39 results about "Category attribute" patented technology

One drawback of generative language models, e.g. large language models (LLMs), is their tendency to “hallucinate”. To address at least this technical problem of hallucination, in some examples, an input may be classified to determine a category associated with the input. In some examples, the input may be an image. A grammar may be obtained based on the determined category. The grammar may define valid sequences of symbols describing attributes of the category. A generative language model may be used to generate a sequence of symbols describing one or more attributes associated with the input. The sequence may be based on the input and conform to the grammar. In some examples, the generative language model may be an LLM.

The invention relates to the technical field of user behavior prediction and marketing strategy optimization, and discloses a cross-platform user behavior prediction and marketing strategy generation system and method, and the system comprises a heterogeneous data alignment module, a three-dimensional behavior prediction network and a collaborative marketing strategy generator. Intelligent separation and reconstruction of three types of features of user interests, category attributes and interactive behaviors are realized through a multi-modal feature decoupling unit; a three-dimensional cross-correlation model is constructed through an adaptive cross attention fusion unit, and dynamic feature fusion is realized; a probability distribution calibration unit uses a multi-task expert network and a dynamic calibration mechanism to output high-precision three-dimensional behavior probability distribution, so that the technical problems of feature entanglement, complex correlation modeling, prediction uncertainty quantization and the like in cross-platform user behavior prediction are solved, and accurate user behavior prediction and intelligent marketing strategy generation are realized; the method has important application value in the fields of e-commerce, advertisement putting and the like.

For at least a selected class attribute of the multiple class attributes, one or more bias metrics are determined that estimate a degree to which a particular workflow (having a set of processing stages) is biased in association with the class attribute. Each user of a set of users is associated with a set of user data to be processed by the particular workflow. At least one of the set of processing stages includes executing a machine-learning model. It can be detected that a bias-mitigation option corresponding to a specific class attribute has been selected. For each of at least two of the set of processing stages: a de-biasing technique is selected; and the processing stage is modified by applying the de-biasing technique. A modified version of the particular workflow (which includes the modified processing stages) is applied to each of a set of input data sets.

The invention discloses an object recognition method and device, a storage medium and electronic equipment. The method comprises the following steps: determining target category attribute description information of a target object from a first object interaction platform; obtaining candidate category attribute description information of candidate objects collected from a second object interaction platform according to the target category attribute description information; performing description conversion on the candidate category attribute description information according to a first category attribute description tree to obtain converted reference category attribute description information; comparing a reference category attribute feature generated based on the reference category attribute description information with a target category attribute feature generated based on the target category attribute description information to obtain a feature similarity; and identifying the object of which the feature similarity indicated in the comparison result meets a threshold condition as a similar object of the target object. According to the object recognition method and device, the technical problem that the accuracy of an object recognition result is low in an object recognition method provided by the related technology is solved.

Embodiments of the present application disclose a decision tree model generation method and a data recommendation method based on the decision tree model. The method comprises: obtaining a training data set formed by feature information of a plurality of training samples, the training samples having known category attributes; in a process of generating a decision tree model according to the training data set, iteratively calculating information gain of each feature attribute under each node, and dividing a data set contained by a current node according to a feature attribute corresponding to maximum information gain until a category attribute can be determined according to the data set contained by the node; if information gains of a plurality of feature attributes under the current node are equal and are maximum information gain, then calculating respective correction information gains of the plurality of feature attributes, and determining a feature attribute for dividing the data set contained by the current node according to the calculated correction information gains; and outputting the decision tree model formed according to the training data set. The decision tree model generated by the present application has more accurate classification effect.

This disclosure provides a data processing method and training method, a data processing device, a training device, an electronic device, and a computer-readable medium. The data processing method includes: acquiring input features of input data corresponding to a target network layer of a target neural network; inputting the input features corresponding to the target network layer into a prediction model to predict the category attribute corresponding to the target network layer, wherein the category attribute characterizes whether the target network layer is suitable for quantization processing of the input data. According to the technical solution of this disclosure, the quantization scheme of the network layer of the target neural network can be determined efficiently and accurately, which is beneficial for the quantized target neural network to ensure that it simultaneously possesses fast inference speed and high inference accuracy during inference operations.

The application discloses an agent-based sensitive data grading method and device, equipment and a storage medium, and relates to the technical field of information security. The method comprises the following steps: performing quality filtering on original data to be graded to obtain candidate sensitive data; pre-training and fine-tuning an original large language model through an industry corpus dataset to obtain a data security large model comprising a data classification agent and a data grading agent; calling the data classification agent to classify the candidate sensitive data based on a classification grading standard to obtain data category attributes; and calling the data grading agent to grade the candidate sensitive data according to the data category attributes and the classification grading standard to generate a classification grading list. The original large language model is pre-trained and fine-tuned through the industry corpus dataset, which provides the agents with the ability of natural language understanding and generation, and can adapt to sensitive data recognition in different scenarios, thereby improving the recognition accuracy.

System (100) for the autonomous reconciliation of product hierarchies in distributed retail systems, comprising: (a) a data entry module configured to collect product hierarchy data from a variety of heterogeneous retail data sources; b) a hierarchy extraction and structuring module configured to extract, normalise and standardise hierarchical relationships and category attributes from the input data; c) an AI-based hierarchy matching module configured to detect similarities, inconsistencies, and conflicts between the hierarchical structures using machine learning and semantic analysis; d) an autonomous reconciliation engine configured to resolve the identified conflicts and map disparate hierarchies into a unified structure using adaptive rules and learned mappings; (e) a feedback and governance module configured to log decisions, enable audit trails, and provide exception handling with optional manual overrides; (f) a unified hierarchy repository configured to store the agreed product hierarchy and make it accessible through APIs and user interfaces; and (g) a monitoring and adaptation module configured to continuously track changes across source systems and improve reconciliation performance over time; h) the system operates without manual intervention to achieve scalable and consistent product hierarchy reconciliation across distributed retail platforms in real time.

The invention discloses a recommended article style discrimination method and device, equipment and a storage medium, and belongs to the technical field of computers and Internet. Comprising the steps that attribute information of a first recommended object is acquired, and the attribute information of the first recommended object comprises a plurality of object attributes of the first recommended object; according to a category attribute rule and the attribute information of the first recommended object, identification information of the first recommended object is determined, the category attribute rule is a rule for determining the same type of recommended object under a first category to which the first recommended object belongs, and the identification information of the first recommended object is used for representing the first recommended object under the first category; according to the identification information of the first recommended object and the identification information of the second recommended object, a style judgment result is determined, the determination mode of the identification information of the second recommended object is the same as the determination mode of the identification information of the first recommended object, and the style judgment result is used for representing the style similarity between the first recommended object and the second recommended object. According to the scheme, the adaptability of the method for discriminating the style of the recommended object to different scenes can be improved.

The invention provides an association rule analysis method and device for multi-valued attribute data, and is applied to the technical field of operation and maintenance systems.The method comprises the steps that original log data is obtained; based on the original log data, attribute types are identified, and the attribute types comprise numeric attributes, interval attributes and category attributes; based on the identification result of the attribute type, performing conversion processing on the numeric type attribute, the interval type attribute and the category type attribute to obtain a numeric type feature vector, an interval type feature vector and a category type feature vector; performing feature fusion based on the numerical feature vector, the interval feature vector and the category feature vector to obtain a unified feature matrix; performing rule extraction and pruning based on the unified feature matrix to obtain an association rule; and on the basis of the association rules, a rule knowledge base is constructed, and the rule knowledge base adopts a hierarchical storage architecture. According to the invention, the data processing efficiency can be improved while the resource consumption is reduced.

The invention discloses a continuous generalized zero sample learning method and system based on a generation method. The method comprises the following steps: acquiring category attributes and Gaussian noise; based on the category attribute and the Gaussian noise, a generator and a discriminator are combined, and pseudo features and identifier projections are obtained; the pseudo features and the identifier projection are input to a zero sample learning model, a learning result is obtained, and the zero sample learning model comprises a diversified attribute enhancement module and a cross-task feature distillation module; the diversified attribute enhancement module is used for enhancing the expression ability of unseen category features by introducing diversified semantic descriptions generated by a large language model; and the cross-task feature distillation module is used for aligning the feature space of the old task and the feature space of the new task, so that the model keeps existing knowledge while adapting to the new task. According to the method, the problems of insufficient diversity and disastrous forgetting existing in unseen category generation features can be solved.

The application discloses a debiasing method based on local search, an electronic device and a medium, and the method comprises the following steps: acquiring original data, marking a category attribute in the original data, and obtaining a marked data set, which is recorded as a data set X; performing local search-based clustering analysis in the data set X selected in step (1), using P to represent a point set of clustering, and for a point x in the point set P with a size of n, r(x) is a radius, so that at least n / k points from P are in a ball with x as the center and r(x) as the radius. According to the method, sample points in different cluster groups in the data set are as different as possible, that is, the distance in the class is as small as possible, and the distance between classes is as large as possible. The distribution of sensitive attributes of each cluster group is as uniform as possible, and the fairness of the data set is improved.

The invention discloses a picture classification method and device and a storage medium, and relates to the technical field of image processing. According to the method, the picture text is extracted through the OCR and the sentences are divided, the sentences are converted into the vectors by utilizing the pre-training vector generation model, the core semantic information associated with the category attributes in the text is fully mined by virtue of the semantic capture capability of the pre-training model, and the sentence text weight is determined by calculating the word frequency and the inverse document frequency after word segmentation. According to the method, the sentences corresponding to the important semantics are made to obtain higher weights, then the weights and the vector sequence are subjected to weighted average to generate the document vectors, the document vectors are made to accurately focus on the core semantics, the effectiveness of text representation is improved, and finally accurate classification is achieved through the classification model.

The invention discloses a commodity recommendation method, system and device of an aspect category enhancement architecture and a medium, and relates to the field of commodity recommendation. The method comprises the following steps: constructing an aspect graph based on a user comment data set, and according to explicit features and implicit preference features of the aspect graph, constructing an aspect category enhancement architecture; extracting a user aspect term subset and a commodity aspect term subset from the user comment data set; based on a Transform architecture and a factorization machine, combining category attributes of aspect term nodes in the aspect graph to construct an aspect category enhanced recommendation model; using the user-commodity pair set as a label, and inputting the user aspect term subset and the commodity aspect term subset into an aspect category enhanced recommendation model for training to obtain a user commodity recommendation model; and the user commodity recommendation module is used for outputting corresponding recommended commodities when the user is received. According to the invention, more accurate commodity recommendation is realized.

The disclosure provides a text recognition method and a model training method and device, relating to image processing, deep learning and natural language understanding in the artificial intelligence technology. The specific implementation scheme is: performing optical character recognition on the obtained to-be-recognized image to obtain initial text of the to-be-recognized image, performing analysis and processing on the initial text to obtain a category attribute of the initial text, if the category attribute of the initial text represents that the initial text is an incorrect text, performing error correction processing on the incorrect text to obtain correct text for correcting the incorrect text, and generating text content of the to-be-recognized image according to the initial text and the correct text, thereby avoiding the drawbacks of text errors caused by the OCR recognition technology, and improving the accuracy and reliability of text recognition.

The invention discloses a travel trajectory data generation method and device, a medium and equipment, and relates to the technical field of information security. The method comprises the following steps: constructing a data set based on collected travel track data containing geographic coordinates, time and travel track category attributes; constructing a generative adversarial network, and training the generative adversarial network by minimizing the space similarity, the time similarity and the category similarity between the generated travel trajectory data output by a generator in the generative adversarial network and the real travel trajectory data in the data set; and generating new travel trajectory data by using a generator of the trained generative adversarial network. The training mode forces the generator to learn spatial features and time features of real travel trajectory data and category features of different categories of real travel trajectory data, and the multi-dimensional feature learning enables the trained generator to generate travel trajectory data with more authenticity in multiple dimensions. And the generated travel track data is more deceptive.

The invention relates to the technical field of remote sensing image intelligent processing and analysis, and discloses a zero-sample hyperspectral remote sensing image classification method based on knowledge graph semantic representation, and the method comprises the steps: firstly constructing a remote sensing domain knowledge graph which comprises the types, attributes and relationships of ground features, and obtaining structured semantic features through knowledge graph representation learning; meanwhile, visual features of the hyperspectral image are extracted by using a convolutional neural network. And performing cross-modal distribution alignment on the visual features and the semantic features in the shared hidden layer space through a dual variational auto-encoder. And finally, taking the hidden layer semantic features corresponding to the unknown category in the knowledge graph as category prototypes, generating enhanced features and training an unknown category classifier, and realizing effective identification of the unknown ground feature category under the condition that only known category labeling samples exist. According to the method, the dependence on a large amount of labeled data is remarkably reduced, and the classification precision and generalization ability in a zero sample scene are improved.

This invention provides a zero-shot method for identifying space targets using attribute vectors. The method involves acquiring ISAR images of the target from radar equipment; generating a category attribute vector for space targets belonging to the same class; using a trained attribute model network to identify the target in the ISAR image, obtaining a predicted attribute vector; and comparing the predicted attribute vector with the category attribute vector to determine the category of the space target in the ISAR image. Because this invention utilizes attribute knowledge from training data and sample data information, it can predict the attributes of the target in the ISAR image without requiring an ISAR image sequence, thus having wide applicability. Furthermore, this invention leverages the attribute knowledge of space target categories to identify space target categories that lack training samples, overcoming the problem of existing network-based space target identification methods requiring a large number of training samples for each target class.

The present application is suitable for the field of geographic information processing and spatial data mining technology, and provides a kind of geographic feature causal mining method considering topological neighborhood, comprising the following steps: obtaining the data of multiple types of geographic features in the target area, and pre-processing, obtaining the geographic feature set containing spatial position coordinates and category attributes;Based on the geographic feature set, the target area is divided into research units, and a plurality of basic research units are obtained, and based on the category attribute of the geographic feature, the spatial distribution of different category geographic features in each basic research unit is counted, and the distribution characteristic value is obtained.The quantitative identification of the causal action direction and the causal action intensity between geographic features can be realized;The structural connection between geographic features can be effectively described, the influence of pseudo correlation is reduced, and the accuracy and stability of causal relationship identification are improved.

The invention discloses an intelligent detection method for vehicles in a tunnel, and specifically belongs to the technical field of intelligent traffic safety monitoring. The method comprises the following steps: performing target detection and multi-target tracking on a real-time video stream collected by a camera in a tunnel based on an ROI region, and determining a target category attribute list; based on the ROI region, carrying out screening and coordinate conversion on the target category attribute list to obtain an ROI category plane list; performing parking state identification on a target category in the ROI category plane list, determining target vehicles in a parking state, obtaining a first parking list, and calculating a first parking occupancy rate of each lane; normal parking vehicles are removed, and a second parking list is determined; and carrying out accident parking identification on the target category in the second parking list, determining a vehicle in an accident state, and carrying out accident alarm. According to the method, the slight traffic accident can be accurately defined, misjudgment is reduced, the calculation complexity is low, and vehicle accident alarming is carried out in time.

The embodiment of the invention discloses an intelligent braking method, device and equipment applied to a vehicle and a medium. The method comprises the following steps: collecting target data in a visual field range; inputting the target data into a pre-trained target classification model, so that the target classification model outputs a preset category attribute corresponding to each category and an output result confidence coefficient; determining a sample matching attribute of the target data and each sample image in a sample library; according to the category attribute, the output result confidence coefficient and the sample matching attribute, determining whether the target data comprises a target object or not; and if the target object exists in the target data, the vehicle is braked. Through the technical scheme of the embodiment of the invention, whether the target object exists in the field of view in the vehicle driving process or not can be accurately and conveniently identified, vehicle braking processing is effectively carried out on the basis of accurately identifying that the target object exists, and the identification precision of the target object and the accuracy of vehicle braking are improved.

The invention discloses a risk processing method and device, electronic equipment and a storage medium. The method comprises the steps of obtaining hardware detection data of an IT system; performing feature processing on the hardware detection data based on the plurality of category attributes to which the hardware detection data belongs to obtain a classification embedding vector corresponding to the hardware detection data; generating a tense data graph of the hardware detection data based on the classified embedded vector, wherein the tense data graph is used for indicating time sequence information of the hardware detection data; performing risk prediction based on the classification embedding vector and the tense data graph to obtain a risk prediction result of the IT system; and root cause analysis is carried out on the risk prediction result based on a target knowledge graph to obtain a target root cause node corresponding to the risk prediction result, and the target knowledge graph is used for indicating multiple devices in the IT system and an association relationship among the multiple devices. Therefore, full-life-cycle risk management and control of the IT system can be effectively realized, and the risk prediction capability and the operation and maintenance efficiency of the IT system are improved.

The invention belongs to the technical field of image information processing, and discloses a cross-domain few-sample remote sensing target classification method based on discriminant text prompt learning. Firstly, a support set and a query set are sent into an image encoder to obtain an encoded feature sum; through a semantic prompt generation module and a support set prototype feature guide prompt generator, a discriminative semantic prompt S adaptive to a few-sample task is generated, and a fixed template is replaced to accurately represent category attributes; then processing a discriminative semantic prompt S through a text encoder, performing intra-class contrast regularization, constructing a same-class multi-view prompt pair and a different-class negative sample pair, and adopting temperature modulation contrast loss to suppress instance noise and improve cross-domain robustness; and finally, bimodal decision hybrid prediction is provided, and the defect that domain invariance and detail discrimination under few samples cannot be considered in a conventional method is overcome through vision-text semantic similarity and vision-vision feature matching degree.

The embodiment of the present application provides a text identification method, device, equipment and storage medium based on deep learning, which relates to the field of artificial intelligence and cloud technology. The method includes: obtaining a target text to be identified, the target text to be identified includes a text title and text content; calling a pre-trained text classification model to perform category attribute identification processing on the target text to be identified, so as to obtain the category attribute to which the target text to be identified belongs, including: performing word and sentence parsing processing on the text title and text content of the target text to be identified, so as to obtain each character of the target text to be identified; performing word vector conversion processing on each character of the target text to be identified, so as to obtain the word vector of each character; performing fusion feature extraction on the word vector of each character to obtain the text vector of the target text to be identified, so as to obtain the category attribute to which the target text to be identified belongs based on the text vector. This can enable the model to better understand the target text to be identified and improve the classification accuracy of the model.

The invention discloses a geographic entity construction method based on a framing topographic map. The method comprises the following steps: performing quality inspection on original topographic map data, removing data with unqualified data quality, unifying a data space coordinate system and an elevation reference, executing a storage operation, and integrating multiple pieces of standard framing stored topographic map data into an SQLite database; designing all data layers and attribute fields of the geographic entity, establishing a conversion rule comparison table from the original topographic map elements to the target geographic entity, and performing data category conversion; analyzing attribute information of geographic entity categories, extracting data from the SQLite database according to an entity attribute similarity principle, and obtaining a data set of line elements and surface elements with the same basic attributes; aiming at the line element set and the surface element data set, carrying out spatial data fusion by setting tolerance parameters, and combining lines or surfaces with similar attributes and adjacent spaces into a complete geographic entity; and performing element integrity, logic consistency, topology error check, spatial precision and attribute check on the constructed geographic entity data. According to the method, the overlapping and gap problems among the vector data are intelligently processed based on the attribute consistency and the spatial proximity, the geographic entity production efficiency is improved, and the labor cost is reduced.

The invention discloses a disease control big data analysis method and system, and the method comprises the following steps: obtaining a multi-source heterogeneous disease control data set, carrying out the preprocessing of the multi-source heterogeneous disease control data set, carrying out the independent optimal sequence search of each class attribute in a cluster to which the class attribute belongs, obtaining the optimal sequence of the class attribute, and obtaining the optimal sequence of the class attribute; determining a distance structure suitable for disease control big data analysis based on the optimal sequence; according to the obtained optimal sequence, distance measurement between the samples and the clusters is calculated, and a clustering model is constructed; alternately updating a sample attribution matrix and a category attribute sequence by minimizing a target function, and dynamically adjusting a clustering result; and outputting a case group, a propagation chain identification result and a risk assessment report based on a clustering result. According to the method, the clustering result is adaptively adjusted through an attribute sequence learning and sample clustering optimization conjoint analysis method, and the problem that an epidemic prevention and control decision is seriously affected due to the fact that the clustering result in the traditional technology cannot reflect business logic such as a real propagation chain and risk population layering is solved.

The present application relates to the technical field of smart grid, and discloses a large-scale power grid high-order tensor network construction method, device, equipment and medium, comprising: acquiring a plurality of category attribute sets of heterogeneous nodes in a large-scale power grid; performing feature extraction on the plurality of category attribute sets to obtain attribute features of the heterogeneous nodes; establishing multiple feature subspaces according to the attribute features of the heterogeneous nodes by using a deep hash mapping model; aligning the multiple feature subspaces to a unified feature dimension by using a breadth learning strategy; calculating distance measurement values between the heterogeneous nodes according to attribute feature vectors of the unified feature dimension by using a distance measurement method; determining a base tensor for representing an interaction relationship of the heterogeneous nodes according to the attribute feature vectors of the unified feature dimension and a weight value of the interaction relationship of the heterogeneous nodes; and constructing a large-scale power grid high-order tensor network according to the base tensor and the weight value of the interaction relationship of the heterogeneous nodes by using a tensor multiplication operation rule. The problem of lacking a model for efficiently representing a large-scale power grid is solved.

The application belongs to the technical field of image information processing, and discloses a cross-domain few-shot remote sensing target classification method based on discriminative text prompt learning. First, the support set and the query set are sent into an image encoder to obtain encoded features. Through a semantic prompt generation module, a support set prototype feature guides a prompt generator to generate discriminative semantic prompts S that adapt to the few-shot task, and replace the fixed template to accurately represent the category attributes. Then, the text encoder processes the discriminative semantic prompts S, the intra-class contrast regularization is used to construct the same-class multi-view prompt pair and the different-class negative sample pair, the temperature modulation contrast loss is used to suppress instance noise and improve cross-domain robustness. Finally, the application proposes a bimodal decision hybrid prediction, which overcomes the problem that the previous method cannot balance the domain invariance and the detail discrimination under few-shot.

The application relates to the field of data processing, in particular to a data processing method and device, electronic equipment and storage medium, which are used for accurately obtaining associated data of objects in a delivery platform. The method comprises the following steps: clustering a plurality of to-be-analyzed objects based on an object category attribute and a preset number of a to-be-analyzed object, and obtaining at least one object cluster; respectively sending a corresponding acquisition request to the delivery platform for each object cluster, wherein each acquisition request is used for acquiring associated data of each to-be-analyzed object in the corresponding object cluster; receiving each feedback message returned by the delivery platform, wherein, when one feedback message representing acquisition failure is received, re-clustering the corresponding object cluster based on a comparison object number carried in the received feedback message, obtaining an updated object cluster, and sending an acquisition request to the delivery platform again based on the updated object cluster until the corresponding associated data is obtained. In this way, the associated data of the objects in the delivery platform can be accurately obtained.

This invention provides a visual language model image classification method based on attribute anchor cue word fine-tuning, relating to the fields of computer vision, deep learning, and visual language model technology. The invention generates attribute text based on the image dataset to be identified and constructs a training sample set; it constructs hybrid cue words, inputs the hybrid cue words and the image dataset to be identified into a visual language model, and identifies the category to which each image in the image dataset belongs. Under the condition of freezing the pre-trained visual language model, this invention utilizes an efficient parameter fine-tuning method for category attribute-enhanced text cueing to solve the problems of unclear semantics and weak generalization ability in existing cue learning, ensuring that attribute information is not overwhelmed by category names and improving image recognition performance on both base and new classes in multi-dataset scenarios.