Patents
Literature
60 results about "Multiple classification" 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
MULTIPLE CLASSIFICATION: "Classifying an object, creature or 'thing' in more than one dimension, such as both colour and their shape is otherwise known as multiple classification.". Related Psychology Terms.
Label-assisted report generation method and device
The invention relates to a method and a device for generating a report under the assistance of a label, and the method comprises the following steps: 1) extracting a structured label set from a text report of a sample based on a large language model, wherein the structured label set comprises a multi-classification group consisting of dichotomous labels and mutual exclusion options; 2) aggregating the labels in batches, after a threshold value is reached, merging and de-weighting, performing specification and mutual exclusion group merging on synonymous, near-synonymous and redundant labels, and converging into a unified label library; 3) based on the text report and the tag library, outputting a tag subset of each sample through a large language model; 4) multi-modal multi-label classification model training: extracting each visual modal feature, performing weighted aggregation and splicing, and outputting each label group logits through a classification head to perform weighted group loss optimization; (5) carrying out joint training on the multi-modal large language model by using samples of'only images-reports' and'images + labels-reports', and (6) carrying out label prediction and screening on the images by using the classification model, and inputting'images + prediction labels' into the multi-modal large language model to obtain a final report.
Owner:ZHEJIANG UNIV
Model training method, audio classification method, device, medium and program product
ActiveCN115309935BNeural learning methodsAudio data clustering/classificationClassification methodsNetwork model
The application provides a model training method, an audio classification method, a device, a medium and a program product, and mainly relates to machine learning technology in the field of artificial intelligence. The training method comprises the following steps: obtaining a first audio, actual classification results and actual position encoding results of the first audio in multiple classification dimensions; inputting the first audio into a target neural network model to obtain predicted classification results and predicted position encoding results of the first audio in the multiple classification dimensions; obtaining a classification loss according to the actual classification results and the predicted classification results; fusing the actual position encoding results of the first audio in the multiple classification dimensions to obtain actual fusion results, and fusing the predicted position encoding results of the first audio in the multiple classification dimensions to obtain predicted fusion results; obtaining a position encoding loss according to the actual fusion results and the predicted fusion results; and training the target neural network model according to the classification loss and the position encoding loss, so that the classification accuracy can be improved.
Owner:TENCENT TECHNOLOGY (SHENZHEN) CO LTD
Software type detection method and apparatus, terminal device, and storage medium
ActiveCN118965354BPlatform integrity maintainanceNeural learning methodsData classSoftware engineering
The application is suitable for the technical field of data processing, and provides a software type detection method and device, terminal equipment and storage medium, the method comprises the following steps: obtaining software data of a to-be-detected software; extracting software features of the to-be-detected software from the software data, wherein the software features comprise operation code features, text features, permission features and image features; classifying the to-be-detected software according to the software features to obtain multiple classification results of the to-be-detected software; determining composite software features of the to-be-detected software according to the multiple classification results; and determining the software type of the to-be-detected software according to the composite software features. The application can solve the defect that the use of a single data type leads to low detection accuracy in the related art, help to improve the accuracy and robustness of detection, and does not need to update a feature library, thus helping to reduce the maintenance cost.
Owner:EAST CHINA JIAOTONG UNIVERSITY
Product classification recognition method and system based on AI technology
ActiveCN120873855BDigital data information retrievalSemantic analysisSemantic vectorSemantic alignment
This invention discloses a product classification and recognition method and system based on AI technology, belonging to the field of AI recognition technology. By constructing a semantic alignment master vector Edis, the system achieves structured extraction and ambiguity resolution of the overall semantic representation of the product. On this basis, the system introduces the construction of a product classification label semantic vector set Gvec and a semantic similarity score set Sset, enabling the system to perform fine-grained matching of multiple classification labels based on contextual semantic features. This constructs a confidence score Conf reflecting the classification credibility, which is then compared with a preset classification judgment threshold Thrs to ensure that product labels are output only when the classification judgment credibility is sufficient. Through the connection of the above processes and the linkage of output results, this method not only significantly reduces the probability of misjudgment caused by expression ambiguity and word order reversal, but also provides a reliable rejection mechanism when accurate classification is not possible, effectively improving the overall robustness and practical value of the classification system.
Owner:ARTICLE NUMBERING CENT OF CHINA +1
Fund multi-dimensional quantitative ranking method and system
The invention relates to a fund multi-dimensional quantitative ranking method and system, and the method comprises the steps: collecting the real-time position data of a target fund, and dividing the fund types based on all ranking framework dimensions; calculating each quantitative index value of the target fund in each time window, and obtaining the similar ranking, ranking level and similar mean value of the target fund according to the quantitative index values; obtaining structured input data of the target fund based on the structured cue word, and submitting the structured input data to the large language model to generate a natural language report, the natural language report including fund analysis, risk cue and adaptation suggestion; and repeating the steps based on each ranking framework dimension, generating a natural language report of the same target fund under different fund categories, and constructing a multi-dimensional ranking system of the target fund. A fund multi-dimensional quantitative ranking system is constructed by covering a plurality of classification dimensions, a plurality of time periods and a plurality of indexes and giving appropriate comments to each index data.
Owner:CHINA CITIC BANK CO LTD
A material handling control method and system with classified goods calling function
The application discloses a kind of material handling control method and system with classified goods allocation function, it is related to the process control technical field based on big data analysis, based on high priority attribute subset, fusion path distance calculation generates optimization path sequence;On the basis of extracting path node conflict, temperature cross influence and time window overlap and other mutual restraint relations, iteratively adjust storage occupancy and handling order, get balanced order list;Finally, combined with material scheduling demand parameters and cargo classification dimension determines target handling instruction set, effectively eliminates the core contradiction of emergency material delay, temperature-sensitive material deterioration and low handling efficiency, realizes when receiving a batch of goods allocation instruction containing multiple attribute constraints, according to current specific circumstances automatically determine the relative importance of each dimension, and generate a set of reasonable handling instruction that meets multiple classification requirements simultaneously.
Owner:深圳市中卫信息技术有限公司
A method and system for mining and analyzing the correlation of clinical comorbidities of discharged patients
The application relates to a method and system for mining and analyzing the correlation of clinical comorbidity of discharged patients. The method comprises the following steps: constructing an individual diagnosis and treatment narrative graph for each patient according to the discharge medical record text, generating a time sequence transaction sequence corresponding to each patient, and constructing a sequence database; based on the database, the original time sequence frequent pattern set is constructed by analyzing through an improved generalized sequence pattern algorithm; each pattern in the set is classified to obtain multiple classification clusters, and the original time sequence frequent pattern in each classification cluster is processed through multi-sequence alignment to construct a generalized clinical path graph, and the information in the generalized clinical path graph is extracted to generate a natural language abstract. The method improves the time sequence logic, knowledge abstraction degree and clinical interpretability of the clinical comorbidity correlation mining by constructing a diagnosis and treatment narrative graph, mining a time sequence frequent pattern, and constructing a generalized clinical path graph through clustering and multi-sequence alignment.
Owner:WEST CHINA HOSPITAL SICHUAN UNIV
A ship type classification prediction method and system based on K-means and XG-Boost
The application provides a ship type classification prediction method and system based on K-means and XG-Boost, acquires ship data and carries out pretreatment, then adopts a K-means clustering algorithm to respectively cluster each kind of data in the pretreated ship data to obtain multiple clusters, calculates error sum of squares of all data in each cluster, calculates the classification number of each kind of data according to the error sum of squares, selects a certain classification number by using an elbow method, marks the ship type of all clustered ships according to the classification number, then takes the ships with marked ship types as training set samples, adopts an XG-Boost classification algorithm to train the training set samples to obtain multiple classification prediction models, verifies the multiple classification prediction models to obtain an optimal classification prediction model, and predicts the ship types of all ships in the world according to the optimal classification prediction model and marks the ship types. The application can accurately classify all ships in the world, and can avoid overfitting and underfitting of the model while ensuring the accuracy.
Owner:COSCO SHIPPING TECH CO LTD +1
Marine organism collecting and classifying device
The utility model provides a marine organism collecting and classifying device, and mainly relates to the field of organism collecting and classifying equipment. A marine organism collecting and classifying device comprises a collecting barrel. The fish collecting device has the advantages that fishes of different sizes are classified according to the sizes by arranging the collecting barrels separated by the multiple stages of classifying plates, the rotating structures and the baffles are further arranged so that all the partitions can rotate, and influences between the fishes of different sizes are prevented through separation of the baffles; furthermore, a water inlet and outlet structure is arranged in the device, so that seawater flow can be controlled from the upper part to flow into the classification device to provide a living environment for marine fishes and discharge the internal water flow, and a multi-stage classification plate is arranged to be of a spiral structure, so that the multi-stage classification plate can be matched with a baffle to rotate to drive the fishes to be moved out step by step in a multi-stage manner; mutual influence caused by simultaneous discharge of fishes with different sizes is avoided.
Owner:YANTAI VOCATIONAL COLLEGE
A waste storage box for cell culture
ActiveCN224297965UWaste collection and transferRefuse receptaclesStructural engineeringProcess engineering
The utility model discloses a kind of waste storage box for cell culture, belong to cell culture technical field, including outer frame, the inside of the outer frame is provided with storage box, the side of storage box is provided with square groove, the inside of square groove is provided with classification mechanism, the side of outer frame side and classification mechanism is adjacent and is fixedly connected with pull rod, the top of outer frame is provided with top cover, the bottom of outer frame is provided with support mechanism, the side of outer frame bottom close to support mechanism is provided with universal wheel, the top of storage box is provided with discharge chute, the utility model is provided with classification mechanism, and storage box can be inserted multiple classification boxes by square groove, hold classification box, simultaneously, the slider of classification box bottom is inserted into sliding slot, multiple classification boxes can be fixed below discharge chute, each classification box corresponds to a kind of waste, push the plug into jack, and fixed in fixed groove, classification box can be closed, prevent waste overflow, the classification storage of waste through classification box can improve the rigor of experiment.
Owner:BOPIN (SHANGHAI) BIOMEDICAL TECH CO LTD
A digital asset management method, device and equipment and storage medium
This invention discloses a method, apparatus, device, and storage medium for managing digital assets. The method includes: real-time traffic monitoring of a target database; hierarchical parsing of detected network traffic data to obtain target chain classification results matching each network traffic data; the chain classification results including multiple classification categories arranged in a progressive order; obtaining target data levels matching each target chain classification result based on the mapping relationship between the chain classification results and data levels; and classifying and grading the digital assets in the target database according to the target chain classification results and target data levels for each network traffic data. This invention solves the problem of managing tens of thousands of digital assets and achieves secure monitoring of data information on network traffic.
Owner:EVERSEC BEIJING TECH
Model training method, image classification method, device, medium and program product
PendingCN121437934ACharacter and pattern recognitionNeural learning methodsClassification methodsSample image
The embodiment of the invention provides a model training method, an image classification method, equipment, a medium and a program product, and relates to the technical field of computers, and the method comprises the steps: carrying out the classification processing of a sample image through a preset classification model, and obtaining a prediction classification result of a plurality of classification levels of the sample image; the classification model comprises a plurality of classification components, and any classification component is used for outputting a prediction classification result of one classification level of the sample image; constructing a loss function of each classification level according to the prediction classification result of the plurality of classification levels and the actual annotation classification result of the sample image in the plurality of classification levels, and determining a gradient corresponding to the loss function; adjusting the gradient corresponding to each classification hierarchy based on the similarity between the gradients corresponding to each classification hierarchy to obtain the adjusted gradient corresponding to each classification hierarchy; and updating model parameters of the classification model by using the adjusted gradient corresponding to each classification level to obtain a trained classification model. According to the method, the model performance can be improved.
Owner:BEIJING UNIV OF POSTS & TELECOMM
Regional lithology interpretation method, system, device and medium based on multi-property characteristics
The application discloses a regional lithology interpretation method, system, device and medium based on multiple physical characteristics, and the method comprises the following steps: acquiring regional logging data and geophysical data; preprocessing the regional logging data to obtain a target regional logging data set; establishing multiple classification models associated with the input of multiple physical characteristics; training and verifying the multiple classification models by using the target regional logging data set to determine the best classification model; performing multidimensional inversion on the geophysical data to obtain an inversion result; and inputting the inversion result into the best classification model for processing to obtain a lithology interpretation result. Therefore, by establishing multiple classification models associated with the input of multiple physical characteristics, and determining the model with the optimal classification effect, i.e. the best classification model, from the multiple classification models, the best classification model can be applied to the lithology interpretation of the geophysical inversion result, so that the lithology interpretation result with relatively better accuracy, efficiency and intelligence can be obtained.
Owner:CENT SOUTH UNIV
Classification method for training sample expansion according to multi-classifier recognition result
This invention discloses a classification method for expanding training samples based on multi-classifier recognition results, comprising: selecting multiple classification methods; selecting initial training samples including each category from the dataset to be classified to form an initial training sample set; classifying and recognizing the dataset data using each classification method and the training sample set to obtain the classification results of the dataset to be classified by each classification method; calculating the classification result acceptance rate for each data based on the classification results of the dataset to be classified; comparing the classification result acceptance rate with its preset threshold to obtain new training samples and expand the training sample set; determining and executing the next iteration classification based on the expanded training sample set; and taking the maximum acceptance classification result of the last iteration as the final data classification result of the dataset to be classified. This invention improves classification accuracy by iteratively verifying the results of multiple classifiers and gradually expanding the training samples.
Owner:GUANGXI ZHUANG AUTONOMOUS REGION ACAD OF AGRI SCI
Air energy storage power station information platform data management method and device, equipment and medium
ActiveCN119513379BData processing applicationsSpecial data processing applicationsData setPower station
This application relates to a data management method, apparatus, equipment, and medium for an air-based energy storage power station information platform. The method includes: acquiring a newly added dataset and current data filtering rules corresponding to the current dataset; partitioning the newly added dataset based on the current data filtering rules to obtain multiple partition data; processing each partition data independently based on a distributed computing framework to obtain classification evaluation results corresponding to each partition data; if any of the multiple classification evaluation results contain unqualified results, then modifying the current data filtering rules based on the unqualified classification evaluation results and the multiple partition data to obtain modified data filtering rules, thereby providing the user with first-target data based on the modified data filtering rules. This application dynamically modifies the current data filtering rules according to the classification evaluation results, enabling the data filtering rules to adaptively adjust and ensure the convenience, accuracy, and relevance of the filtering results.
Owner:BEIJING NORTH STAR DIGITAL REMOTE SENSING TECH CO LTD
Rock core image classification method based on multi-modal information fusion and related equipment
PendingCN121459026ABiological modelsNatural language data processingFeature extractionClassification methods
The invention discloses a core image classification method based on multi-modal information fusion and related equipment. The method comprises the following steps: acquiring a core image and general text description; wherein the general text description comprises a plurality of classification categories and text description of each classification category; inputting the core image and the general text description into a pre-trained visual language model to generate a preliminary classification result; wherein the preliminary classification result comprises the classification probability of each classification category; matching from an expert text description pool based on the preliminary classification result to obtain expert text description; and inputting the core image and the expert text description into a visual language model to generate a final classification result. By fusing image and text multi-modal information, the limitation of purely depending on visual information is reduced, the classification accuracy is improved, the pre-trained visual language model can automatically process the image and the text, the complex process of traditional manual feature extraction is avoided, the overall efficiency is improved, and the method can be widely applied to the technical field of rock core classification.
Owner:GUANGZHOU MARINE GEOLOGICAL SURVEY
Positioning method based on multi-source information fusion and electronic equipment
The invention discloses a positioning method and electronic equipment based on multi-source information fusion, and the method comprises the steps: obtaining a plurality of classification regions and the number of classification types according to channel frequency response, received signal strength and angle time delay response; obtaining a target feature according to the channel frequency response, the angle time delay response, the first feature extractor and the second feature extractor; and carrying out regression operation on the target features according to the plurality of classification regions and the number of divided categories to obtain positioning information, so that the positioning precision can be improved.
Owner:HUAWEI TECH CO LTD
Wafer defect classification method and related apparatus
This application discloses a wafer defect classification method and related apparatus. By introducing a multi-branch classification tree, decision nodes can have at least three branching scenarios, resulting in multiple classification results corresponding to multiple leaf nodes. The target classification result is then determined based on these multiple results. Since the multi-branch classification tree allows multiple leaf nodes to point to the same classification result, the same defect can obtain the same classification result through different decision paths. This allows the classification result to be verified through multi-path logic, thereby improving the fault tolerance of the defect classification process, reducing the probability of misclassification due to errors in single feature judgments, and improving the accuracy and stability of wafer defect classification results.
Owner:FEICESIKAIPU (SHANGHAI) SEMICONDUCTOR TECHNOLOGY CO LTD
Loyalty extraction machine
The present invention provides a loyalty extraction machine, wherein “quadratic multiform separation” (QMS) is modified and executed multiplicatively in an even generalized way. In each execution, the characteristic of one single membership is either enhanced or reduced. This process is performed in turn to each membership. Thus, every sample data (or element) receives multiple classification results. Then, the multiple classification results are collected and analyzed by an “eclectic classifier” to reach a final decision. The combination of the generalized QMS and the eclectic classifier therefore develops the loyalty extraction machine. Moreover, a label called “loyalty type” of the element is introduced to describe the effectiveness of membership recognition with respect to a training set.
Owner:KO-HUI MICHAEL FAN
Multi-agent based data classification and grading method, device, electronic equipment and computer program product
The application discloses a multi-agent-based data classification and grading method and device, electronic equipment and computer program product. The method comprises the following steps: obtaining data to be classified and graded; using multiple decision agents to classify and grade the data to be classified and graded respectively to obtain multiple classification and grading results; determining the voting results of the multiple decision agents according to the multiple classification and grading results, wherein the voting results represent the number of classification and grading results that are the same through the number of votes; in the case that the voting results have the same number of votes, using a coordination agent to divide the multiple decision agents into multiple target debate groups for multi-agent debate to obtain a target classification and grading result, wherein the classification and grading results obtained by the decision agents in the same target debate group are different. The application solves the technical problem of low classification and grading accuracy of the traditional data classification and grading method.
Owner:PEKING UNIV +1
Classification result verification method and device, equipment, storage medium and product
The embodiment of the invention provides a classification result verification method and device, equipment, a storage medium and a product, and relates to the technical field of artificial intelligence. The method comprises the following steps: receiving a verification request; generating a first feature matrix corresponding to the to-be-classified information and a second feature matrix corresponding to the plurality of classification results; performing fusion processing on the first feature matrix and the second feature matrix through a gated cross attention mechanism to obtain a fused feature matrix; and according to the fusion feature matrix, performing verification processing on the plurality of classification results to obtain a plurality of verification results corresponding to the plurality of classification results. According to the scheme, feature extraction and feature fusion are performed on the to-be-classified information and the classification result through the algorithm in the model to determine the matching value between the classification information and the classification result, and whether the to-be-classified information and the classification result are matched or not is determined through the matching value, so that a manual verification mode is replaced; and the verification accuracy of the classification result is improved.
Owner:INDUSTRIAL AND COMMERCIAL BANK OF CHINA
Gradual assistance optimization method, system, device, medium for fashion design
The application relates to a clothing design step-by-step auxiliary optimization method, system, device and medium. The method first introduces a knowledge graph, converts general optimization intentions into specific design categories and types by analyzing the correspondence between multiple classification tags of clothing in a clothing design drawing and multiple optimization intention keywords corresponding to optimization information, and effectively avoids design direction deviation. Secondly, a double sorting mechanism is adopted to comprehensively consider the correlation with the classification tags and the optimization intention keywords, accurately sort the design categories and the design types, intuitively display the modification content most consistent with the optimization intention, and greatly improve the clothing design decision efficiency. Finally, the step-by-step user self-optimization experience is realized by sequentially pushing the drawing optimization page, the user can independently adjust the categories, and each step is based on the previous results, ensuring the coherence of the modification. The step-by-step guidance mode reduces the design threshold, solves the problems of blindness and black box in the optimization process, and effectively improves the design quality and user experience.
Owner:ZHEJIANG FUSHENG IND CO LTD
X-ray fitting analysis method and x-ray fitting analysis system based on multi-physics variable model
PendingUS20260016290A1Using wave/particle radiation meansMaterial analysis by transmitting radiationData classComputational physics
An X-ray fitting analysis method and an X-ray fitting analysis system based on a multi-physics variable model. The X-ray fitting analysis method includes: obtaining multiple X-ray measurement signals of an inspection target by an X-ray measurement apparatus; establishing multiple fitting models according to a target architecture; performing an initial fitting procedure to generate multiple initial fitting results and multiple initial parameter ranges; performing a first co-fitting procedure to obtain multiple first fitting results that satisfy a first fitting condition; performing, according to a data type of the first fitting results, a classification procedure to generate multiple classification fitting results; counting to obtain multiple classification parameter ranges; and performing a second co-fitting procedure to obtain multiple second fitting results that satisfy a second fitting condition and respectively correspond to multiple structural parameters.
Owner:NANOSEEX INC
Encrypted traffic classification method and device, electronic equipment, computer readable storage medium and computer program product
PendingCN121125631AKey distribution for secure communicationStatistical analysisTraffic optimization
The invention provides an encrypted traffic classification method and device, electronic equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: based on acquired first encrypted traffic data, obtaining a first statistical feature corresponding to the first encrypted traffic data; performing gradient classification on the first statistical feature by using a pre-trained gradient boosting tree model to obtain a first classification result and second encrypted traffic data; the second encrypted traffic data is the first encrypted traffic data of which the classification result is not determined; extracting gradient features from the second encrypted traffic data, and dividing the second encrypted traffic data into a plurality of classification clusters; and using a traffic secondary classification discrimination model based on semi-supervised fingerprints and association clustering, and combining the classification cluster to re-classify the second encrypted traffic data to obtain a second classification result. According to the embodiment of the invention, technical support is provided for network security and traffic optimization through hierarchical gradient classification in combination with feature extraction, statistical analysis, gradient boosting tree training and complex traffic deep processing.
Owner:STATE GRID INFORMATION & TELECOMM GRP CO LTD +1
Arrayed face beauty prediction method, device and storage medium
The application discloses an array type face beauty prediction method and device and a storage medium. The method comprises the following steps: extracting multiple different scale face beauty features from a face image through multiple feature extractors; array type fusing the multiple different scale face beauty features to obtain multiple fused features; performing multiple binary classification processing on the multiple fused features through a face beauty classification network to obtain multiple classification results, wherein the face beauty classification network is obtained through supervised training of a cost-sensitive loss function, and the cost-sensitive loss function is a loss function set according to cost-sensitive training labels; and combining the multiple classification results to make a decision and obtain a face beauty prediction result. The application can strengthen information supervision through multi-scale face beauty feature extraction and array type fusion, and can effectively reduce the average cost of classification errors through optimization of the cost-sensitive loss function, thereby improving the accuracy of classification prediction.
Owner:WUYI UNIV
A modality balance training method and system for heterogeneous multimodal domain adaptation
This invention discloses a modality balancing training method and system for heterogeneous multimodal domain adaptation. The method acquires labeled source domain multimodal samples and unlabeled target domain multimodal samples. For each modality, a modality encoder and classifier are constructed to extract unimodal representations and generate multimodal fusion representations. Constrained training is performed on the modality representations based on the source domain samples. Pseudo-labels are generated for the target domain samples using the consistency among multiple classification results. Distribution alignment loss between source domain and target domain representations is calculated for each modality. Modality balancing weights are dynamically determined based on the state of each modality alignment loss during training. Model parameters are updated based on the source domain supervision loss, the target domain pseudo-label loss, and the distribution alignment loss adjusted by the balancing weights. This invention can alleviate the training imbalance problem caused by inconsistent modality offsets in heterogeneous multimodal domain adaptation, and improve the model's prediction performance and generalization ability in the target domain.
Owner:ZHEJIANG LAB
Inclined plate type grading cyclone
The utility model provides an inclined plate type grading cyclone, and belongs to the technical field of coal washing. Comprising a cylindrical section of a columnar structure, a square column section with the section of a rectangular structure is arranged above the cylindrical section, an overflow opening is formed in the top of the square column section, a funnel-shaped conical section is arranged below the cylindrical section, and an underflow opening is formed in the lowest portion of the conical section. A plurality of inclined plates with irregular structures are embedded in the square column section; each interference settlement plate is of a transversely-arranged herringbone structure, and the multiple interference settlement plates are arranged in the same direction to form a fishbone-shaped structure. Coarse particle materials can rapidly settle and return to the grading cyclone to be graded through the inclined plate, so that multiple grading is achieved, coarse particle mismatching substances are reduced, the fine screening treatment capacity is high, and other equipment is not needed for further grading.
Owner:SHENHUA SHENDONG COAL GRP +1
Merchant matching method and device
The invention discloses a merchant matching method and device, and the method comprises the steps: searching a plurality of candidate merchants related to a semantic feature word based on the semantic feature word of a target merchant, and enabling each candidate merchant in the plurality of candidate merchants and the target merchant to form a plurality of initial candidate pairs of the target merchant; performing multi-layer filtering processing on each initial candidate pair so as to screen a plurality of candidate pairs to be processed by the classification model from the plurality of initial candidate pairs according to a multi-layer filtering processing result; the text similarity features of the candidate pairs are processed through a classification model, a plurality of classification results are obtained, and the classification model is obtained through training of positive and negative samples formed by semantic feature words; and according to the plurality of classification results, a target candidate pair in the plurality of candidate pairs is determined, and a target candidate merchant in the target candidate pair is matched with the target merchant. According to the invention, differentiated processing can be carried out on the merchant matching process, the accuracy and efficiency of merchant matching can be improved, and the cost and difficulty of merchant matching can be reduced.
Owner:NETSUNION CLEARING CORP
Asthma symptom description classification method and system based on multi-task learning
The invention relates to an asthma symptom description classification method and system based on multi-task learning. The method comprises the following steps: constructing a multi-task learning module based on shared parameters; specifically, the multi-task learning module comprises a shared task and a plurality of classification tasks, and the shared task generates universal feature vectors for all the classification tasks; the asthma symptom description text is input into a multi-task learning module, and each classification task generates a classification result vector and a loss value; a total classification result vector is generated based on the classification result vector. According to the method, analysis of the same asthma symptom is subjected to multi-dimensional splitting, so that strong association information in the same symptom is effectively utilized, mode information in a classification task can help a model to better learn another task, the flexibility and configurability of a multi-task learning module are improved, and the learning efficiency is improved. Knowledge is migrated from other tasks with rich data through the shared encoder, so that better performance is obtained.
Owner:HANGZHOU HEALTH ONLINE INFORMATION TECH CO LTD
Archive classification frame
Owner:保定市中心血站