159 results about "Best matching" patented technology

The invention belongs to the technical field of cloud computing and edge computing collaboration, and particularly discloses an intelligent scheduling and optimizing method for a cloud-edge collaboration network based on reinforcement learning. By constructing the state sensing matrix and generating the action decision vector, the problem that a traditional scheduling method is insufficient in correlation analysis of multi-dimensional operation state data in a complex network environment is solved, and the comprehensive sensing capability of the operation state of the network node is improved; a dynamic mapping mechanism among the running state, the resource limitation and the task allocation strategy is established, the task allocation and resource scheduling strategy is automatically and differentially adjusted according to the real-time state of the node, and the optimal matching between the task demand and the resource supply and the dynamic balance between the performance and the efficiency are realized; through performance index monitoring and closed-loop feedback optimization, the scheduling effect is mastered in real time, continuous iterative optimization is performed on the reinforcement learning model according to objective data, and resource waste and scheduling delay are reduced.

An example device for decoding video data includes a memory configured to store video data; and a processing system comprising one or more processors implemented in circuitry, the processing system being configured to: generate a prediction block for a current block of video data using a neural network; determine an equivalent intra mode for the prediction block from a set of available intra-prediction modes using decoder-side intra mode derivation (DIMD), the equivalent intra mode representing one of the intra-prediction modes that would generate an intra-prediction block that would best match the prediction block generated using the neural network; decode a residual block for the current block of the video data based on the equivalent intra mode; and combine the prediction block with the residual block to decode the current block of the video data.

The invention relates to the field of artificial intelligence, in particular to an intelligent diagnosis system for crop diseases and insect pests and a self-evolution method of the intelligent diagnosis system. The diagnosis system extracts features from a disease and insect pest image and a description text of a user for disease and insect pests, performs multi-modal fusion, generates a diagnosis report by using the fused features, and further comprises a gating network and a LoRA adapter which is specific for various disease and insect pest tags. And the gating network selects a corresponding LoRA adapter according to the description text of the plant diseases and insect pests, the corresponding LoRA adapter is dynamically injected into a Transform layer of the basic model, a reasoning path of the LoRA adapter is temporarily reconstructed, and the model with updated parameters is utilized to generate a diagnosis text. According to the method, the most matched expert capability can be loaded in time, so that the adaptability and the diagnosis accuracy of complex and changeable diagnosis scenes are remarkably improved.

The invention provides a three-dimensional flow velocity field inversion method and system based on coherence-assisted SAR adaptive offset tracking, and a storage medium, so as to improve the problem of decoherence caused by time heterogeneity in a synthetic aperture radar interferometry (InSAR) technology, and relates to the technical field of electric power engineering geological disaster monitoring. The method comprises the following steps: establishing a space-time baseline network between SAR images; obtaining coherence data after geocoding; selecting a pair of SAR images from the space-time baseline network as a reference unit, calculating an offset result through a pixel offset tracking technology for the reference unit, and determining an optimal matching window parameter of the reference unit; establishing a mapping function between coherence and optimal matching window parameters; according to the established mapping function, calculating an optimal matching window parameter of each remaining SAR image pair; and in combination with the line-of-sight average deformation rate fields of the ascending and descending rails, inverting a surface three-dimensional flow velocity deformation field of the target area through a geometric vector decomposition model.

The invention discloses a storage goods allocation dynamic optimization method based on AI, and belongs to the technical field of intelligent storage and logistics management.The method comprises the steps that firstly, multi-dimensional data of goods, storage, business and the like are obtained; then, calculating a time-sensitive goods location-goods matching degree score integrating future delivery frequency of goods, association degree with adjacent goods, physical distance and a dynamic time urgency factor changing along with approaching of expected sorting time; a reinforcement learning model is used, the matching degree score is used as a key decision basis, a composite reward function considering overall layout gain, physical handling cost and high-urgency-degree task reward is used for guidance, and a dynamic goods allocation optimization scheme with globality and perspectiveness is output; according to the method, continuous self-adaptive adjustment of the storage layout can be realized, and the sorting efficiency and the response capability of the warehouse are remarkably improved.

The invention relates to the technical field of conveyor regulation and control, in particular to a pipe chain conveyor operation regulation and control method suitable for bulk material conveying. Obtaining time sequence data of motor working states and monitoring point running states of the pipe chain conveyor in a preset time period and a historical time period, and screening out historical reference time most matched with working conditions at all moments in the preset time period by comparing difference characteristics of the motor working states in the historical time period and the preset time period; carrying out multi-dimensional operation state characteristic comparison on the material blocking probability and each moment in a preset time period, dynamically calculating the material blocking probability, and screening material blocking moments; the material blocking probability difference of all the monitoring points at the material blocking moment and the time relation of the material blocking moment are analyzed, an abnormal transmission chain is constructed, and a material blocking propagation path is revealed; on the basis, feature vectors are constructed, misjudgment is reduced in combination with a neural network, and a refreshed abnormal transmission chain is obtained; and finally, calculating the self-adaptive regulation and control intensity at the current moment based on the refreshed abnormal transmission chain to realize closed-loop optimization of a regulation and control strategy.

The invention provides a cross-platform WebView packaging device and method based on conditional compiling. The cross-platform WebView packaging device comprises a WebView packaging layer, a platform adaptation layer and a bottom platform layer, the WebView encapsulation layer provides a unified API interface as an entry directly called by a developer, and shields the difference of a bottom platform; the platform adaptation layer detects the type of a target operating system in a compiling stage by utilizing a Qt preprocessor macro, and dynamically selects and instantiates a bottom browser view most matched with the corresponding platform according to a detection result; and the bottom platform layer comprises a browser engine implementation class exclusive to each platform. According to the method, a unified packaging layer and an API interface are established, platform differences are digested at a bottom layer, and upper-layer service codes can be written once and run everywhere.

The invention discloses a semantic tree index-based reasoning enhanced RAG retrieval method and system. The method comprises the following steps of: S1, reading and analyzing a document; s2, structuring the semantic knowledge tree by the document; s3, carrying out tree search and retrieval; and S4, enhancing the answer by the RAG system. According to the method, a long document of a specific field task is structured into an indexable semantic knowledge tree, and a large language model is guided to perform multi-step reasoning and path search on the tree in a retrieval process, so that a logic node which is most matched with query is dynamically locked, fundamental transformation from semantic similarity matching to structured logic positioning is realized, and the search efficiency is improved. The method does not need to depend on a vector database and embedded model training.

Recommending machine learning models is provided. The method comprises training machine learning models, wherein each machine learning model is trained with a unique respective dataset. Metadata associated with each machine learning model is extracted, wherein the metadata includes properties of the respective dataset used to train the machine learning model. The machine learning models and metadata are stored in a model catalog. Upon receiving a new dataset, similarity scores are calculated between the new dataset and the machine learning models in the model catalog according to the properties of the datasets in the metadata of the machine learning models. A closest match machine learning model is identified from the model catalog for the new dataset according to similarity score. Responsive to a determination that the closest match machine learning model exceeds a similarity threshold, predictions for the new dataset are generated with the closest match machine learning model.

The invention belongs to the field of control, relates to a resin production waste gas recovery control method and system, and aims to solve the technical problems that in the prior art, working condition division cannot be matched with the real state of a multivariable system, and PID parameters are balanced among a plurality of control targets. The neurons are divided into stable working condition neurons, transition working condition neurons and disturbance working condition neurons; s2, obtaining a Pareto optimal solution set composed of a PID parameter group, wherein the Pareto optimal solution set comprises an ITAE optimal solution, a control output change rate optimal solution and an equilibrium solution; s3, determining an optimal matching unit of the potential feature vector in the self-organizing mapping network; s4, when the optimal matching unit is a stable working condition neuron and the absolute value of the current control error is greater than a high-order threshold value, selecting an ITAE optimal solution; when the absolute value of the current control error is smaller than a low-level threshold value, the optimal solution of the control output change rate is selected; and a balance solution is selected between the two. Accurate identification and division of the operation state of the resin production waste gas recovery process are realized.

A real time translation method for a game includes extracting features from video frames using computer vision and a database, performing an association process to find a machine learning model best matching the features for translation, obtaining texts in the game through optical character recognition (OCR), preprocessing the texts, translating the texts using the machine learning model to generate translated texts, and rendering the translated texts to images of the video frames for displaying the images with the translated texts on a display device.

Methods, systems, and computer storage media for providing security incident prioritization management using an adaptive incident prioritization engine of a security management system are described. The adaptive incident prioritization engine provides security incident prioritization based on an adaptive incident prioritization (AIP) framework built using a ranking algorithm. In particular, the adaptive incident prioritization framework employs a Best Matching 25 (BM25) algorithm and strategically and programmatically adapts the algorithm (e.g., an adaptive incident prioritization model) to rank security incidents based on a local security incident relevance metric (an adaptation of Term Frequency—TF—in BM25) and a global security incident rarity metric (an adaptation of Inverse Document Frequency—IDF—in BM25) associated with security incidents. A prioritization score for a security incident is calculated based on aggregating weighted frequencies of security incident ranking components (e.g., security incident metadata) within a security incident to determine an overall significance of the security incident.

The invention relates to an electric power document analysis and generation method and device based on a large language model and a medium, and the method calls the large language model to execute the following steps to perform document analysis and generation: obtaining a to-be-processed electric power document, and performing structured analysis and feature extraction to obtain first structured data; screening an optimal template which is most matched with the power document from a template library on the basis of the first structured data; and reading second structured data corresponding to the optimal template, verifying the second structured data, adjusting chapter titles according to the first structured data and synchronously updating writing cue words after verification is passed, outputting a document adjustment result, performing format standardization, and generating a final templated power document. Compared with the prior art, the method has the advantages of high power document template generation efficiency, high matching precision, self-adaptive adjustment capability and the like.

The invention provides a full-scene adaptive intelligent recommendation method and system based on a graph neural network, and the method comprises the steps: constructing a parallel candidate recommendation model for generating a diversified candidate pool; a scene label of a user is determined through a CoT inference rule, the scene label is used as a query, a most matched strategy document is retrieved from a strategy knowledge base by utilizing an RAG module, and a currently recommended dynamic execution strategy is formulated; normalizing the original score of the parallel candidate recommendation model based on a dynamic execution strategy, calculating a preliminary fusion score according to a dynamic weight retrieved from a strategy knowledge base, and generating a preliminary sorting list; and performing rearrangement based on big language model reasoning on the preliminary sorting list in combination with scene and strategy guidance, and introducing evidence retrieved from a fact knowledge base based on an RAG module into a rearrangement result to obtain a final recommendation list with language interpretation. According to the method, the recommendation list which is highly personalized, self-adaptive in scene and clear in natural language interpretation can be obtained.

Disclosed herein are embodiments of systems, methods, and products comprises an analytic server, which automatically manages appointment scheduling. The analytic server receives a customer request to schedule an appointment. The analytic server determines the required data from both customer and service provider for making the appointment. The analytic server retrieves customer data comprising requested service attributes, user preferences, users attributes from internal database and external data source. The analytic server retrieves service providers' data comprising provider service attributes, providers' attributes from internal database and external data sources. The analytic server accesses external data source by web crawling various websites. The analytic server executes an artificial intelligence model to predict user preferences and needs. The analytic server determines potential service providers best matching the customer's input or predicted preferences. The analytic server generates an appointment for each matching service provider and transmits an electronic message comprising the appointments to customer device.

The invention relates to the technical field of electric data processing, in particular to a medicine logistics transportation monitoring method and system, and the method comprises the steps: collecting temperature data in a box body on a transport vehicle; the validity of the temperature data at the current moment is calculated, the validity is the bumping strength of the box body at the moment and the optimal matching degree of a temperature change sequence at the current moment in the current state and a standard temperature sequence after the box body is normally opened, and the length of the temperature change sequence at the current moment is consistent with that of the standard temperature sequence; and acquiring a temperature prediction value at the next moment based on a sliding moving average method. According to the method, the size of the moving average window is dynamically adjusted by combining the matching degree of the jolting strength and the temperature change trend, and self-adaptive smoothing processing of the temperature data is achieved. Noise can be suppressed during jolting, sensitivity can be kept during real fluctuation, and therefore prediction accuracy is improved, and the false alarm rate of abnormal detection is reduced.

The present application relates to the technical field of log analysis, and particularly relates to a kind of audit system operation and maintenance log automatic generation method and equipment.The method parses log stream to distinguish business log entries from unassociated abnormal entries, constructs thread execution sequence and calculates the idle time slice between adjacent business log entries in thread execution sequence;Load the resource constraint mapping table defining hardware resource type and minimum duration;Get the candidate business entries of unassociated abnormal entries, and perform mutual exclusivity and capacity physical verification using the resource constraint mapping table to filter out effective candidate business entries;According to the time difference between unassociated abnormal entries and its effective candidate business entries, filter out the best matching business entries, and then backtrack the business context, inject the exception stack and correct the state to generate structured audit records.The present application realizes non-invasive and high-precision exception attribution based on the rules of runtime resource constraints, and improves the integrity and accuracy of audit data.

Systems and methods for using agents are provided. A first message is received from a user to engage with a corresponding session of an automated content logic. The automated content logic is configured to facilitate electronic communication between with the user and a plurality of agents associated with the automated content logic. The first message is applied as an input to a first set of agents to receive a corresponding output for each respective agent in the first set of agents. Each corresponding output identifies an action performed by the automated content logic responsive to the first message that best matches with the first message. The corresponding output for each respective agent in the first set of agents is used to identify a first action. A process is performed in accordance with the first action to generate a second message to engage with the corresponding session.

An example device for decoding video data includes a memory configured to store video data; and a processing system comprising one or more processors implemented in circuitry, the processing system being configured to: generate a prediction block for a current block of video data using a neural network; determine an equivalent intra mode for the prediction block from a set of available intra-prediction modes using decoder-side intra mode derivation (DIMD), the equivalent intra mode representing one of the intra-prediction modes that would generate an intra-prediction block that would best match the prediction block generated using the neural network; decode a residual block for the current block of the video data based on the equivalent intra mode; and combine the prediction block with the residual block to decode the current block of the video data.

This application belongs to the field of control and relates to a control method and system for recovering waste gas from resin production. It addresses the technical problems in existing technologies, such as the inability of operating condition division to match the true state of a multivariable system and the need for PID parameters to balance multiple control objectives. The control method includes the following steps: S1, dividing neurons into stable operating condition neurons, transitional operating condition neurons, and disturbance operating condition neurons; S2, obtaining a Pareto optimal solution set composed of PID parameter groups, including the ITAE optimal solution, the control output rate of change optimal solution, and the equilibrium solution; S3, determining the best matching unit for the latent eigenvectors in a self-organizing map network; S4, when the best matching unit is a stable operating condition neuron, selecting the ITAE optimal solution when the absolute value of the current control error is greater than a high threshold; selecting the control output rate of change optimal solution when the absolute value of the current control error is less than a low threshold; and selecting the equilibrium solution when it is between the two. This achieves accurate identification and division of the operating state of the resin production waste gas recovery process.

When registering with a microservice system, each service provides a natural-language description of itself. An application programming interface (API) is used to generate a vector representation of the description, referred to as a service vector. A user searches for a service by providing a natural-language description of the service. The API is used to generate a vector representation of the provided description, referred to as a search vector. This vector representation is compared to vector representations stored in the database to identify one or more candidate services. Efficiency of the search is improved by clustering the services. A vector representation for each cluster, referred to as a cluster vector, may be generated. The search vector may be compared to the cluster vectors. Once a closest matching cluster vector is identified, the search vector may be compared to the service vectors for the services in the cluster.

The application relates to the technical field of artificial intelligence, and provides an interactive method, device, equipment, medium and program product of a customer service robot, the method comprising the following steps: determining a target operation process matched with a user demand indicated by a conversation context based on the conversation context; the conversation context is determined based on a user query to be replied and historical interaction data belonging to the same conversation as the user query; inputting process content corresponding to the conversation context and the target operation process into a node detection model to obtain a target process node to be executed output by the node detection model; the node detection model is used for determining a target process node most matched with the conversation context from a plurality of process nodes through semantic understanding capability; and generating reply information of the user query based on the target process node. The application solves the problem that a process execution is rigid and cannot adapt to user jump expression, realizes nonlinear and dynamic execution of an operation process, and further improves the interactive experience of the customer service robot.

This application relates to the field of artificial intelligence, and more particularly to a method, apparatus, device, and program product for controlling document editing. The method includes: receiving a document editing request from a large language model agent; if no exact matching sequence is found in the target file for the content to be replaced, calculating the similarity between the content to be replaced and the sequence in the target file, and locating the best matching segment based on the similarity; performing a difference analysis between the content to be replaced and the best matching segment and generating an error diagnosis report including difference information; returning the error diagnosis report to the large language model agent, guiding the agent to correct the document editing request. This method finds the best matching segment for sequences that fail to match precisely and generates an error diagnosis report including difference information, facilitating rapid location of differences, reducing invalid attempts, and improving the agent's repair efficiency.

The invention relates to a power enterprise data asset cross-department sharing intelligent cooperation method and system, and the method comprises the steps: S1, obtaining multi-source data of each business system, an Internet of Things terminal and a historical database of a power enterprise, and constructing a multi-dimensional knowledge graph; s2, semantic understanding is carried out based on an NLP large model according to data requirements of a user, the requirements are automatically converted into structured asset query conditions, key entities in the requirements are automatically analyzed and expanded in combination with a multi-dimensional knowledge graph, and standardized requirements are obtained; s3, recommending the most relevant and most matched data assets and responsible departments by adopting a graph neural network algorithm, judging whether the demand relates to cross-department and multi-data fusion or not, and automatically decomposing to generate a multi-node collaborative work order; s4, generating a deliverable data set through AI driven approval, automatic processing suggestion and sensitivity detection according to the collaborative work order; and S5, according to the user characteristics, determining an optimal data delivery mode to carry out data delivery. According to the method, the data sharing and cross-department cooperation efficiency is remarkably improved.

The invention discloses a crowd sensing task allocation method based on direction constraint, and the method is suitable for a multi-participant and multi-task mobile crowd sensing scene, and is superior to a traditional task allocation method in the aspects of task allocation efficiency and resource utilization rate. The method comprises the following steps of: establishing a task sorting list at each vertex of a sensing area, and comprehensively considering a participant path direction, a distance budget and a dynamic progress constraint; and finding out a region vertex which is most matched with the path direction of the participant, traversing a task list of the vertex, performing feasibility verification on each candidate task, and calculating a dynamic progress constraint value to obtain a final task allocation scheme. According to the method, the influence of the moving direction of the participant and the distance budget is considered, the allocation strategy is dynamically adjusted according to the real-time position and the task distribution, the task completion rate and the data acquisition quality are effectively improved, and the method has good practical guiding significance.

Systems and methods for the selection, verification and recommendation of cohort sample sets is provided. In some embodiments, a dataset selection optimization includes first receiving at data stewards classes of data required by the data consumer. The data stewards process their data (or a subset of their data) into a vector set within a sequestered computing node. These vector sets are transferred to a core management system for minimizing a difference between a target vector and any combination of the data stewards' vector sets. A cost function may also be applied to the vector sets during this optimization. Once the data steward(s) that best match the target vector are identified, they may be placed in contact with the data consumer for access of their information.

The application discloses a CAD drawing equipment type identification method and system based on an embedded model, relates to the technical field of equipment type identification, and comprises the following steps: constructing a CAD equipment dataset; constructing a general embedded model comprising a feature extraction structure and an auxiliary head structure, and training the general embedded model by using the CAD equipment dataset; screening various types of equipment image templates to construct a CAD equipment template library, embedding all images in the template library into a vector set by using the trained general embedded model, and importing the vector set into a vector database; inputting a CAD equipment image to be identified into the trained general embedded model to obtain a corresponding embedded vector; calculating the similarity between the embedded vector of the CAD equipment image to be identified and the vector set in the vector database, and determining the most matched equipment type. The application solves the problem of obvious defects of the CAD drawing equipment type identification method in the prior art in terms of scalability, small sample adaptability and data cost control.

The invention provides a safety behavior regulation matching method and system based on a large model and a knowledge graph, and relates to the technical field of safety intelligent management, and the method comprises the steps: enabling a multi-modal large model to receive a field image, and generating a behavior description and a candidate regulation direction; the multi-modal large model extracts visual features from a field image based on a thinking chain template, retrieves a multi-modal knowledge graph based on the visual features and behavior description, and screens a candidate term range based on the visual features and candidate term directions; the multi-modal large model screens candidate regulations from a relational database on the basis of the direction of the candidate regulations, converts the candidate regulations into multi-modal semantic vectors on the basis of visual features and behavior description, and screens the candidate regulations from a vector database; the multi-modal large model is based on a multi-modal knowledge graph, and the most matched security behavior regulation and image-text-regulation cross-modal interpretation are output. The method breaks through the limitation of single-mode recognition, improves the recognition precision of complex scenes, and achieves the cross-mode recognition of safety behaviors and the precise matching of regulations.

The application discloses a dynamic object moving distance measurement method and system based on a laser point cloud, a terminal and a medium, and relates to the technical field of point cloud analysis. The method comprises the following steps: acquiring current frame point cloud data of a moving target, registering the current frame point cloud data with a target point cloud template sequence to obtain a best matching template; moving the best matching template to a target position corresponding to the current frame point cloud data to obtain a reference point of the current frame point cloud data; judging whether the current frame point cloud data is end frame point cloud data, and if the current frame point cloud data is end frame point cloud data, determining the moving distance of the moving target. The application realizes coarse registration by finding the best matching template, and then realizes fine registration by moving the best matching template to the target position corresponding to the current frame point cloud data, so that the moving target can be accurately positioned, the moving distance of the dynamic object can be accurately measured, and the application has high robustness.

The application discloses a kind of intelligent matching conversion methods and systems from HTML to PPTX document, method includes: obtaining HTML document and carrying out parsing, determine hierarchy, style and semantic structure, and extract element;Extract the visual features, structural features, semantic features and functional features of element;Classify elements according to categories using convolutional neural networks, and use a matching model based on the Transformer architecture to perform similarity matching to obtain the best matching result;Convert the element to a PPTX element, layout the PPTX element, and convert the PPTX element to an animation effect based on the functional features of the element;Render and standardize the converted and laid out PPTX elements to generate a PPTX document. Through the technical solution of the present application, the element matching accuracy is significantly improved, the layout conversion quality is greatly improved, the complete style information is preserved, and the conversion efficiency is significantly improved.