Correlation Of Features Research Articles

Accelerating biopharmaceutical cell line selection with label-free multimodal nonlinear optical microscopy and machine learning

The selection of high-performing cell lines is crucial for biopharmaceutical production but is often time-consuming and labor-intensive. We investigated label-free multimodal nonlinear optical microscopy for non-perturbative profiling of biopharmaceutical cell lines based on their intrinsic molecular contrast. Employing simultaneous label-free autofluorescence multiharmonic (SLAM) microscopy with fluorescence lifetime imaging microscopy (FLIM), we characterized Chinese hamster ovary (CHO) cell lines at early passages (0–2). A machine learning (ML)-assisted analysis pipeline leveraged high-dimensional information to classify single cells into their respective lines. Remarkably, the monoclonal cell line classifiers achieved balanced accuracies exceeding 96.8% as early as passage 2. Correlation features and FLIM modality played pivotal roles in early classification. This integrated optical bioimaging and machine learning approach presents a promising solution to expedite cell line selection process while ensuring identification of high-performing biopharmaceutical cell lines. The techniques have potential for broader single-cell characterization applications in stem cell research, immunology, cancer biology and beyond.

Polarization-dependent control of ionization pathways in atomic strong-field nonsequential double ionization

Utilizing a classical ensemble model (CEM), we have studied the rescattering behavior of quasi-free tunneling electron and correlation dynamics of electron pair during nonsequential double ionization (NSDI) of Xe atoms induced by a counter-rotating, two-color, elliptically polarized (TCEP) laser field. Our numerical results reveal that the momentum of the first tunneling electron is predominantly located in the third quadrant of momentum spectrum, whereas the momentum spectrum of the second electron presents a triangular structure along the direction of the negative vector potential of driving field. This phenomenon occurs because the first electron has greater emission kinetic energy than the second one, rendering the latter more sensitive to the influence exerted by the electric field vector potential. Moreover, we also observe that increasing ellipticity significantly alters ion momentum distributions by reducing weights at non-zero peaks, which indicates a weakened electron correlation. The large ellipticity field progressively decreases recollision energy, reflected in broadening delays between recollision and double ionization time, ultimately modifying ionization pathways and correlation features of electron pair.

A Stacking Ensemble Model with Enhanced Feature Selection for Distributed Denial-of-Service Detection in Software-Defined Networks

The proliferation of Distributed Denial of Service (DDoS) attacks poses a significant threat to network accessibility and performance. Traditional feature selection methods struggle with the complexity of network traffic data, leading to poor detection performance. To address this issue, a Genetic Algorithm Wrapper Feature Selection (GAWFS) is proposed, integrating Chi-squared and Genetic Algorithm (GA) approaches with a correlation method to select the most correlated features. GAWFS effectively reduces feature dimensions, eliminates redundancy, and identifies crucial and correlated features for classification. Detection accuracy is further improved by employing a stacking ensemble model, combining Multi-Layer Perceptron (MLP) and Support Vector Machine (SVM) as base models, with Random Forest (RF) as the metamodel. The proposed classifier achieves impressive accuracies of 99.86% for training data and 98.89% for test data, representing improvements of approximately 5% and 40%, respectively, over previous studies. The training time was also reduced to 2,593 s, a substantial improvement of approximately 29.92%. Validation on various benchmark datasets confirmed the efficacy of the proposed approach, underscoring the importance of the enhanced feature selection method and the stacking ensemble model against DDoS attacks.

Retinal optical coherence tomography intensity spatial correlation features as new biomarkers for confirmed Alzheimer’s disease

BackgroundThe nature and severity of Alzheimer’s disease (AD) pathologies in the retina and brain correspond. However, retinal biomarkers need to be validated in clinical cohorts with confirmed AD biomarkers and optical coherence tomography (OCT). The main objective of this study was to investigate whether retinal metrics measured by OCT aid in the early screening and brain pathology monitoring for confirmed AD.MethodsThis was a case–control study. All participants underwent retinal OCT imaging, and neurological examinations, including amyloid-β (Aβ) positron emission tomography. Participants were subdivided into cognitively normal (CN), mild cognitive impairment (MCI), and AD-derived dementia (ADD). Except retinal thickness, we developed the grey level co-occurrence matrix algorithm to extract retinal OCT intensity spatial correlation features (OCT-ISCF), including angular second matrix (ASM), correlation (COR), and homogeneity (HOM), one-way analysis of variance was used to compare the differences in retinal parameters among the groups, and to analyze the correlation with brain Aβ plaques and cognitive scores. The repeatability and robustness of OCT-ISCF were evaluated using experimental and simulation methods.ResultsThis study enrolled 82 participants, subdivided into 20 CN, 22 MCI, and 40 ADD. Compared with the CN, the thickness of retinal nerve fiber layer and myoid and ellipsoid zone were significantly thinner (P < 0.05), and ASM, COR, and HOM in several retinal sublayers changed significantly in the ADD (P < 0.05). Notably, the MCI showed significant differences in ASM and COR in the outer segment of photoreceptor compared with the CN (P < 0.05). The changing pattern of OCT-ISCF with interclass correlation coefficients above 0.8 differed from that caused by speckle noise, and was affected by OCT image quality index. Moreover, the retinal OCT-ISCF were more strongly correlated with brain Aβ plaque burden and MoCA scores than retinal thickness. The accuracy using retinal OCT-ISCF (AUC = 0.935, 0.830) was better than that using retinal thickness (AUC = 0.795, 0.705) in detecting ADD and MCI.ConclusionsThe study demonstrates that retinal OCT-ISCF enhance the association and detection efficacy of AD pathology compared to retinal thickness, suggesting retinal OCT-ISCF have the potential to be new biomarkers for AD.

A novel correlation feature self-assigned Kolmogorov-Arnold Networks for multi-energy load forecasting in integrated energy systems

A novel correlation feature self-assigned Kolmogorov-Arnold Networks for multi-energy load forecasting in integrated energy systems

Evaluation of Surface Weathering Degree of Brick Historical Buildings Based on GLCM-Entropy Weight TOPSIS: A Case Study in the Kaifeng City Wall

ABSTRACT The identification of the weathering degree of the surface of brick historical buildings is the key to determine the rationality of the repair plan. Image processing technology is a common method for quantitative weathering assessment, and mortar joint is a difficult point that affects the weathering assessment of brick ancient buildings. Based on the image chunking method, the GLCM-entropy weight TOPSIS method is proposed to evaluate the surface weathering degree of brick ancient buildings. The method includes three parts: The image chunking technology is proposed to effectively avoid the influence of mortar joint on the evaluation results. Second, GLCM is used to extract the four texture features of contrast, homogeneity, entropy and correlation of bricks, which can effectively distinguish image differences. Finally, the features are used as the EW-TOPSIS evaluation indices for comprehensive evaluation of 500 sample images obtained from Kaifeng city wall. The evaluation result S of Kaifeng city wall is 0.452, which indicates that the degree of weathering is moderate weathering. The results show that this method solves the problem of subjectivity in weathering recognition, and the introduction of image chunking processing can effectively avoid the error caused by image quality.

Fault diagnostic method based on transfer dynamic deep learning for few shot temporal–spatial correlation industry process

AbstractData‐based fault diagnosis plays a crucial role in ensuring the safety of industrial processes. However, the complex industry process often has temporal–spatial correlation with insufficient labelled fault data. To settle these problems, a new transfer dynamic deep learning strategy that combines autoencoder (AE) with gate recurrent unit (GRU) is proposed. First, dynamic AE networks are introduced to extract the single‐attribute time series features, and the dynamic GRU is employed to extract the spatial correlation features among multiple feature dimensions and temporal correlation among samples. Then, to solve the problem of insufficiently labelled industrial data, the model‐based transfer learning between the sufficient laboratory data and insufficient labelled industrial data is executed. Experimental results based on the Tennessee Eastman (TE) process and the benchmark simulation model 1 (BSM1) process show that the proposed approach has excellent performance.

Efficient diagnosis of diabetes mellitus using an improved ensemble method

Diabetes is a growing health concern in developing countries, causing considerable mortality rates. While machine learning (ML) approaches have been widely used to improve early detection and treatment, several studies have shown low classification accuracies due to overfitting, underfitting, and data noise. This research employs parallel and sequential ensemble ML approaches paired with feature selection techniques to boost classification accuracy. The Pima India Diabetes Data from the UCI ML Repository served as the dataset. Data preprocessing included cleaning the dataset by replacing missing values with column means and selecting highly correlated features using forward and backward selection methods. The dataset was split into two parts: training (70%), and testing (30%). Python was used for classification in Jupyter Notebook, and there were two design phases. The first phase utilized J48, Classification and Regression Tree (CART), and Decision Stump (DS) to create a random forest model. The second phase employed the same algorithms alongside sequential ensemble methods—XG Boost, AdaBoostM1, and Gradient Boosting—using an average voting algorithm for binary classification. Evaluation revealed that XG Boost, AdaBoostM1, and Gradient Boosting achieved classification accuracies of 100%, with performance metrics including F1 score, MCC, Precision, Recall, AUC-ROC, and AUC-PR all equal to 1.00, indicating reliable predictions of diabetes presence. Researchers and practitioners can leverage the predictive model developed in this work to make quick predictions of diabetes mellitus, which could save many lives.

Tool wear prediction based on XGBoost feature selection combined with PSO-BP network

To address the challenge of accurately capturing tool wear states in small sample scenarios, this paper proposes a tool wear prediction method that combines XGBoost feature selection with a PSO-BP network. In order to solve the problem of input feature selection and parameter selection in BP neural network, a double-layer programming model of input feature and parameter selection is established, which is solved by XGBoost and PSO. Initially, vibration and cutting force signals from CNC machining are preprocessed using time-domain segmentation, Hampel filtering, and wavelet denoising. Subsequently, time-domain, frequency-domain, and time–frequency domain features are extracted from the preprocessed data using FFT and wavelet packet decomposition, followed by feature screening for tool wear mapping via Pearson correlation and XGBoost feature importance analysis as model input. Finally, PSO is employed to optimize BPNN parameters. Experimental results show that PSO outperforms other algorithms in training the tool wear prediction model, with XGBoost feature selection reducing model construction time by 57.4% and increasing accuracy by 63.57%, demonstrating superior feature selection capabilities over Decision Tree, Random Fores, Adaboost and Extra Trees. These findings suggest that the proposed method can effectively predict tool wear in real-world CNC machining, contributing to improved production efficiency, reduced tool replacement frequency, and lower maintenance costs, thereby providing valuable insights for industrial applications.

Predicting Epileptic Seizures Using EfficientNet-B0 and SVMs: A Deep Learning Methodology for EEG Analysis

Seizure prediction is a critical challenge in epilepsy management, offering the potential to improve patient outcomes through timely interventions. This study proposes a novel framework combining a convolutional neural network (CNN) based on EfficientNet-B0 and an ensemble of six Support Vector Machines (SVMs) with a voting mechanism for robust seizure prediction. The framework leverages normalized Short-Time Fourier Transform (STFT) and channel correlation features extracted from EEG signals to capture both spectral and spatial information. The methodology was validated on the CHB-MIT dataset across preictal windows of 10, 20, and 30 min, achieving accuracies of 96.12%, 94.89%, and 94.21%, and sensitivities of 95.21%, 93.98%, and 93.55%, respectively. Comparing the results with state-of-the-art methods, we highlight the framework’s robustness and adaptability. The EfficientNet-B0 backbone ensures high accuracy with computational efficiency, while the SVM ensemble enhances prediction reliability by mitigating noise and variability in EEG data.

A deep learning approach for early prediction of breast cancer neoadjuvant chemotherapy response on multistage bimodal ultrasound images

Neoadjuvant chemotherapy (NAC) is a systemic and systematic chemotherapy regimen for breast cancer patients before surgery. However, NAC is not effective for everyone, and the process is excruciating. Therefore, accurate early prediction of the efficacy of NAC is essential for the clinical diagnosis and treatment of patients. In this study, a novel convolutional neural network model with bimodal layer-wise feature fusion module (BLFFM) and temporal hybrid attention module (THAM) is proposed, which uses multistage bimodal ultrasound images as input for early prediction of the efficacy of neoadjuvant chemotherapy in locally advanced breast cancer (LABC) patients. The BLFFM can effectively mine the highly complex correlation and complementary feature information between gray-scale ultrasound (GUS) and color Doppler blood flow imaging (CDFI). The THAM is able to focus on key features of lesion progression before and after one cycle of NAC. The GUS and CDFI videos of 101 patients collected from cooperative medical institutions were preprocessed to obtain 3000 sets of multistage bimodal ultrasound image combinations for experiments. The experimental results show that the proposed model is effective and outperforms the compared models. The code will be published on the https://github.com/jinzhuwei/BLTA-CNN.

An optimized method for short-term load forecasting based on feature fusion and ConvLSTM-3D neural network

As renewable energy continues to penetrate modern power systems, accurate short-term load forecasting is crucial for optimizing power generation resource allocation and reducing operational costs. Traditional forecasting methods often overlook key factors such as holiday load variations and differences in user electricity consumption behavior, resulting in reduced accuracy. To address this, we propose an optimized short-term load forecasting method based on time and weather-fused features using a ConvLSTM-3D neural network. The Prophet algorithm is first employed to decompose historical electricity load data, extracting feature components related to time variables. Simultaneously, the SHAP algorithm filters weather variables to identify highly correlated weather features. A time attention mechanism is then applied to fuse these features based on their correlation weights, enhancing their impact within the time series. Finally, the ConvLSTM-3D model is trained on the fused features to generate short-term load forecasts. A case study using real-world data validates the proposed method, demonstrating significant improvements in forecasting accuracy.

P0517 Role of radiomics in predicting the degree of intestinal fibrosis in patients with Crohn’s Disease

Abstract Background During their lifetime, patients with Crohn’s disease (CD) frequently develop strictures as a complication, where various degrees of inflammation and fibrosis coexist. Distinguishing between inflammatory and fibrotic strictures is crucial for deciding on surgery or treatment changes, yet it is currently challenging in clinical practice. Cross-sectional imaging techniques (e.g. magnetic resonance imaging MRI) provide a reasonably accurate estimate of bowel wall fibrosis. However, only histopathological examination can determine its true extent. Radiomics uses artificial intelligence to analyze quantitative features from diagnostic images and can predict clinical outcomes. This study aims to assess the role of radiomics in classifying patients according to the degree (no/mild vs severe) of intestinal fibrosis in a cohort of CD patients who underwent surgery for strictures disease Methods We retrospectively selected 48 CD patients who underwent MR-enterography and subsequent surgery within 4 months between 2017-2023. An expert pathologist evaluated the degree of fibrosis in histopathological specimens from surgical samples using the Chiorean scoring system from 0 (no fibrosis) to 2 (severe fibrosis), and the degree of inflammation using Geboes score. For each MR-enterography, typical lesions of CD were searched by an expert radiologist on T2-weighted axial images generating a region of interest (ROI) segmentation for each lesion. We extracted 100 radiomic features from each ROI of the pre-processed MRI images using Pyradiomics. Feature selection included univariate analysis (Wilcoxon-Mann-Whitney test) and Spearmann correlation to exclude highly correlated features. A logistic regression model with stepwise regression was built with the selected features and evaluated by computing the area under the curve (AUC) of the receiver operating characteristic (ROC) curve. Model’s sensitivity and specificity were calculated using the best cut-off according to the Youden index method. Results Among the 48 patients, 48% (n=23) showed severe fibrosis on the surgical specimen, whereas 52% (n=25) had either no or mild fibrosis. Two radiomic features were selected and resulted statistically significant in the fitted logistic regression model(p &amp;lt; 0.05): original_shape_Maximum2DDiameterColumn and original_firstorder_90Percentile. Results with their 95%CI were: AUC 0.79(0.64-0-90), sensitivity 0.69(0.50-0.75) and specificity 0.84(0.66-0.94). Bootstrap corrected (mean optimism) values of AUC and metrics were: AUC 0.77 (0.02), sensitivity 0.68(0.01), specificity 0.81(0.02). Conclusion Our study suggests that MRI-based radiomics modeling could help predict the degree of intestinal fibrosis. Future research will confirm these findings in a larger cohort

Industrial multivariate time-series data anomaly detection incorporating attention mechanisms and adversarial training

ABSTRACT To address the challenges faced in industrial anomaly detection, including data sample imbalance, lack of anomaly labels, and complex spatiotemporal relationships in high-dimensional data, this paper proposes a novel multi-modal time-series anomaly detection model that combines attention mechanisms and adversarial training. In this model, the first step involves utilizing graph attention mechanisms to extract sequence correlation features from multi-modal time-series data, which are then summed with the original data to form a dual-feature-based data representation. Subsequently, a self-supervised learning approach is employed to input this data representation into a variational autoencoder’s encoding-decoding network for reconstruction. Anomaly detection is performed by analyzing the error between the input and reconstructed data. The model also employs spatiotemporal attention mechanisms and adversarial training during reconstruction to enhance feature extraction and model generalization. By comparing our proposed model to five commonly used baseline models, we demonstrate its effectiveness in detecting anomalies in scenarios involving high-dimensional data and imbalanced abnormal samples, demonstrating superior anomaly detection performance, as well as excellent performance on real industrial production and processing datasets.

Mixing Data Cube Architecture and Geo-Object-Oriented Time Series Segmentation for Mapping Heterogeneous Landscapes

The conflict between environmental conservation and agricultural production highlights the need for precise land use and land cover (LULC) mapping to support agro-environmental-related policies. Satellite image time series from the Moderate Resolution Image Spectroradiometer (MODIS) sensor are essential for current LULC mapping efforts. However, most approaches focus on pixel data, and studies exploring object-based spatiotemporal heterogeneity and correlation features in its time series are limited. The objective of this study is to mix the data cube architecture (analysis-ready data—ARD) and the geo-object-oriented time series segmentation via Geographic Object-Based Image Analysis (GEOBIA) to assess its performance in identifying natural vegetation and double-cropping practices over a crop season. The study area was the state of Mato Grosso, Brazil. Results indicate that, by combining GEOBIA and time series analysis (materialized by the multiresolution segmentation algorithm to derive spatiotemporal geo-objects of the MODIS data cube), representative training data collected after a quality control process, and the Support Vector Machine to classify the ARD, the overall accuracy was 0.95 and all users’ and producers’ accuracies were higher than 0.88. By considering the heterogeneity of Mato Grosso’s landscape, the results indicate the potential of the approach to provide accurate mapping.

Detecting and Mapping Peatlands in the Tibetan Plateau Region Using the Random Forest Algorithm and Sentinel Imagery

The extensive peatlands of the Tibetan Plateau (TP) play a vital role in sustaining the global ecological balance. However, the distribution of peatlands across this region and the related environmental factors remain poorly understood. To address this issue, we created a high-resolution (10 m) map for peatland distribution in the TP region using 6146 Sentinel-1 and 23,730 Sentinel-2 images obtained through the Google Earth Engine platform in 2023. We employed a random forest algorithm that integrated spatiotemporal features with field training samples. The overall accuracy of the peatland distribution map produced is high, at 86.33%. According to the classification results, the total area of peatlands on the TP is 57,671.55 km2, and they are predominantly located in the northeast and southwest, particularly in the Zoige Protected Area. The classification primarily relied on the NDVI, NDWI, and RVI, while the DVI and MNDWI were also used in peatland mapping. B11, B12, NDWI, RVI, NDVI, and slope are the most significant features for peatland mapping, while roughness, correlation, entropy, and ASM have relatively slight significance. The methodology and peatland map developed in this work will enhance the conservation and management of peatlands on the TP while informing policy decisions and supporting sustainable development assessments.

Multi-dimensional failure risk load prediction of electric motor based on physical model and data-driven approach

Dynamic prediction of failure risk loads is a critical prerequisite for system damage monitoring and service life evaluation. As the failure mechanism of electric vehicle drive motors under multi-source load excitations is complex, and the risk load poses challenges in rapid acquisition, a physics-based and data-driven method for multi-dimensional failure risk load prediction of motors is proposed. Based on actual operational data collected from users, we analyze the failure risk loads of various components of a motor. A physical simulation model of the motor is established to extract multi-dimensional failure risk load time history, and the effectiveness of the simulation model is verified using data from bench tests. Moreover, using the physical model simulation data as a foundation, a two-stage data-driven model is constructed. In the first stage, the genetic algorithm optimized back propagation neural network (GA-BPNN) model is implemented for multi-dimensional loss prediction of the motor, and serves as the primary input for the second stage model, which integrates prediction accuracy and training efficiency. In the second stage, by comprehensively considering the spatial and temporal correlation features with the input layer reconstructed using correlation matrix weights, an improved convolutional neural network with bidirectional long short-term memory (CNN-BiLSTM) model is employed to predict multi-scale loads. Compared with traditional models, the proposed method exhibits significantly improved prediction accuracy, with R2 greater than 0.95. Furthermore, the damage error between the predicted load and the expected load is within 10%, which provides technical support for reliability evaluation and life prediction of electric motor.

Online Physical Education Teaching Resources Integration Method Based on Multi-Objective Optimization Algorithm

In order to improve the sharing level of online physical education teaching resources, the integration design of online physical education teaching resources is carried out by using MOOC resource information fusion technology and closed frequent item fusion calculation method, and the intelligent scheduling and management ability of online physical education teaching resources is improved. An online physical education teaching resource integration model based on big data information fusion and cluster scheduling is proposed. The factors of online physical education teaching demand scale are initialized, the structure of common factors of online physical education teaching resources is determined by factor analysis method, the orthogonal rotation of indicators by maximum variance method is adopted, the cloud computing model is adopted to analyze the integration structure of online physical education teaching resources, and the feature quantity of cross frequent items rules of online physical education teaching resources is extracted. The parallel K-means clustering model of online physical education resources integration is constructed, the fuzzy correlation degree features of online physical education resources are calculated, the extracted correlation features of online physical education resources are processed by C-means clustering method for big data fusion, the multi-objective optimization technology is adopted for online physical education resources integration and adaptive scheduling, and the adaptive scheduling ability of online physical education resources is improved. The SPSS statistical software is used for empirical analysis, and the results show that the integration degree of online physical education teaching resources by this method is high, which promotes the development of online physical education teaching.

Spatial Signatures of Electron Correlation in Least-Squares Tensor Hypercontraction.

Least-squares tensor hypercontraction (LS-THC) has received some attention in recent years as an approach to reduce the significant computational costs of wave function-based methods in quantum chemistry. However, previous work has demonstrated that LS-THC factorization performs disproportionately worse in the description of wave function components (e.g., cluster amplitudes T̂2) than Hamiltonian components (e.g., electron repulsion integrals (pq|rs)). This work develops novel theoretical methods to study the source of these errors in the context of the real-space T̂2 kernel, and reports, for the first time, the existence of a "correlation feature" in the errors of the LS-THC representation of the "exchange-like" correlation energy EX and T̂2 that is remarkably consistent across ten molecular species, three correlated wave functions, and four basis sets. This correlation feature portends the existence of a "pair point kernel" missing in the usual LS-THC representation of the wave function, which critically depends upon pairs of grid points situated close to atoms and with interpair distances between one and two Bohr radii. These findings point the way for future LS-THC developments to address these shortcomings.

TMVF: Trusted Multi-View Fish Behavior Recognition with correlative feature and adaptive evidence fusion

TMVF: Trusted Multi-View Fish Behavior Recognition with correlative feature and adaptive evidence fusion