Patents
Literature
22results about How to "Easy to explain" patented technology
Efficacy Topic
Property
Owner
Technical Advancement
Application Domain
Technology Topic
Technology Field Word
Patent Country/Region
Patent Type
Patent Status
Application Year
Inventor
Method for predicting brain task activation patterns using combined structural and functional magnetic resonance imaging
The application discloses a method for predicting brain task activation patterns by combining structural and functional magnetic resonance imaging, and belongs to the cross field of biomedical engineering and pattern recognition. First, the method obtains brain function features, structural features and Prompt text through data preprocessing, and then respectively establishes a Prompt text coding model for embedding the text of the Prompt into a text feature space, extracts high-level semantic features of the brain, and fuses a backbone network model of features of two modalities of functions and structures, and a Prompt interaction module for fusing Prompt information so that the prediction result of the model can be more easily explained and understood. The model is optimized based on a classification-contrast loss objective function, and the brain task activation patterns in a task state are predicted from the brain task activation patterns in a non-task state. The application improves the prediction accuracy of the model, and is helpful for predicting the brain task activation patterns of patients with mental diseases, the elderly or infant groups.
Owner:UNIV OF ELECTRONICS SCI & TECH OF CHINA
A nonlinear data classification method and system based on deep transfer learning visual recognition capability
This invention belongs to the field of nonlinear data classification technology and discloses a nonlinear data classification method based on deep transfer learning and visual recognition capabilities. This invention transforms high-dimensional numerical features into structured images, enabling convolutional neural networks to leverage their powerful visual recognition capabilities to handle tabular data classification tasks. This image-based representation naturally reveals the implicit nonlinear relationships within the data. By further integrating deep transfer learning, this method effectively alleviates the data sparsity problem in small-sample scenarios and improves generalization performance through a series of training optimization strategies. The framework proposed in this invention achieves an effective connection between the numerical feature space and the image-based deep learning architecture, providing theoretical innovation and practical value for complex nonlinear data classification tasks.
Owner:ZHEJIANG POLICE COLLEGE
Data-driven intelligent analysis methods, systems, storage media, and computer equipment
ActiveCN122088710Aenhance reasoning abilityimprove interpretabilityBiological modelsInference methodsLinguistic modelData description
This application discloses a data-aware intelligent analysis method, system, storage medium, and computer device. The method includes: receiving a target problem and initializing state information, wherein the state information includes key sub-conclusions, a reasoning state logic diagram, and key data descriptions; a language model-based agent iteratively performs the following operations: obtaining input information for the current step; updating the state information of the current step based on the action and environmental feedback information of the previous step, and making decisions based on the state information of the current step to obtain the action and tools to be invoked for the current step; invoking the tools to execute the action of the current step, and obtaining environmental feedback information for the current step after the action is completed; checking whether the target problem has been analyzed; if not, proceeding to the next loop; if completed, generating analysis results. This method can improve the reasoning ability, response consistency, and interpretability of deep reasoning models.
Owner:RAJAX NETWORK &TECHNOLOGY (SHANGHAI) CO LTD
Attribute decomposition-aggregation based open vocabulary semantic segmentation method and system
ActiveCN116935399BComplete segmentation resultsThe segmentation result is accurateCharacter and pattern recognitionNeural learning methodsFeature extractionImage segmentation
The application provides an open vocabulary semantic segmentation method and system, comprising: decomposing target category text into a set of attribute texts; using a visual feature extractor to encode a target picture into visual features; using a text feature extractor to encode the attribute texts into attribute features; feeding the visual features and the attribute features into an attribute aggregation network to obtain a segmentation probability map, and converting the segmentation probability map into a segmentation result using a threshold method; calculating the loss of the predicted segmentation result and a real segmentation mask, and training using a gradient descent method; and using the trained model to infer test images and test attribute texts to obtain test image segmentation results. The application proposes a decomposition-aggregation strategy, which greatly improves the accuracy of open vocabulary semantic segmentation.
Owner:SHANGHAI JIAOTONG UNIV
A method for fan imaging of ground penetrating radar data
ActiveCN115616673Bdescribe clearly and accuratelyprecise positioningDetection using electromagnetic wavesThree-dimensional spaceBoundary extension
The application provides a geological radar data fan imaging method, which performs boundary extension on two-dimensional rectangular profile data of a geological radar, and then performs third-order interpolation between channels and triangular projection transformation. Through conversion of the two-dimensional rectangular profile to the fan profile, the morphological characteristics of the geological radar measuring line in the three-dimensional space can be more clearly and accurately described, and target positioning and interpretation in the three-dimensional space can be realized.
Owner:CHINA UNIV OF MINING & TECH (BEIJING)
Indoor design map expression method and display device thereof
The application discloses an indoor design picture performance method and a display device thereof and relates to the technical field of display equipment. The technical scheme is as follows: a rectangular storage box and a rectangular collecting box are arranged, a display unit is arranged between the rectangular storage box and the rectangular collecting box, a treading lifting mechanism is arranged below the rectangular storage box, a plurality of picture frames arranged linearly are stored in the rectangular storage box, and the treading lifting mechanism is used for driving the picture frames in the rectangular storage box to ascend. The application has the advantages that the device is designed ingeniously and is easy to operate. The picture frames can be adsorbed by the treading lifting mechanism, the adsorbed picture frames can be turned to the front side by the groove wheel mechanism, the collection frame for adsorbing the picture frames is turned over by the rotation of the second motor, the device is convenient for explanation and observation, the collection frame for adsorbing the picture frames is attached to the rectangular collecting box when the collection frame is turned over, the picture frames are collected, and the practicability of the device is improved.
Owner:TIANJIN POLYTECHNIC UNIV
A spatial noise annoyance modeling method based on binaural psychoacoustic parameter difference
ActiveCN115630480BImprove forecast accuracyEasy to explainSignal processingHeadphones for stereophonic communicationSound sourcesSpatial noise
The application provides a spatial noise annoyance modeling method based on binaural psychoacoustic parameter difference, and the method uses the binaural psychoacoustic parameter difference as a bridge, establishes a spatial noise annoyance model based on noise self annoyance by using multivariate linear regression, and realizes correction of noise self annoyance in the sound source direction, so that the annoyance of noise in different directions is more reasonably and accurately predicted, the use range of the noise annoyance model is widened, the physical mechanism of the spatial noise annoyance is explained, and the application value of the noise annoyance model is enhanced.
Owner:NORTHWESTERN POLYTECHNICAL UNIV
Transformer mechanical state in-situ sensing method based on space magnetic field modal evolution
PendingCN121978591AAvoid glitchy upgradesResolving ambiguous diagnostic issuesMagnetic measurementsElectric winding testingComputational physicsCondensed matter physics
The invention provides a transformer mechanical state original taste sensing method based on spatial magnetic field modal evolution, which comprises the following steps: extracting a steady-state magnetic field component and a vibration magnetic field component in a transformer, and obtaining a first detection feature and a second detection feature according to the steady-state magnetic field component, diagnosing whether the magnetic circuit state of the transformer core is normal based on the first detection feature and the second detection feature; and diagnosing whether the axial and radial vibration modes of the transformer winding are normal or not according to the vibration magnetic field component. According to the invention, the internal magnetic field space modal characteristics and the vibration frequency characteristics of the transformer are fused, and the external measurement signals of the transformer are strongly correlated with the change of the internal mechanical state, so that the fault types of the winding and the iron core can be effectively distinguished, the fault area can be preliminarily positioned, and the diagnosis accuracy is greatly improved.
Owner:EAGLERISE MAGNETOELECTRIC TECH (JI AN) CO LTD
Multistage CDN flow scheduling method based on linear constraint
The invention relates to the technical field of computer networks, and discloses a multistage CDN (Content Delivery Network) traffic scheduling method based on linear constraint, which comprises the following steps of: receiving an access request analyzed by a local DNS (Domain Name System) to form a scheduling instance to be optimized; carrying out complexity discrimination on the scheduling instance, and determining an applicable scheduling category of the scheduling instance; establishing a performance upper limit model related to server hardware resources; functional linear constraints are introduced, and Big-M parameters related to the constraints are finely set; and server cluster information, server information, quality constraint information and region hijacking information are obtained, joint solving is performed on the performance upper limit model and the functional constraint, a target cache node IP address meeting all constraint conditions is determined and fed back to a local DNS, and response and scheduling of an access request are completed. According to the invention, through a multi-stage scheduling framework and refined Big-M parameter setting, the computational efficiency and resource utilization precision of CDN scheduling in a complex network environment are improved.
Owner:凌川峰(贵州)信息技术有限公司 +1
Function annotation abundance sequence-based base model training method and device
PendingCN121963875Areduce duplicationlow costBiostatisticsBiological modelsOpen reading frameAlgorithm
The invention relates to a base model training method and device based on a functional annotation abundance sequence. The method comprises the following steps: S1, carrying out function annotation on an open reading frame of a genome or metagenome sample; s2, counting the occurrence frequency of each annotation and constructing a sequence according to an abundance descending order; s3, after the sequence is subjected to token processing, inputting the sequence into a model based on Transform, and adopting joint training of language modeling, comparative learning and classification loss to obtain species-level and token-level fixed dimension embedding; s4, on the basis of the embedding, completing downstream tasks such as phylogenetic tree construction, species identification and phenotype prediction, BGC / MGC recognition and key gene positioning in three levels, namely a genome level, a gene cluster level and a gene / protein level. In the embodiment of the invention, good uniformity, expandability and interpretability are displayed, and the dependence on a reference database is reduced. The corresponding device comprises a data processing module, a model training module and an application module, and can be realized by program instructions in a computer readable storage medium.
Owner:ZHEJIANG LAB
Automatic test system and method based on relay protection device
The invention discloses an automatic test system and method based on a relay protection device, and relates to the technical field of relay protection device analysis. In the relay protection test process, electrical time sequence data and device state signals in the test stage are collected in real time, data are processed, and a scene data unit capable of being independently stored and analyzed is generated; constructing a test path transfer directed graph; predicting a subsequent stage candidate set according to the test path transfer directed graph; forming a to-be-executed candidate list according to the predicted subsequent stage candidate set in combination with the current available resource set; analyzing a resource conflict relationship among the test items, and generating an optimized test sequence; when it is detected that the test process is abnormally interrupted, a continuous test execution instruction is generated by conducting mode matching on the context and a historical successful continuous test case, the test is recovered from the breakpoint, and meanwhile the interruption and recovery mode is stored in a historical continuous test case library. And the success rate and the safety of continuous testing are improved.
Owner:ZHEJIANG ZHENENG LANXI POWER GENERATION CO LTD
A method for product search with query-dependent multi-faceted explainability
ActiveCN117911110BEasy to explainImprove retrieval performanceTheoretical computer scienceSearch intent
The application realizes a commodity search method with query-related multi-aspect explainability by means of a method in the field of artificial intelligence. The search information of an input commodity is searched and the query commodity result is fed back by using a multi-path query perception graph convolution network for explainable commodity search. The multi-path query perception graph convolution network for explainable commodity search comprises two components: a query perception graph convolution sorter and a query perception multi-path reasoner. The query perception graph convolution sorter models the representation of a user and a commodity according to different knowledge relationship domains in a knowledge graph by using a graph convolution network. The query perception multi-path reasoner is responsible for exploring a query-specific multi-path from the knowledge graph to meet the search intention of the user. The two components share basic parameters and are collaboratively trained to constitute a complete network. The method can significantly improve the retrieval performance and generate better explanations for the search result.
Owner:BEIJING NORMAL UNIVERSITY
Automobile paint performance testing device
ActiveCN224263202Ureduced simulation timelow costMachine part testingVehicle testingProgrammable logic controllerElectric machinery
The utility model discloses an automobile paint performance testing device. The automobile paint performance testing device comprises a brush, a motor, a sample tool, a PLC (Programmable Logic Controller) and a liquid spraying mechanism, the sample tool is arranged below the brush and is used for fixing an automobile paint plate sample; the brush is connected with an output shaft of the motor, the liquid spraying mechanism is used for spraying water flow containing smoke dust to the automobile panel during testing, and the PLC is a controller of the motor. According to the utility model, a sample can be repeatedly scrubbed under different environments and working conditions of a vehicle, such as mixed water added with dust and a vehicle washing medium, and the paint layer condition of the automobile panel sample after different scrubbing times can be observed, so that a reliable basis is provided for the real use performance of automobile paint, and research and development and production data can be provided for OEM.
Owner:INTERTEK TESTING SERVICES
Traditional Chinese medicine planting site selection evaluation method based on big data
PendingCN121961107AEnsure data coverageensure continuityData processing applicationsData miningBiology
The invention relates to the technical field of agricultural big data and intelligent traditional Chinese medicine planting evaluation, and discloses a traditional Chinese medicine planting site selection evaluation method based on big data. According to the scheme, sunshine and soil salt content collection devices are deployed in candidate places, sampling is carried out daily to form two types of sequences, and a mean value is solved; normalizing a global minimum value and a global maximum value into standardized sunlight and low salt indexes, and carrying out arithmetic averaging to obtain baseline suitability; a perturbation step length is set according to the range and the candidate number, positive and negative disturbance and amplitude limiting are carried out on sunlight and salt, and four types of disturbance baselines are reconstructed after normalization; the sunlight sensitivity and the salt-containing sensitivity are added according to the ratio of the actual standardized step length to the suitability change, a penalty term is generated by taking the maximum value of the whole, the robustness suitability is obtained through deduction and amplitude limiting, and then quantile grading is used; and finally summarizing the longitude and latitude to form a structured result table.
Owner:新疆农业职业技术大学
Method for constructing a model for regulating analysis of sinusitis based on data of inhibition of nasal polyp cells
This invention discloses a method for constructing a sinusitis regulation analysis model based on nasal polyp cell inhibition data, belonging to the field of CRSwNP disease analysis technology. The method includes the following steps: collecting Lund-Kennedy scores and corresponding nasal polyp cell inhibition data from CRSwNP patients to obtain raw sample data; screening the raw sample data to remove unqualified raw sample data from CRSwNP patients to obtain standard sample data; performing index preprocessing based on the standard sample data and obtaining the optimal lag order to obtain reference sample data; constructing an LK analysis prediction model based on the reference sample data and analyzing and predicting the Lund-Kennedy scores of CRSwNP patients. This invention addresses the problem that existing CRSwNP disease analysis technologies cannot provide clinicians with objective and reliable analytical and predictive tools when analyzing and predicting the Lund-Kennedy scores of CRSwNP patients, thus failing to help clinicians reliably predict the disease progression of CRSwNP patients during treatment.
Owner:TIANJIN MEDICAL UNIVERSITY GENERAL HOSPITAL
A multi-modal medical recommendation method based on a double-layer gating mechanism and related devices
ActiveCN121709125Bovercome limitationsguaranteed retentionMedical data miningTherapiesData classData set
The application discloses a multi-modal medical recommendation method based on a double-layer gating mechanism and related devices, and relates to the technical field of artificial intelligence medical treatment. The method obtains text data, numerical data and category data of a patient, and pre-processes the data to obtain a data set; the data set is input into a multi-task learning model for training until a training round is reached or a preset condition is met, a trained multi-task learning model is obtained, the multi-task learning model extracts features from the text data, the numerical data and the category data through a text feature branch, a numerical feature branch and a category feature branch respectively to obtain text features, numerical features and category features; the text features, the numerical features and the category features are simultaneously input into a double-layer gating network for mode-level gating and feature-level gating processing to obtain final feature representations; and the final feature representations are input into a multi-task head network to obtain classification prediction results of a plurality of tasks. The application can provide accurate personalized medical recommendations for patients.
Owner:HAINAN UNIV
Code merging conflict automatic resolution method based on editing script identification and pre-training supervision fine tuning
PendingCN121996289AStrong explainabilityThe final solution is reasonableCode refactoringBiological modelsAlgorithmTheoretical computer science
The invention relates to a code merging conflict automatic resolution method based on editing script recognition and pre-training supervision fine tuning, and mainly solves the problem of code merging conflicts in the field of software development. The method comprises the following steps: firstly, extracting all merging submissions in a submission history from a screened open source Git code warehouse; replaying and merging to extract information construction features such as contents and contexts of all conflict blocks, extracting an editing script positioning solution by using an algorithm, sending the editing script into a pre-training model to output embedding of code change, sending the editing script into a downstream recurrent neural network to output a prediction condition accepted by the script, and meanwhile, finely adjusting the pre-training model; and processing the prediction result to generate a recommended solution. The method has the advantages of being high in interpretability, capable of adapting to the production environment in a non-invasive mode, high in expansibility and the like, automatic resolution of code conflicts can be achieved, and the working efficiency of developers for solving code merging conflicts is improved.
Owner:NANJING UNIV
A Few-Shot Spatial Transcriptome Cell Labelling System and Method Based on Graph Hints
PendingCN122090966AReduce labeling costsreduce dependenceBiostatisticsBiological modelsAlgorithmConnectivity
This invention relates to the field of spatial transcriptomics technology and provides a few-sample spatial transcriptomics cell annotation system and method based on graph cue learning, comprising: a preprocessing module; a graph construction module, which constructs spatial adjacency relationships based on cell spatial coordinates and constrains or weights graph connectivity relationships by combining inter-cell transcriptomics expression similarity; a graph representation learning module, which learns discriminative cell embedding representations from the cell graph structure under unsupervised or weakly supervised conditions; a cue fine-tuning module, which injects the category information contained in a small number of labeled samples into a pre-trained graph representation model in the downstream cell type annotation task; and a cell type prediction module, which completes cell type annotation based on a small number of labeled samples. This invention can fully integrate gene expression information and multi-source spatial relationships under conditions of very few labeled samples to construct highly discriminative cell representations, achieving high-precision automatic annotation of large-scale spatial transcriptomics data.
Owner:JILIN UNIVERSITY
A seismic data low-frequency signal inversion method and system
The present application relates to the oil and gas geophysical exploration seismic data processing technical field, specifically relates to a kind of seismic data low-frequency signal inversion method and system, method includes: obtaining initial seismic data;According to initial seismic data, determine the highest frequency that needs to be carried out low frequency recovery;Based on initial seismic data, estimated seismic wavelet is obtained;The gradient corresponding to different directions in initial seismic data is calculated, and the structure tensor is obtained according to the gradient;The structure tensor is carried out eigenvalue decomposition, and the eigenvector corresponding to the minimum eigenvalue is obtained;According to seismic wavelet, the objective function about reflection coefficient is established, the objective function is numerically solved, and the reflection coefficient is obtained;The reflection coefficient is carried out low frequency filtering with high cut frequency as the highest frequency, and the recovered low-frequency signal is obtained.The method effectively improves the quality and resolution of seismic data.
Owner:PETROCHINA CO LTD
A molecular dynamics simulation method for co2-ionized water flooding
ActiveCN120808914BImprove the development effectOptimization effect development effectGeometric CADDesign optimisation/simulationEnergy minimizationWater flooding
The application discloses a CO2-ion water flooding molecular dynamics simulation method and belongs to the technical field of CO2 oil displacement. The method comprises the following steps: establishing oil molecule models, water molecule models, CO2 molecule models, CaCl2 molecule models and graphene plate models corresponding to the characteristics of a target oil reservoir; optimizing the geometric structures of the models; assembling the optimized molecular models into a CO2-CaCl2-H2O ternary displacement system and performing molecular dynamics simulation; selecting an NVT ensemble during simulation, applying a periodic boundary condition, adopting a COMPASS II force field and configuring atomic potential parameters, running energy minimization and temperature initialization, and then performing displacement process simulation; quantitatively evaluating the oil displacement effect of the displacement medium of the CO2-CaCl2-H2O ternary system based on a trajectory file output by simulation; and the application can better explain the behavior of molecules under the synergistic action of multiple displacement media, reflect the oil displacement microscopic mechanism under the synergistic action of multiple displacement media, and has important significance for optimizing the development effect of tight oil reservoirs and improving the recovery rate.
Owner:XI'AN PETROLEUM UNIVERSITY
OVT surface element attribute extraction method and device and storage medium
PendingCN121954573AUnderstand spatial distribution characteristicsfacilitate communicationPreparing sample for investigationSeismic signal processingGeophysicsExtraction methods
The invention provides an OVT surface element attribute extraction method and device and a storage medium, and belongs to the field of seismic exploration and seismic data processing. The method comprises the following steps: acquiring seismic data; carrying out OVT surface element division on the seismic data, and obtaining transverse and longitudinal groups of each OVT surface element; calculating coordinates of the center of each OVT surface element; calculating an offset distance and an azimuth angle of each OVT surface element; and displaying the OVT surface element attribute graph. The OVT piece attribute graph is drawn and output according to the coordinates, the offset distance and the azimuth angle of the surface element center obtained through calculation, the seismic data are converted into visual graphs from abstract numbers, analysis and interpretation are facilitated, a processing result can be displayed to related personnel conveniently, and data communication and sharing are promoted.
Owner:CHINA PETROLEUM & CHEMICAL CORP +1
