33 results about "Spectral clustering" patented technology

Liquid chromatography coupled with tandem mass spectrometry (LC-MS / MS) is commonly adopted in large-scale glycoproteomic studies involving hundreds of disease and control samples. Current methods for glycopeptide identification in such data analyze the individual datasets and do not exploit redundant spectra of glycopeptides present in related datasets. A concurrent approach is provided for glycopeptide identification in multiple related glycoproteomic datasets by using spectral clustering and spectral library searching.

This invention relates to the field of task scheduling technology, and discloses a task scheduling method and system based on spectral clustering and auction algorithms. The method includes: acquiring registered resource information and task information to be scheduled; performing spectral clustering on the resource information and task information respectively to form resource group sets and task group sets; based on the auction algorithm, performing inter-group matching and task allocation on the resource group sets and task group sets to generate an initial allocation scheme; performing global balance adjustment on the initial allocation scheme to obtain a final allocation scheme, and performing task scheduling according to the final allocation scheme. This invention solves the problems of complex process scheduling easily getting trapped in local optima and insufficient dynamic response and cross-departmental collaboration in large-scale continuous task input scenarios, improving the global balance of resource allocation, real-time accuracy of scheduling, and business continuity assurance capabilities.

The application provides a lifting pump main motor bearing vibration lightweight intelligent prediction method, realizes accurate prediction of future state of lifting pump main motor bearing vibration, collects power consumption, power, current, temperature and water flow of the lifting pump main motor bearing as model input variables through a sensor, establishes a lifting pump main motor bearing vibration lightweight intelligent prediction model, constructs a multi-layer convolution module stacking structure to extract deep features of input data, relieves influence of insufficient generalization on model prediction performance, realizes accurate prediction of lifting pump main motor bearing vibration, designs a convolution module self-organizing reconstruction mechanism, groups similar convolution kernels by using spectral clustering, self-organizes a low-rank similar convolution kernel result matrix, and assigns the low-rank similar convolution kernel result matrix to a weight vector of a newly-built convolution kernel in each lightweight convolution module, so that the number of parameters of the prediction model is reduced.

A spectral clustering method and system based on unified anchor and subspace learning is provided. The spectral clustering method based on unified anchor and subspace learning includes: S1: acquiring a clustering task and a target data sample; S2: performing unified anchor learning on multi-view data corresponding to the acquired clustering task and the acquired target data sample, and adaptively constructing an objective function corresponding to an anchor graph according to a learned unified anchor; S3: optimizing the constructed objective function by using an alternating optimization method to obtain an optimized unified anchor graph; and S4: performing spectral clustering on the obtained optimized unified anchor graph to obtain a final clustering result.

The application relates to a social network text target topic detection method based on sparse subspace clustering, and belongs to the technical field of computer data processing of natural language processing and social network data mining.The method writes text identification into target platform text content and interaction event records, carries out word segmentation, denoising and vectorization processing, reduces short text noise and event mismatch interference on subsequent analysis, constructs a social relationship graph, calculates edge confidence weight to form a graph regular constraint parameter, reduces the weight of a low-confidence screen edge in the constraint, suppresses relationship noise caused by organized manipulation from the source, solves sparse representation coefficients, constructs a similarity matrix, performs spectral clustering, enhances the separability of samples with similar semantics but different propagation modes, calculates semantic cohesion and propagation deviation, determines a target topic cluster by using a double threshold, extracts a key text set and a propagation evidence set, forms a reviewable target topic detection result, and improves target topic detection rate and reduces false positives.

The application is suitable for the technical field of computers, and provides a federated learning method for heterogeneous clients, comprising: distributing an initial global model to all client models for training, and collecting model update parameters of all client models after training; based on the model update parameters, constructing multi-view feature representation; based on the multi-view feature representation, using a spectral clustering method for client clustering grouping to obtain multiple client clusters; using a two-stage aggregation method to aggregate each client cluster to obtain global gradient parameters, iteratively update the initial global model, and obtain a final global model. The application solves the problems of poor single-view gradient clustering accuracy in the first round of existing federated learning and non-discriminative aggregation of old client gradients, realizes high-precision clustering grouping and dynamic fair aggregation under the premise of not leaking original data privacy, and significantly improves the generalization ability of the global model and the training efficiency of large-scale systems in a heterogeneous environment.

Techniques are described for providing a ML data analytics application including guided ML workflows that facilitate the end-to-end training and use of various types of ML models, where such guided workflows may also be referred to as ML “experiments.” For example, the ML data analytics application may enable users to create experiments related to prediction of numeric fields (for example, using linear regression techniques), predicting categorical fields (for example, using logistic regression), detecting numerical outliers (for example, using various distribution statistics), detecting categorical outliers (for example, using probabilistic statistics), forecasting time series data, and clustering numeric events (for example, using k-means, density-based spatial clustering of applications with noise (DBSCAN), spectral clustering, or other techniques), among other possible uses of various types of ML models to analyze data.

This invention discloses a method and system for intelligent scheduling of garbage trucks based on reinforcement learning, specifically relating to the field of intelligent scheduling technology. The method involves acquiring urban garbage collection road network topology, historical garbage production spatiotemporal distribution data, and garbage truck trajectory data; calculating the negative coupling coefficient of each intersection segment to the scheduling strategy of adjacent routes to form a negative coupling matrix; performing spectral clustering on the negative coupling matrix to divide the coupling-sensitive subgraph and extracting a set of key intersection segments; inputting the set of key intersection segments and historical garbage production spatiotemporal distribution data into a multi-agent reinforcement learning framework to construct a scheduling strategy model and output initial strategy parameters; generating a collaborative scheduling scheme based on the initial strategy parameters in a simulation environment and incrementally correcting the initial strategy parameters to obtain a converged scheduling strategy; determining the garbage truck departure time and travel path based on the converged scheduling strategy and writing it back to the route interaction graph for iterative optimization.

This invention discloses a method, system, and device for single-cell transcriptome clustering based on deep autoencoders. The method includes: data preprocessing and feature engineering, construction of a deep autoencoder network, design of a multi-task loss function, model training and feature extraction, dimensionality reduction visualization, and spectral clustering analysis. The system includes: a data preprocessing module, a deep autoencoder module, a loss function calculation module, a model training module, a dimensionality reduction visualization module, a clustering analysis module, and an evaluation module. This invention significantly improves the clustering accuracy of single-cell data by introducing a contrastive learning mechanism and optimizing the clustering strategy.

This invention proposes a method and system for ECG clustering based on multi-lead semantic consistency, belonging to the field of ECG signal processing technology. The method includes: constructing a similarity correlation matrix for an ECG signal set using adaptive graph learning; learning the spectral representation of each lead of the ECG signal using spectral clustering on the similarity correlation matrix, whereby spectral clustering utilizes multi-lead shared spectral embedding to measure the correlation between ECG signals, obtained through an optimized minimum edge weight objective function; wherein the optimized minimum edge weight objective function includes the multi-lead shared semantic similarity matrix; after obtaining the spectral embedding matrix, constructing a spectral rotation objective function by introducing an orthogonal rotation factor matrix; merging the minimum edge weight objective function and the spectral rotation objective function into a total objective function, and optimizing the total objective function using the alternating direction multiplier method to obtain the ECG clustering result. This invention significantly improves the clustering quality through the learning of semantically consistent graphs.

The application relates to a single-phase ground fault early warning method and system based on dynamic clustering analysis, in the system, a data acquisition module collects real-time recording wave data of three-phase current and zero sequence current, and a characteristic waveform is constructed based on the recording wave data; a multi-scale feature extraction module calculates a plurality of dimensions of current feature vectors based on the characteristic waveform, normalizes the plurality of dimensions of current feature vectors, and generates a high-dimensional feature vector; an adaptive dynamic clustering module performs offline spectral clustering based on historical data to obtain an initial cluster; a local neighborhood graph is established with the initial cluster center as a node; a mode migration tracking module captures the dynamic change path of a new sample in a feature space based on the high-dimensional feature vector, calculates a mode migration recognition index and a Mahalanobis distance value; and a multi-level early warning decision module executes multi-level early warning determination based on the mode migration recognition index and the minimum Mahalanobis distance value, and generates corresponding early warning events when different early warning conditions are met.

This invention discloses an intelligent extraction method and system for typical micro-topography in icing areas of transmission lines, relating to the field of intelligent operation and maintenance technology for power systems. The method includes extracting meteorological data from a standardized matrix, calculating effective degrees of freedom and gradient values, correcting the meteorological data to obtain corrected meteorological data, updating the enhanced feature vector using the corrected meteorological data to obtain an updated feature vector, obtaining feature components through multi-scale wavelet decomposition to generate a normalized multi-scale vector, obtaining candidate regions through K-means clustering, and generating a typical micro-topography label map and an icing sensitivity matrix. By constructing a standardized spatial fusion matrix and adaptive modeling of multimodal temporal features, the method achieves accurate quantification and dynamic extraction of typical micro-topography features. Combined with scale entropy enhancement and spectral clustering optimization, the method improves the spatial resolution and classification accuracy of typical micro-topography in icing areas.

The application discloses a data insight report generation method and system based on knowledge enhancement and fact verification. A domain index relationship knowledge graph is constructed by training a large language model; the insight is mapped to the graph, a weighted insight graph is constructed, and spectral clustering is performed to divide the insight theme area; a viewpoint unit is generated by an intelligent agent, and consistency verification and correction are performed by a fact verification intelligent agent; user intent is analyzed, candidate themes are screened, a report outline is generated by heuristic search arrangement based on SRCI score, semantic progression and logic entropy calculation; finally, verified viewpoint units are integrated according to the outline, and the final report is output. The application realizes the generation of a fully automatic and high-credibility report, and improves accuracy, logic and explainability.

This invention belongs to the field of multi-unmanned surface vessel (USV) path planning technology. It uses rectangular decomposition results as basic units to construct a topological intensity map representing spatial adjacency density, and utilizes Laplace eigenmaps to capture the topological similarity between rectangles. Subsequently, a weight discretization allocation strategy is proposed, which transforms the weighted graph partitioning problem into a linear sum allocation problem by splitting rectangles with heterogeneous areas into isomorphic virtual nodes. This ultimately yields a region allocation scheme that balances strict load balancing and spatial compactness. Finally, through connectivity repair and boundary fine-tuning, the final multi-USV coverage path is output. By maximizing the path compactness within sub-regions, this invention ensures that adjacent survey lines belong to the same operational unit, providing a high-quality trajectory foundation for sonar data fusion using overlapping areas and compensating for edge imaging defects. This helps improve the coverage detection path quality of multi-USV systems in underwater target search missions.

This invention discloses a method and system for assessing depression risk based on big data, relating to the field of big data behavioral analysis technology. The method includes constructing feature vectors for each time point, generating a vector set, setting a scale set using linear sampling and generating domain labels using a clustering algorithm, calculating a core score, generating difference sequences from the homology dimension and stacking them into a sequence matrix, extracting the first principal component using PCA to obtain a topological kernel vector and constructing a topological kernel matrix, obtaining residual perturbation vectors through singular value decomposition and performing discrete Fourier transform to obtain spectral values, calculating the split point index, and combining the stability score to obtain the depression risk. By constructing a manifold point cloud of behavioral vectors, the evolution trajectory of depression risk is accurately assessed. By extracting topological stability indicators and spectral clustering results, the detection accuracy and assessment reliability of persistent state anomalies are improved. The introduction of residual perturbation and frequency domain analysis mechanisms enhances the timeliness and accuracy of early warning of depression risk.

The present application relates to the technical field of image generation, in particular to a method and device for anti-homogenization training of a diffusion model, wherein the method comprises: obtaining a group of images generated by the same prompt word, and constructing a distance matrix of the group of images; performing spectral clustering on the distance matrix to divide the group of images into different semantic clusters; adaptively assigning an exploration reward according to the size of each semantic cluster to obtain a plurality of semantic clusters after the exploration reward; and inputting the plurality of semantic clusters after the exploration reward into the diffusion model after structure perception regularization to obtain a group of images after generalization. Thus, the mode collapse and other problems caused by the existing group relative strategy optimization method in image generation are solved.

This invention provides a short text clustering and fuzzy recognition algorithm for large-scale online subgraph sampling, comprising the following steps: Step S1, extraction and preprocessing of training samples; Step S2, construction of the neural network; Step S3, overall clustering prediction; Step S4, fuzzy sample recognition; Step S5, retraining of the neural network. This invention combines short text clustering with a large language model, which not only improves clustering accuracy but also enables the handling of clustering tasks with different themes and classification requirements, significantly reducing the manual cost of data annotation. Furthermore, this invention can annotate fuzzy samples for classification, using K-nearest neighbors combined with minimum spanning trees to assist subgraph sampling in the selection scheme. This utilizes sparse structure to reduce computational costs and exposes the fluctuations of boundary samples through spectral clustering, providing a more comprehensive perspective for fuzzy sample selection and improving the accuracy and interpretability of the clustering results.

The application discloses a high-speed rail second-level wind speed prediction and early warning method based on wind process dynamic division and physical guidance, comprising the following steps: in stage one, wind process dynamic division and offline construction of a knowledge base, historical wind field data are divided into wind process types with clear physical meaning through a graph attention network and adaptive spectral clustering; in stage two, hierarchical online wind speed prediction based on physical guidance, macro-, meso- and micro-scale decoupling deduction is carried out based on a hierarchical state space model with physical guidance, and a multi-physical constraint loss function is used to ensure that the prediction conforms to the fluid mechanics law; in stage three, dynamic adaptive envelope line online early warning, a dynamic safety envelope line is generated by using the extreme value theory and real-time scene information, and scene-adaptive accurate hierarchical early warning is realized. The method can significantly improve the accuracy of high-speed rail wind speed prediction and the reliability of early warning.

The application discloses a brain representative functional tissue exploration method based on multi-view function gradient density, comprising the following steps: obtaining resting-state functional magnetic resonance imaging data and preprocessing to obtain standardized resting-state image data; processing the standardized resting-state image data based on resting-state functional connectivity to generate a function gradient density map; adopting a spherical wavelet transform method to perform multi-scale decomposition on the function gradient density map to obtain view data corresponding to multiple spatial frequency scales; calculating cross-individual differences of function gradient density features of different individuals under each view based on a Wasserstein distance, and constructing a similarity matrix corresponding to each view; adopting a multi-view spectral clustering method to perform fusion analysis on the similarity matrix to identify a brain representative functional tissue mode; and visually displaying the representative functional tissue mode; and fully considering the cerebral cortex geometric structure and multi-scale functional features, realizing stable identification and individual difference analysis of the brain functional tissue mode.

This invention designs an image classification method based on machine learning and spectral clustering of grains, comprising: acquiring an image dataset; extracting the feature vector of each image in the image dataset using a feature extraction model; creating an initial grain as a whole by treating the feature vectors of all images in the image dataset, and calculating the variable sparsity measure (VSM) value of the initial grain; if the VSM value is lower than a preset threshold, dividing the grain into two smaller grains using a 2-means clustering algorithm; continuing until the VSM values ​​of all grains meet the condition; calculating the similarity between the generated grains based on the center feature vector and the radius of the grains; dividing the generated grains into multiple clusters using spectral clustering based on the similarity between the grains, where each cluster represents a category; calculating the similarity between the feature vector of the image sample to be tested and the center feature vector of each grain, and taking the category corresponding to the cluster of the most similar grain as the classification result of the image sample to be tested. This invention provides more reliable and efficient technical support for the field of image classification.

The present application belongs to the technical field of power system operation and optimization, and particularly relates to a multi-scenario generation method based on probability sampling and improved clustering. In view of the problem that single prediction cannot cover the operation boundary due to the strong space-time coupling of power grid source and load under extreme weather, by determining the source and load parameters and the correlation control model, a large number of initial scenarios are generated by using Latin hypercube sampling, and are reduced by using the Canopy-spectrum clustering-Kmeans hybrid algorithm, and the typical scenario set and the occurrence probability are output by combining double indexes, so as to provide quantifiable uncertainty support for power grid random planning, and the method has the advantages of authenticity and high efficiency.

The present application provides a kind of for image registration based on neighborhood information and greedy search deterministic model fitting method, comprising: S1, the key point and descriptor of given image pair are extracted, and initial matching set is constructed according to descriptor similarity;S2, high probability inlier set is identified and obtained;S3, neighborhood set is constructed for each inlier;S4, the best model hypothesis is generated using the hypothesis optimization strategy based on greedy search;S5, the noise scale of best model hypothesis and inlier set I are obtained;S6, if the data volume of temporary inlier set is less than k, then it is equal to inlier set;S7, the residual vector between initial matching set and best model hypothesis is calculated, and affine matrix H is calculated;S8, if current sampling times is less than η, then return S4;Otherwise, S9 is executed;S9, distance matrix A is calculated by H and A is sparsified to obtain G;S10, cluster label is obtained using spectral clustering on G;S11, if only one model instance is contained in data, then label is updated by post-processing step.

This invention discloses an intelligent classification method for the maturity of Torreya grandis (Chinese nut) from the perspective of unmanned aerial vehicles (UAVs), relating to the fields of smart agriculture and remote sensing image processing. It simultaneously acquires RGB visible light and near-infrared images of the Torreya grandis canopy, performing dark channel prior dehazing and canopy masking. Gaussian difference filtering is used to extract spectral features from the near-infrared images, generating a fruit saliency heatmap. A dual-channel YOLOv8-Lite lightweight network is constructed, using the visible light branch for small target localization and the near-infrared branch for maturity discrimination. Cross-modal gating attention enables adaptive fusion of dual-modal features. A decoupled three-classification decision head is designed based on spectral thresholds and color gradients, and a multi-task loss function is introduced for classification. Spectral clustering is used to optimize non-maximum suppression, and combined with UAV POS data, accurate mapping from pixel coordinates to WGS84 geographic coordinates is achieved. This invention can effectively improve the accuracy and robustness of Torreya grandis maturity recognition in complex environments, achieving lightweight real-time inference and geographic location output.

This invention relates to a method and system for generating a unified spatial multi-omics data latent representation. The method includes: acquiring spatial multi-omics data containing at least two omics modalities and preprocessing it to obtain spatial multi-omics features; constructing a modality-specific variational autoencoder for each omics modality to obtain a modality-specific latent representation; introducing a consistency alignment constraint by aligning the variational autoencoders to force the latent representations of different modalities to be consistent in a shared latent space; constraining the latent representations in the shared latent space using spectral clustering loss; fusing the posterior distributions of each modality in the shared latent space using an expert product-based mechanism; and reparameterizing the cross-modal integrated posterior distributions to obtain a unified latent representation as the unified latent representation output for the spatial multi-omics data, enabling different omics modalities to achieve semantically, structurally, and probabilistically consistent collaborative representations in a unified latent space.

The application discloses a job intelligent supervision method and system based on multi-modal data fusion, relates to the technical field of petrochemical operation supervision, and comprises the following steps: collecting multi-dimensional anti-explosion risk data, normalizing and calculating the overall risk degree according to a mapping rule table, and generating a high-risk alignment frame when a certain number of frames continuously exceed a threshold value; subsequently collecting personnel coordinates and electronic operation ticket data of UWB / BLE combined positioning, generating a three-modal atomic event with a 'high-risk grid-ticket number' double key, forming a three-modal continuous time sequence after alignment logic time stamping, and extracting an abnormal merged section; then selecting a multi-level sliding window according to the section length, calculating the cross-modal average correlation degree in the window, and generating an abnormal cluster by using spectral clustering or DBSCAN; calculating the confidence score of the sliding window, judging the final abnormal event and whether to switch the window level, and dynamically adjusting the confidence threshold and the window weight according to the pass rate; and the application realizes second-level accurate early warning of high-risk no-ticket operation in petrochemical industry, reduces the missed detection rate, and simplifies operation and maintenance.

This invention discloses a multi-domain voltage fingerprinting method and system for identifying transformer-household relationships using a synergistic computing module, belonging to the field of power distribution network management technology. The method involves acquiring the voltage time-domain waveform via high-speed sampling using an HPLC synergistic fusion module, followed by adaptive variational mode decoupling, multi-scale permutation entropy-scattering feature joint characterization, and continuous homology topological feature extraction. After confidence-weighted concatenation, transformer-household relationship discrimination is achieved through Riemannian manifold combined with spectral clustering. The system consists of a distributed HPLC synergistic fusion module and a transformer concentrator. The module performs local feature extraction and uploads compressed feature vectors, while the concentrator performs feature fusion, distance calculation, and cluster recognition. This invention introduces continuous homology into transformer-household identification for the first time. Pure voltage-driven operation requires no additional hardware, edge computing significantly reduces communication overhead, and the system boasts high accuracy, strong noise robustness, and adaptability to existing equipment upgrades, improving the level of refined management in transformer substations.

This invention discloses a method, system, and device for hyperspectral anomaly detection based on a multi-scale grid window, belonging to the field of hyperspectral image processing technology. The method includes: acquiring a hyperspectral image; constructing a spectral gradient tensor to calculate a multi-scale spectral abrupt change intensity map and determining the optimal window size; extracting heterogeneous spectral core regions; filtering pure background regions through spectral clustering and correcting the generalized likelihood ratio statistic to obtain anomaly scores; constructing an attention weight map to generate a multi-scale enhanced anomaly score map and calculating saliency weights for fusion; optimizing spatial consistency through a Markov random field and combining historical data with dynamic thresholds to output the detection results. This invention solves the problems of local contrast calculation being affected by multiple anomalies, background estimation ignoring spectral purity and spatial correlation, and poor detection robustness due to lack of historical time-series data utilization. It achieves accurate hyperspectral anomaly identification from coarse localization to fine detection, and from single-frame analysis to historical verification.

This invention discloses a real-time health monitoring system and method for concrete structures based on multi-source sensor fusion, comprising the following steps: S1, deploying sensors and collecting data to construct a spatiotemporal index sequence; S2, performing time warping and frequency domain correction to eliminate asynchronous offsets and generate a spatiotemporal feature tensor; S3, uploading to a cloud platform, constructing a hypergraph propagation network, and generating a structural mechanics coupling feature map; S4, performing graph convolution operations to construct a temporal propagation chain and generate a structural damage matrix; S5, performing spectral clustering processing to construct a health status map and complete damage trajectory inference; S6, performing threshold mapping processing to generate early warning results and output real-time monitoring results. This invention achieves real-time and accurate monitoring and adaptive early warning of the health status of concrete structures through multi-source sensor data acquisition, spatiotemporal feature fusion, hypergraph network modeling, and intelligent damage inference.

This invention provides a non-technical line loss status assessment method based on multi-scale spatiotemporal feature fusion in low-voltage distribution network environments, relating to the fields of power big data analysis and intelligent operation and maintenance of distribution networks. The invention constructs a fusion architecture based on multi-scale temporal convolutional networks and long short-term memory networks; extracts multi-scale spatiotemporal features of users from instantaneous power consumption fluctuations to periodic load patterns using MSTBlock units; designs a cluster balance constraint mechanism to ensure that sparse but critical non-technical line loss anomaly warning signals are not obscured by massive amounts of normal power consumption data; adaptively selects graph segmentation or spectral clustering integration strategies based on data scale to output clustering labels, which are then mapped to the evolution trajectory of user power consumption behavior. This invention can identify the level of power consumption anomalies from raw load signals with random fluctuations and interference, and calculates the investigation priority by combining transformer area correlation analysis, significantly improving the accuracy and interpretability of unsupervised assessment decisions for non-technical line losses in distribution networks.

This application discloses an adaptive weighted constraint subspace block diagonal representation image clustering method and system. The method includes: acquiring image data to be clustered and establishing a data matrix; calculating class weighting constraints based on the data matrix and Pearson correlation weighting constraints based on the image data; establishing a diagonal representation model objective function based on the class weighting constraints and Pearson correlation weighting constraints, and optimizing the self-representation coefficient matrix of the diagonal representation model objective function; constructing an adjacency matrix based on the self-representation coefficient matrix, and establishing an undirected weighted graph based on the adjacency matrix and image data; and performing spectral clustering processing on the undirected weighted graph to obtain cluster labels. This application integrates the block diagonal regularization of the self-representation coefficient matrix obtained from the linear reconstruction process with the class weighting constraints and Pearson correlation weighting constraints describing data relationships, thereby further improving the block diagonal structure of the self-representation coefficient matrix, enhancing robustness to outliers, and improving clustering performance.