality and reducing the time required for data acquisition and analysis. This method can minimize acquisition time, conserve microscope usage, and reduce sample drift and deterioration, leading to more accurate characterizations with fewer deformation artifacts.

Quadratic Matrix Factorization With Applications to Manifold Learning.

Matrix factorization is a popular framework for modeling low-rank data matrices. Motivated by manifold learning problems, this paper proposes a quadratic matrix factorization (QMF) framework to learn the curved manifold on which the dataset lies. Unlike local linear methods such as the local principal component analysis, QMF can better exploit the curved structure of the underlying manifold. Algorithmically, we propose an alternating minimization algorithm to optimize QMF and establish its theoretical convergence properties. To avoid possible over-fitting, we then propose a regularized QMF algorithm and discuss how to tune its regularization parameter. Finally, we elaborate how to apply the regularized QMF to manifold learning problems. Experiments on a synthetic manifold learning dataset and three real-world datasets, including the MNIST handwritten dataset, a cryogenic electron microscopy dataset, and the Frey Face dataset, demonstrate the superiority of the proposed method over its competitors.

Gender Classification Using Viola Jones, Orthogonal Difference Local Binary Pattern and Principal Component Analysis

Facial recognition is currently a widely discussed topic, particularly in the context of gender classification. Facial recognition by computers is more complex and time-consuming compared to humans. There is ongoing research on facial feature extraction for gender classification. Geometry and texture features are effective for gender classification. This study aimed to combine these two features to improve the accuracy of gender classification. This research used the Viola-Jones and Orthogonal Difference Local Binary Pattern (OD-LBP) methods for feature extraction. The Viola-Jones algorithm faces issues in facial detection, leading to outliers in geometry features. At the same time, OD-LBP is a new descriptor capable of addressing pose, lighting, and expression variations. Therefore, this research attempts to utilize OD-LBP for gender classification. The dataset used was FERET, which contained various lighting variations, making OD-LBP suitable for addressing this challenge. Random Forest and Backpropagation were employed for classification. This research demonstrates that combining these two features is effective for gender classification using Backpropagation, achieving an accuracy of 93%. This confirms the superiority of the proposed method over single-feature extraction methods.

SLRfinder: A method to detect candidate sex-linked regions with linkage disequilibrium clustering.

Despite their critical roles in genetic sex determination, sex chromosomes remain unknown in many non-model organisms, especially those having recently evolved sex-linked regions (SLRs). These evolutionarily young and labile sex chromosomes are important for understanding early sex chromosome evolution but are difficult to identify due to the lack of Y/W degeneration and SLRs limited to small genomic regions. Here, we present SLRfinder, a method to identify candidate SLRs using linkage disequilibrium (LD) clustering, heterozygosity and genetic divergence. SLRfinder does not rely on specific sequencing methods or a specific type of reference genome (e.g., from the homomorphic sex). In addition, the input of SLRfinder does not require phenotypic sexes, which may be unknown from population sampling, but sex information can be incorporated and is necessary to validate candidate SLRs. We tested SLRfinder using various published datasets and compared it to the local principal component analysis (PCA) method and the depth-based method Sex Assignment Through Coverage (SATC). As expected, the local PCA method could not be used to identify unknown SLRs. SATC works better on conserved sex chromosomes, whereas SLRfinder outperforms SATC in analysing labile sex chromosomes, especially when SLRs harbour inversions. Power analyses showed that SLRfinder worked better when sampling more populations that share the same SLR. If analysing one population, a relatively larger sample size (around 50) is needed for sufficient statistical power to detect significant SLR candidates, although true SLRs are likely always top-ranked. SLRfinder provides a novel and complementary approach for identifying SLRs and uncovering additional sex chromosome diversity in nature.

A painting authentication method based on multi-scale spatial-spectral feature fusion and convolutional neural network

The scientific authentication of paintings holds significant importance within the realm of art collection. Employing convolutional neural networks for the classification of authentic and counterfeit painting images based on color images is a viable but suboptimal choice. This study investigates the potential for authenticating paintings by incorporating high-spectral images alongside high-resolution spatial images. High-resolution and high-spectral images of genuine and counterfeit paintings were acquired using a push-broom digital scanning system. The processing methods presented in this approach for the acquired images are: 1) The study utilized the circular local binary pattern (LBP) and principal component analysis (PCA) to extract surface texture and spectral data from Chinese character images in paintings, encompassing both spatial and spectral dimensions. 2) The technique utilizing non-subsampling Shearlet transform (NSST) and pulse-coupled neural network (PCNN) was employed to integrate the spatial and spectral characteristics of the images into a pseudo-color image, producing a dataset of feature data for genuine and counterfeit paintings. 3) The experiments aimed to achieve the authenticity of artworks using EfficientNet v2-s, the hyperparameters of which were fine-tuned accordingly. The experimental findings demonstrate that this approach attained a 90.8 % accuracy on the test dataset, representing a 3.5 % enhancement over the existing top-performing three-dimensional convolutional neural network (3D-CNN).

Data fusion of Raman spectra in MSPC for fault detection and diagnosis in pharmaceutical manufacturing

This study investigates the use of Raman spectroscopy fused with other types of data (e.g., pH, temperature and turbidity) for multivariate statistical process control of two pharmaceutical case studies: one simulated industrial-scale fed-batch process for the production of penicillin and one real lab-scale crystallization process. The monitoring schemes are built on local principal component analysis models and hyper-parameters are tuned with regards to highest accuracy in fault detection. Accuracies above 90% are obtained for all types of data and level of DF. Furthermore, for the first case study the model built solely on spectra achieves higher fault detection rates, when only considering faults that also result in off-specification quality. This is supported by the fact that the fault is not necessarily detected when it occurs, but rather when it starts to affect quality variables as measured by the spectra.

The prognostic value of circulating miRNAs in patients with localized/locally advanced prostate cancer after radical prostatectomy and radiotherapy.

184 Background: The outcome after curative intended management of local prostate cancer (PC) is promising. However, there is still a risk of biochemical relapse (BR) after either radical prostatectomy (RP) or radiotherapy (RT). MicroRNAs (miRNAs) have showed encouraging results as alternative biomarkers in different settings of PCa. The aim of this study was to investigate the prognostic value of circulating miRNAs after management of local PCa. Methods: In total 107 newly diagnosed patients with local or locally advanced PCa were included in the study. The cohort was divided into an interventional group with 59 patients (RP (39 patients) or RT (20 patients)) and an observational group with 48 patients (active surveillance (AS)). Both baseline samples at time of diagnosis and follow-up samples six months after treatment were collected. The relative expression of four miRNAs (miRNA-21, -93, -125b, and miRNA-221) was assessed in plasma using real-time polymerase chain reaction. Results: In the interventional cohort, the decrease in miRNA-125b after RT was significantly associated with a longer time to biochemical recurrence (BR) (HR= 0.17, p=0.03), and a higher level of miRNA-125b at baseline samples showed a tendency to higher risk for BR in the same population (HR= 2.82, p=0.09). On the other hand, lower levels of miRNA-221 at both baseline and follow-up samples tended to be associated with lower risk of BR in patients managed by RT (HR=0.14, p=0.07, and HR=0.13, p=0.07, respectively). None of these four miRNAs was associated with a significant risk of progression in patients managed by AS. Conclusions: Both baseline and the dynamics of miRNA-125b was informative as to the risk of BR in patients with PCa treated by RP. Validation in independent cohorts is called for.

Golden eagle based improved Att-BiLSTM model for big data classification with hybrid feature extraction and feature selection techniques

ABSTRACT The remarkable development in technology has led to the increase of massive big data. Machine learning processes provide a way for investigators to examine and particularly classify big data. Besides, several machine learning models rely on powerful feature extraction and feature selection techniques for their success. In this paper, a big data classification approach is developed using an optimized deep learning classifier integrated with hybrid feature extraction and feature selection approaches. The proposed technique uses local linear embedding-based kernel principal component analysis and perturbation theory, respectively, to extract more representative data and select the appropriate features from the big data environment. In addition, the feature selection task is fine-tuned by using perturbation theory through heuristic search based on their output accuracy. This feature selection heuristic search method is analysed with five recent heuristic optimization algorithms for deciding the final feature subset. Finally, the data are categorized through an attention-based bidirectional long short-term memory classifier that is optimized with a golden eagle-inspired algorithm. The performance of the proposed model is experimentally verified on publicly accessible datasets. From the experimental outcomes, it is demonstrated that the proposed framework is capable of classifying large datasets with more than 90% accuracy.

Automated adaptive chemistry for Large Eddy Simulations of turbulent reacting flows

Large Eddy Simulations (LES) of turbulent reacting flows carried out with detailed kinetic mechanisms have a key role for the discovery of the physical and chemical processes occurring in combustion systems, and are essential for the development of efficient, stable, and non-pollutant technologies. Nevertheless, these simulations require a large amount of computational resources, making their utilization for large-scale systems, such as industrial burners and gas turbines, impractical. In this work, we combine state-of-the-art machine learning algorithms and model reduction methods to deliver a fully automated strategy for performing LES with adaptive chemistry. This strategy is based on the Sample-Partitioning Adaptive Chemistry (SPARC) algorithmic procedure, which consists of four steps: the generation of a training dataset, its partitioning in clusters, the generation of a set of reduced chemical mechanisms specifically tailored to each cluster and, lastly, the numerical simulation of the case of interest with adaptive chemistry enabled by an on-the-fly classification of every grid point.The SPARC approach has already been demonstrated to substantially reduce the computational effort of reactive flows simulations. However a non-negligible level of user interventions is needed, upon which the method’s success critically depend. Therefore, with the goal of boosting the performance of this workflow and minimise the user-specified degrees of freedom, we plug in and exploit the Local Principal Component Analysis augmented with an automated Bayesian-optimised search for optimal clustering solutions, and the Computational Singular Perturbation method with an additional layer of automation based on the Tangential Stretching Rate for minimally-sized reduced mechanisms. We employ a cheap and easy-to-generate 1-dimensional-flames training database and we demonstrate the efficiency, accuracy and robustness of this strategy with an application to LES of the Adelaide Jet in Hot Coflow (AJHC) burner, a turbulent reacting flow exhibiting intense turbulence-chemistry interactions.

Leukemia Blood Cells Detection using Neural Network Classifier

Image segmentation is an image processing operation performed on the image in order to partition the image into some images based on the information contained in the original image. Image segmentation plays an important role in many medical imaging applications, image segmentation facilitates the anatomy process in a particular body of human body. Classification and clustering are the methods used un data mining for analyzing the data sets and divide them on the basis of some particular classification rules. There are many image segmentation tools that used for medical purpose, so it is necessary to define and/or to improve the image segmentation methods in order to get the best method. In this study, the image of leukemia and red blood cells will be used as samples to determine the best algorithm in image segmentation. The procedure for doing segmentation itself is clustering image, edge detection on image, and image classification. The clustering is to extract important information from an image. The edge detection is to determine the existence of edges of lines in image in order to investigate and localize the desired edge features. Moreover, the classification analyzes the properties of some images and organizes the information into certain categories. In this study, the Neural Network and K-Nearest Neighbor are used for image classification by paired with Local Binary Pattern and Principal Component Analysis. The results revealed that the best method of proven in classifying images is from Local Binary Pattern feature extraction with the average accuracy of 94%.

Adaptive local Principal Component Analysis improves the clustering of high-dimensional data

In local Principal Component Analysis (PCA), a distribution is approximated by multiple units, each representing a local region by a hyper-ellipsoid obtained through PCA. We present an extension for local PCA which adaptively adjusts both the learning rate of each unit and the potential function which guides the competition between the local units. Our local PCA method is an online neural network method where unit centers and shapes are modified after the presentation of each data point. For several benchmark distributions, we demonstrate that our method improves the overall quality of clustering, especially for high-dimensional distributions where many conventional methods do not perform satisfactorily. Our online method is also well suited for the processing of streaming data: The two adaptive mechanisms lead to a quick reorganization of the clustering when the underlying distribution changes.

A method for detecting tomato canopies’ phenotypic traits based on improved skeleton extraction algorithm

Automatic acquisition of tomato canopies’ phenotypic traits is essential for tomato varieties’ selection and scientific cultivation. Due to the infinite growth characteristics of tomato, its organ development is stochastic and the canopies' internal structure of is also complex. These make it challenging to obtain organs’ detailed phenotypic traits. Thus, this work proposed a method for detecting tomato canopies’ phenotypic traits based on improved skeleton extraction algorithm (ISEA). Firstly, after collecting tomato canopies' point cloud data from multiple perspectives, this work reconstructed its three-dimensional (3D) model accurately. Secondly, the least squares method was used to simplify the spatial contraction model of the Laplace Skeleton Extraction algorithm to obtain the tomatoes’ skeleton point set. On this basic, Greedy algorithm was used to optimise the Edge Collapse algorithm to extract more accurate and reliable skeleton structure. Then, in conjunction with the canopy growth characteristics of the crop and the Intrinsic Shape Signatures (ISS) principle, the simplified skeleton structure was subjected to local principal component analysis (PCA), which achieved the separation of tomatoes’ main stem and leaves. Finally, a modelling algorithm based on Delaunay triangulation was applied to construct the separated organs’ model to calculate phenotypic traits such as stem diameter, leaf area index and average leaf inclination. The calculated results were also compared with the measured values at different growth stages for performance evaluation. The average precision, average recall, average accuracy and micro F1 score of ISEA were 0.9144, 0.7751, 0.7243 and 0.8306, respectively. The overall R2 between calculated values and measured values for stem diameter, leaf area index and average leaf inclination were 0.9638, 0.9067, and 0.9428, and the root mean square errors (RMSE) were 0.3922, 0.0029, and 0.0186, respectively. As a result, the proposed method can extract a simple skeleton from the complex tomato canopy and calculate phenotypic traits accurately. It also provided solid technical support for studying the interaction about “genotype-phenotype-envirotype” in tomato.

Lie PCA: Density estimation for symmetric manifolds

We introduce an extension to local principal component analysis for learning symmetric manifolds. In particular, we use a spectral method to approximate the Lie algebra corresponding to the symmetry group of the underlying manifold. We derive the sample complexity of our method for a variety of manifolds before applying it to various data sets for improved density estimation.

Privacy preserving identification of population stratification for collaborative genomic research.

The rapid improvements in genomic sequencing technology have led to the proliferation of locally collected genomic datasets. Given the sensitivity of genomic data, it is crucial to conduct collaborative studies while preserving the privacy of the individuals. However, before starting any collaborative research effort, the quality of the data needs to be assessed. One of the essential steps of the quality control process is population stratification: identifying the presence of genetic difference in individuals due to subpopulations. One of the common methods used to group genomes of individuals based on ancestry is principal component analysis (PCA). In this article, we propose a privacy-preserving framework which utilizes PCA to assign individuals to populations across multiple collaborators as part of the population stratification step. In our proposed client-server-based scheme, we initially let the server train a global PCA model on a publicly available genomic dataset which contains individuals from multiple populations. The global PCA model is later used to reduce the dimensionality of the local data by each collaborator (client). After adding noise to achieve local differential privacy (LDP), the collaborators send metadata (in the form of their local PCA outputs) about their research datasets to the server, which then aligns the local PCA results to identify the genetic differences among collaborators' datasets. Our results on real genomic data show that the proposed framework can perform population stratification analysis with high accuracy while preserving the privacy of the research participants.

Unsupervised long-term damage detection in an uncontrolled environment through optimal autoencoder

Unsupervised damage detection in the presence of both regular environmental variations, such as a daily change in temperature and humidity, and irregular environmental variations, such as rain and snow, remains a critical challenge in guided wave structural health monitoring. This paper proposes an optimal autoencoder-based damage detection strategy to solve this problem. Although the autoencoder and other neural networks have been used to detect anomalies in structural health monitoring, the training data has been also required to be collected from a known intact structure, which increases the time and labor cost for inspection and limits the application of such methods. Instead, our autoencoder is trained with guided wave data that contain uncontrolled regular, irregular, and damage variations. We investigate hyperparameters that will negatively influence autoencoder-based damage detection, including the training time and damage duration. We also propose sparsity, dropout, and weight decay regularization strategies to enhance the robustness of the autoencoder to these hyperparameters. Results show that our optimal autoencoder method can achieve an area under the receiver operating curve score near 0.92 for detecting damage present in the last 16 days, which improves previous local principal components analysis reconstruction methods with a score of 0.88.

Abstract A070: Determining gene expression regulated by androgen receptor variant 7 in co-loss of MAP3K7/CHD1 prostate cancer

Abstract Prostate cancer (PCa) is the second leading cancer diagnosis in men. The ACS estimates 268,490 new cases of PCa and 34,500 deaths in 2022. Most patients with local PCa are disease free at the 5-year mark. Patients with metastatic PCa only have a 31% 5-year disease-free survival rate. While these patients may initially respond to androgen deprivation therapy, many will progress to castration-resistant prostate cancer (CRPC) and fail treatment. There is a need to develop new therapies to treat CRPC. The Cramer lab studies the dual deletion of MAP3K7 and CHD1 PCa subtype. In MAP3K7/CHD1-depleted PCa cell lines the ratio of AR splicing variants shifts to enrich for the constitutively active variant AR-v7, a known driver of CRPC. Other studies demonstrated that AR-v7 has unique target genes and can alter canonical AR target gene expression. I hypothesize that gain of AR-v7 signaling activates atypical gene targets which alter the gene expression of MAP3K7/CHD1-depleted PCa. I will determine the chromatin-binding targets and identify AR-v7 specific candidate genes. Patients with loss of MAP3K7 and CHD1 also enrich in the spliceosome protein U2AF2, therefore I will also perform transcriptomics with/without AR-FL and AR-v7 to characterize the larger impacts of AR-variant signaling at the RNA level. To this end, I will develop an inducible knockdown system in dual shMAP3K7/shCHD1 cells targeting AR-FL, AR-v7, and AR-Total. This will provide insight into the role of AR-v7 in two stages of gene expression and identify candidate therapeutic genes/pathways in loss of MAP3K7 and CHD1 PCa. Citation Format: Claire Gillette, Lauren Jillson, Scott Cramer. Determining gene expression regulated by androgen receptor variant 7 in co-loss of MAP3K7/CHD1 prostate cancer [abstract]. In: Proceedings of the AACR Special Conference: Advances in Prostate Cancer Research; 2023 Mar 15-18; Denver, Colorado. Philadelphia (PA): AACR; Cancer Res 2023;83(11 Suppl):Abstract nr A070.

Estimation and Inference on Time-Varying FAVAR Models

ABSTRACT We introduce a time-varying (TV) factor-augmented vector autoregressive (FAVAR) model to capture the TV behavior in the factor loadings and the VAR coefficients. To consistently estimate the TV parameters, we first obtain the unobserved common factors via the local principal component analysis (PCA) and then estimate the TV-FAVAR model via a local smoothing approach. The limiting distribution of the proposed estimators is established. To gauge possible sources of TV features in the FAVAR model, we propose three L 2 -distance-based test statistics and study their asymptotic properties under the null and local alternatives. Simulation studies demonstrate the excellent finite sample performance of the proposed estimators and tests. In an empirical application to the U.S. macroeconomic dataset, we document overwhelming evidence of structural changes in the FAVAR model and show that the TV-FAVAR model outperforms the conventional time-invariant FAVAR model in predicting certain key macroeconomic series.

Speech Segmentation Based on the Computation of Local Signal Manifold Dimension

A new computational method of unvoiced and voiced speech segmentation is proposed from the perspective of local linear manifold analysis of speech signals. It is based on the estimation of the dimension of short-time linear subspace. The subspace dimensional characteristics of the single phoneme signal are studied. The local signal vector set is analyzed by using the PCA algorithm to estimate the dimension of the data matrix formed by framing. The local PCA is used to analyze the speech signal to achieve the segmentation of unvoiced and voiced pronunciation. Simulation experiments prove the effectiveness of the proposed method.

Local manifold learning and its link to domain-based physics knowledge

In many reacting flow systems, the thermo-chemical state-space is known or assumed to evolve close to a low-dimensional manifold (LDM). Various approaches are available to obtain those manifolds and subsequently express the original high-dimensional space with fewer parameterizing variables. Principal component analysis (PCA) is one of the dimensionality reduction methods that can be used to obtain LDMs. PCA does not make prior assumptions about the parameterizing variables and retrieves them empirically from training data. In this paper, we show that PCA applied in local clusters of data (local PCA) is capable of detecting physically meaningful parameterization of the thermo-chemical state-space. We first demonstrate that utilizing three common combustion models of varying complexity: the Burke–Schumann model, the chemical equilibrium model, and the homogeneous reactor. Parameterization of these models is known a priori which allows for benchmarking with the local PCA approach. We further extend the application of local PCA to a more challenging case of a turbulent non-premixed n-heptane/air jet flame for which the parameterization is no longer obvious. Our results suggest that meaningful parameterization can be obtained also for more complex datasets. We show that local PCA finds variables that can be linked to local stoichiometry, reaction progress and soot formation processes. We shed the light on how data-driven techniques, such as local PCA, can be enhanced by using the available knowledge of the system.

Acetazolamide-augmented BOLD MRI to Assess Whole-Brain Cerebrovascular Reactivity in Chronic Steno-occlusive Disease Using Principal Component Analysis.

Background Exhaustion of cerebrovascular reactivity (CVR) portends increased stroke risk. Acetazolamide-augmented blood oxygenation level-dependent (BOLD) MRI has been used to estimate CVR, but low signal-to-noise conditions relegate its use to terminal CVR (CVRend) measurements that neglect dynamic features of CVR. Purpose To demonstrate comprehensive characterization of acetazolamide-augmented BOLD MRI response in chronic steno-occlusive disease using a computational framework to precondition signal time courses for dynamic whole-brain CVR analysis. Materials and Methods This study focused on retrospective analysis of consecutive patients with unilateral chronic steno-occlusive disease who underwent acetazolamide-augmented BOLD imaging for recurrent minor stroke or transient ischemic attack at an academic medical center between May 2017 and October 2020. A custom principal component analysis-based denoising pipeline was used to correct spatially varying non-signal-bearing contributions obtained by a local principal component analysis of the MRI time series. Standard voxelwise CVRend maps representing terminal responses were produced and compared with maximal CVR (CVRmax) as isolated from binned (per-repetition time) denoised BOLD time course. A linear mixed-effects model was used to compare CVRmax and CVRend in healthy and diseased hemispheres. Results A total of 23 patients (median age, 51 years; IQR, 42-61, 13 men) who underwent 32 BOLD examinations were included. Processed MRI data showed twofold improvement in signal-to-noise ratio, allowing improved isolation of dynamic characteristics in signal time course for sliding window CVRmax analysis to the level of each BOLD repetition time (approximately 2 seconds). Mean CVRmax was significantly higher than mean CVRend in diseased (5.2% vs 3.8%, P < .01) and healthy (5.5% vs 4.0%, P < .01) hemispheres. Several distinct time-signal signatures were observed, including nonresponsive; delayed/blunted; brisk; and occasionally nonmonotonic time courses with paradoxical features in normal and abnormal tissues (ie, steal and reverse-steal patterns). Conclusion A principal component analysis-based computational framework for analysis of acetazolamide-augmented BOLD imaging can be used to measure unsustained CVRmax through twofold improvements in signal-to-noise ratio. © RSNA, 2023 Supplemental material is available for this article.