56 results about "Node splitting" patented technology

The invention relates to the technical field of artificial intelligence, in particular to a big data information analysis method and system based on artificial intelligence, and the method comprises the following steps: obtaining a jump segment direction vector, classifying label nodes with consistent directions, splitting a path lacking a mapping relation to form an interruption segment, recognizing a node with continuous directions as a bridge segment entrance, and obtaining a bridge segment direction vector; according to the method, by means of classification according to the node sequence and the channel number in the segment hopping direction, direction aggregation boundary determination and semantic field combination path position interruption recognition, the node sequence structure positioning precision and the paragraph division reasonability are improved, and the focus analysis set is generated by combining the path segments. Bridge segment connection is screened based on direction consistency, path continuation logic, behavior tracks and aggregation numbers are kept in sequence and hierarchy regularization, a behavior aggregation sequence of a semantic chain is constructed, and the continuity of path recognition in a channel aggregation area and the stability of node attribution are enhanced.

The invention belongs to the technical field of dynamic network risk prediction based on knowledge graph driving, and discloses a dynamic network risk prediction method and system based on knowledge graph driving, and the method comprises the steps: firstly obtaining a network security event and a context to construct a knowledge graph, and then detecting entity semantic drift through a preset rule and a deep semantic model, and carrying out node splitting, fusion, renaming or label updating and other structural remodeling on the knowledge graph based on a detection result, and finally carrying out risk prediction by utilizing a graph neural network model. The problem that in a traditional method, a static model cannot adapt to dynamic semantic changes is solved, the recognition capacity of a novel attack mode is improved, the risk detection false report and missing report rate is reduced, and the risk prediction efficiency is improved.

The invention discloses an online incremental learning method and system based on a self-adaptive B + tree index, and relates to the technical field of artificial intelligence, and the method comprises the steps: constructing an extended self-adaptive B + tree index structure, inserting a new sample, and calculating a trajectory disturbance value; generating a disturbance score by combining the disturbance value and a node state, and adding a new sample and an associated node sample into an incremental learning scheduling queue when a condition is met; when a node splitting, merging or reconstructing event occurs, acquiring an affected node and updating a structural code of the affected node; monitoring sample label track change, adjusting a label confidence mark when a threshold value is exceeded, and limiting participation frequency according to a screening mechanism; and performing forgetting compression operation on the leaf nodes with low access frequency and small label fluctuation degree. By constructing an adaptive B + tree index and combining disturbance-driven scheduling, structure change response and forgetting compression strategies, sample management refinement, learning update high efficiency and structure maintenance controllability are realized, and the stability and resource utilization efficiency of online incremental learning are improved.

The invention discloses a storage and calculation separation key value storage delay optimization method based on prediction driving, and aims to solve the problems of performance bottleneck and high tail delay caused by passive management and high network delay in a key value storage system in a storage and calculation separation scene. The method comprises the following steps: time load prediction: deploying a time sequence prediction model at a client, and predicting a future read-write request based on a historical access sequence; active cache prefetching: according to the predicted read request, actively preloading hotspot data from a server side to a client side for caching so as to improve the cache hit rate and hide network delay; active write-in optimization: according to the predicted write request, executing maintenance at a server side through predictive pre-insertion and active node splitting, and moving high index structure adjustment overhead out of a key request path to eliminate a write delay peak; and structure sensing batch synchronization: pre-fragmenting a local write buffer by using a server index top layer model of a client cache, and combining multiple independent remote insertion operations into one efficient batch update to reduce data synchronization overhead. Compared with an existing passive management system, the characteristics of model prediction and active cooperation are fully utilized, and the average delay and the tail delay of the system are reduced.

The invention provides a water turbine bearing bush temperature prediction model driven by feature importance, relates to the field of energy and power engineering, and provides a feature importance module to quantify the influence of each input feature on a prediction target by analyzing the reduction degree of a decision tree on a mean square error in a node splitting process so as to predict the temperature of a bearing bush of a water turbine. Therefore, effective evaluation of feature importance is realized. The extracted feature importance is converted into learnable weighted parameters, and the learnable weighted parameters are fused into input features of the trunk model to enhance expression of key features and suppress interference of redundant features, and the expression ability and prediction precision of the weighted parameters in a water turbine bearing bush temperature prediction task are enhanced. Experimental results show that the prediction precision of the FETSM in multiple time steps is obviously superior to that of the most advanced model, the FETSM has good performance in two evaluation indexes of mean square error and mean absolute error, and the effectiveness and robustness of the model in water turbine bearing bush temperature prediction are verified.

The invention discloses a valve-side dry-type bushing end screen diagnosis and evaluation method, system, medium and equipment, and the method comprises the steps: data collection and preprocessing: obtaining multi-source data of a converter transformer valve-side dry-type bushing through real-time monitoring or historical recording of a sensor; dividing the data set, and dividing the multi-source data into a training set and a test set according to a proportion; building a random forest model, generating training subsets of a plurality of decision trees from the training set by adopting replacement random sampling, randomly selecting a feature subset for each decision tree, performing node splitting based on an information gain ratio, and generating the decision trees in a depth-first mode until a preset maximum depth is reached; performing model integration and diagnosis, adopting a voting mechanism to integrate classification results of all decision trees, and outputting defect types and state evaluation grades, the state evaluation grades being grade II and grade I; and performing dynamic tuning and deployment, performing model parameter optimization by taking a test set macro average F1 score greater than or equal to 0.9 as a threshold value, and performing real-time diagnosis on the optimized model.

The application discloses a distributed storage index method and system based on multiple hash, and the method is as follows: based on a key-value storage engine, a Master / Slave architecture is adopted to build a distributed storage index system; the Master / Slave architecture comprises one Master node and multiple Slave nodes; according to an access request of a key-value pair, the Master node calls multiple linear hash functions to calculate the key of the key-value pair, and obtains the Slave nodes corresponding to all hash values; the Master node sends corresponding operation requests to the corresponding Slave nodes, including an increase operation, a deletion operation, a modification operation and a search operation; when the Master node detects that the real-time load rate of a certain Slave node exceeds a load critical point triggering node splitting, a node splitting operation is performed. The method utilizes the automatic expansion feature of linear hash, improves the scalability and resource utilization rate; based on multiple hash functions, multiple data backups are realized, data loss is avoided, and the reliability is improved; access and storage are realized by using nodes with low load rates, and load balancing is realized.

The invention provides a distributed computing method and system and related equipment. The method comprises the steps that a management node issues a fusion operator to a computing node; the fusion operator comprises two matrixes and an instruction for realizing matrix multiplication operation of the two matrixes; a plurality of matrix calculation units in the calculation node respectively execute part of the two matrix multiplication operations according to the fusion operator to obtain part of calculation results of the matrix multiplication operations; and executing a communication task to synchronize the partial calculation result. A management node only needs to send a fusion operator to a calculation node, the calculation node can complete calculation of a matrix multiplication according to the fusion operator, and the management node does not need to split the matrix multiplication operation into a plurality of subtasks for calculation and communication and then send the subtasks to the calculation node in sequence. The phenomenon that computing node resources are idle due to management node splitting and task issuing can be avoided, and the utilization rate of the resources and the efficiency of the distributed computing system are improved.

The invention relates to the technical field of medical data analysis, in particular to a heart disease key factor prediction method based on random forest, which comprises the following steps: 1, data preprocessing: preprocessing input data, separating independent variables from dependent variables, and dividing into a training set and a test set; 2, model construction and training: constructing a random forest model, generating a plurality of decision trees through a Bagging algorithm, and performing node splitting by adopting a CART algorithm; 3, performing model evaluation, and adjusting model parameters (such as the number of trees and a node splitting threshold value) to optimize prediction performance; 4, evaluating the generalization ability of the model by using an out-of-bag sample (OOB), and outputting a feature importance analysis result; and 5, model inspection: predicting a new sample based on the trained model, and generating a reliability calibration curve. By optimizing the data preprocessing process, the parameter debugging strategy and the model evaluation system, the accuracy and reliability of heart disease prediction are improved, and a scientific basis is provided for clinical diagnosis.

This solution discloses a new method for identifying RNA pseudouridine sites. This method proposes to use a variety of feature representation techniques to extract sequence features, and then uses the SVM-RFE method for feature selection to compress the feature space and optimize the feature subset. The best feature set after feature selection is input into the kernel method KeMRF based on polynomial random forest to identify pseudouridine sites in the sequence. As a newly proposed classification method, compared with the traditional random forest, KeMRF not only optimizes the discriminant criterion for node splitting, but also combines with an easy-to-interpret kernel method, making the classification performance more superior. This method reduces the training time of the model, improves the classification performance of the model, and further enhances the accuracy of identifying pseudouridine sites.

The application provides a persistent learning type index system for DRAM-NVM hybrid memory, comprising: a model part: dividing data into sub-data sets, expressing each sub-data set by using a linear regression model, collecting maximum keys covered by each linear regression model, constructing a two-layer recursive model index architecture RMI, indexing local maximum keys, and retrieving a model address to which a target key belongs; a data part: comprising two types of data structures, namely a Node node class and a Buffer buffer block class; a structure type adjustment part: according to the data structure type, being divided into buffer block expansion, buffer block conversion, node conversion and node splitting, and being used for ensuring that the index always remains correct and efficient during the whole working load process. The application solves the problem that the index height is not limited, and ensures that the index can correctly and efficiently access data under any type of working load.

This invention discloses a high-precision indoor personnel positioning method based on machine learning, comprising the following steps: collecting samples, labeling locations, and generating a dataset; extracting inertial features and clustering them to obtain behavioral pattern labels; constructing a signal topology map based on wireless data; constructing extreme random trees for each pattern subset, dynamically selecting splitting features and thresholds to generate an environment-adaptive tree; introducing node splitting into the topology map with topological proximity constraints, and performing local weighted regression combining topology and environmental weights to obtain each extreme random tree model; acquiring real-time data, obtaining pattern probabilities from the clustering model, selecting the corresponding tree model for traversal, and obtaining the predicted location for each pattern; fusing the predicted location output results with probabilities as weights, and updating the topology map, environmental statistics, and tree splitting thresholds. This invention achieves high-precision positioning in dynamic and complex indoor environments, improving positioning robustness and environmental adaptability.

The invention provides a test case generation method and device, equipment, a storage medium and a program product, and relates to the technical field of financial science and technology or the technical field of software testing. The method comprises the following steps: determining a splitting condition of nodes of an adaptive decision tree according to a test entropy and a failure rate gradient of a test object; performing node splitting on the adaptive decision tree represented by the test object based on the splitting condition to generate a test path, the test path being a path from a root node to a leaf node in the adaptive decision tree after node splitting; and generating a test case set according to the test path. The method provided by the invention can adapt to real-time updating of the test object and generate the test case with higher test efficiency.

This invention discloses an approximate nearest neighbor connection optimization method for encrypted databases based on PM# trees, belonging to the field of approximate nearest neighbor connection optimization technology for encrypted databases. It solves the problems of low retrieval efficiency and insecure retrieval processes in existing technologies. The method includes: projecting a high-dimensional dataset into a low-dimensional space using hash projection to obtain a low-dimensional dataset; encrypting the high-dimensional dataset to obtain an encrypted high-dimensional dataset; deleting redundant nodes generated during node splitting in the PM# tree to obtain a PM# tree, and using the PM# tree to build an index on the high-dimensional dataset to obtain an index file; encrypting the index file to obtain an encrypted index; and performing an approximate nearest neighbor connection query on the high-dimensional dataset based on the encrypted index, the low-dimensional dataset, and the encrypted high-dimensional dataset to obtain the query results. This achieves encryption of the high-dimensional dataset while reducing computational load, thus accelerating retrieval speed and improving retrieval quality.

The present application relates to the field of database technology, and proposes an operating method, server and storage medium for a distributed database. The present application sets up a main server and at least one sub-server, and divides the data of the data node into the storage area of ​​each sub-server for storage. When a key-value pair needs to be added to the database, the main server will determine the data node where the key-value pair needs to be inserted, and then send the page number and key-value pair corresponding to the data node to the sub-server where the data of the data node is located. After receiving the page number and key-value pair, the sub-server will perform key-value pair merging and data node splitting operations on the data node, and return the obtained data node splitting results to the main server. Finally, the main server performs an index node splitting operation based on the data node splitting result, thereby completing the operation of adding the key-value pair to the database. By setting it in this way, the data transmission pressure brought to each server can be reduced.

This invention provides a pipeline network segmentation method based on a node explosion algorithm, comprising: constructing an original pipeline network topology; node explosion: breaking the original pipeline network topology at designated segmentation nodes and deleting the designated segmentation nodes to obtain multiple subgraph topologies; node interpolation: in each subgraph topology, constructing edges between the designated segmentation nodes and their neighboring nodes, and re-inserting the designated segmentation nodes into each subgraph topology; and correcting the flow direction of all subgraph topologies after node interpolation. This invention solves the technical problem that the traditional DFS traversal segmentation method suffers from differences in segmentation results due to the different root node input order, and improves the efficiency of topology segmentation and the accuracy of simulation calculations.

The invention discloses a steel frame node identification method and system based on a data center, and belongs to the technical field of computer aided design, the steel frame node identification method based on the data center comprises the following steps: S1, intelligent classification and standardization processing of steel structure node members; s2, setting the connection mode and the connection mode of the steel structure node components; s3, the connection relation between the steel structure node components is calculated; s4, calculating a 3D node constructed by the steel structural member; s5, calculation of steel structure component splitting 2D nodes; and S6, calculating, matching and identifying the characteristic value of the basic node of the steel structural member. The method is used for realizing automatic and high-precision identification and classification of various nodes in the steel structure model, the technical bottlenecks of high human error rate, long modeling period and difficulty in adapting to complex node types in the traditional node identification process are solved, and the efficiency and quality of steel structure node design and construction are improved.

The application relates to the field of artificial intelligence, in particular to a big data information analysis method and system based on artificial intelligence, which comprises the following steps: acquiring a jump section direction vector and classifying label nodes with consistent directions, splitting a path lacking a mapping relationship to form an interrupted section, identifying nodes with continuous directions as bridge section entrances, rearranging path sections to generate a sliding connection sequence, and combining semantic fields to sort behavior trajectories to generate a focused analysis set; in the application, the jump section direction is classified according to node order and channel number, the direction aggregation boundary is determined, the semantic field is combined with the path position to identify an interruption, the positioning accuracy of the node sequence structure and the rationality of the paragraph division are improved, the bridge section connection is based on direction consistency screening, the path continuation logic is maintained, the behavior trajectory and the aggregation number are sorted according to the sequence and the level, the behavior aggregation sequence of the semantic chain is constructed, and the coherence of path identification in the channel aggregation area and the stability of node attribution are enhanced.

The application provides a decision tree recommendation method and system based on random sampling, which deduces the conduction path of faults among devices from the device physical topology association relationship in the device production detection scene, determines the sequential association order of each conduction node and the conduction trigger condition between nodes, groups and collects real detection data and virtual detection data according to the conduction node correspondence relationship to obtain a sampling pool, determines the decision tree branch splitting priority based on the upstream and downstream association relationship of the fault conduction path, obtains a decision tree, generates a step-by-step detection action sequence corresponding to the fault conduction path according to the reasoning path, corresponds each reasoning branch to a detection operation link to obtain a step-by-step detection action sequence, obtains an execution feedback result, updates the node splitting weight of the decision tree, adjusts the sample screening condition of the sampling pool, and obtains the sampling pool to optimize the decision tree. The application can realize the linkage adaptation of the decision tree structure and the data source, and continuously improve the accuracy of fault detection.

The invention discloses a data cleaning method and device, electronic equipment and a computer program product. Relates to the field of artificial intelligence, and comprises the following steps: collecting M data records in a target database, and extracting data information of target data corresponding to the data records from each data record, M being a positive integer; data features of the target data are generated based on the data information, the data features are input into the decision tree model, a cleaning strategy of the target data is obtained, the data features at least comprise time features and frequency features, and the decision tree model takes each data feature as a node and takes information gain as a splitting condition of the node; and processing the target data based on the cleaning strategy to obtain a target database after data cleaning. By means of the method and device, the problem that in the related technology, when data in a database is cleaned, mistaken deletion exists is solved.

The application discloses an online incremental learning method and system based on adaptive B+ tree index, relates to the technical field of artificial intelligence, and comprises the following steps: constructing an extended adaptive B+ tree index structure, inserting a new sample and calculating a track disturbance value; combining the disturbance value and a node state to generate a disturbance score, and adding the new sample and associated node samples to an incremental learning scheduling queue when a condition is met; obtaining an affected node and updating the structure code of the affected node when a node splitting, merging or reconstructing event occurs; monitoring sample label track changes, adjusting a label confidence marker when a threshold is exceeded, and limiting the participation frequency according to a screening mechanism; and performing a forgetting compression operation on a leaf node with low access frequency and small label fluctuation degree. Through the construction of the adaptive B+ tree index, the disturbance-driven scheduling, the structure change response and the forgetting compression strategy, the sample management is refined, the learning update is efficient, and the structure maintenance is controllable, so that the stability and resource utilization efficiency of the online incremental learning are improved.

The invention provides an electrical equipment classification rule self-learning method and system, and belongs to the technical field of electrical equipment management. Specifically, a C4.5 decision tree algorithm is adopted to train extracted feature data of the electrical equipment, a feature with the maximum information gain ratio is selected to perform node splitting, a classification rule in a decision tree form is formed, the classification rule is optimized based on feature importance evaluation, a data feedback mechanism is established, and a classification result is obtained. Newly generated equipment data are collected in real time and added into a training data set, the optimized classification rule is retrained and optimized through the training data set containing new data, and related nodes and rules in the decision tree model are updated. Key features are automatically screened based on information gain ratio and feature importance evaluation, low-contribution features are removed, the generalization ability of the model is improved, and overfitting is reduced. Classification rules are continuously and dynamically optimized through data feedback and incremental learning, the method adapts to dynamic factors such as equipment aging and environment change, and high accuracy is kept for a long time.

The invention relates to the technical field of abnormal aggregation, in particular to an abnormal data aggregation method for extra-high voltage equipment, which comprises the following steps: performing node splitting and classification through a random forest, and combining depth and sampling proportion setting and cross division training and verification sets, so that risk classification has hierarchical precision; the method comprises the following steps: establishing a dynamic risk index sequence, enhancing the reliability of a result under a constraint condition, carrying out interval mapping and extreme threshold comparison on the risk result to form the dynamic risk index sequence, introducing a graph neural network to carry out neighbor sampling and feature aggregation on the association strength between indexes, and realizing the conversion from single-point anomaly to a coupling mode between multiple indexes. According to the method, the causal relationship and the propagation path between different types of anomalies can be disclosed, a continuous abnormal chain structure is formed through multi-hop path extraction and sequence combination, scattered abnormal information is converted into chain expression with evolutionary logic, and the capacity of recognizing potential faults of a complex system in advance and guaranteeing overall operation safety is improved.

Embodiments of the present specification provide a method and device for training a risk control model and predicting a risk category. The method for training the risk control model comprises: obtaining each first training sample, the first training sample comprising a first feature value corresponding to an attribute feature of a business object and a category label indicating whether the business object has a business risk; determining a second feature value corresponding to a scenario feature, adding the second feature value to the corresponding first training sample to form a second training sample; constructing a first decision tree through node splitting based on each second training sample, the process of splitting for a current node comprising: splitting according to a splitting purity of any splitting condition in a plurality of candidate splitting conditions of the current node; regarding the scenario feature as a category feature during the splitting process; and determining a risk control model for classifying business objects based on the first decision tree. The complexity of the model system can be simplified, and the model has better performance.

The present application relates to a method, system, device and medium for predicting the rotational mode of interstellar molecules, wherein the method for predicting the rotational mode of interstellar molecules includes: obtaining core parameters of the interstellar molecular structure based on the three-dimensional structure of the interstellar molecule; inputting the core parameters into a decision tree model for training, during the training process, constructing a node splitting criterion of the decision tree based on the core parameters, gradually splitting the nodes of the decision tree based on the node splitting criterion, and adjusting the depth of the decision tree and the node splitting criterion by cross-validation, selecting the optimal splitting feature of the node splitting criterion by information gain; obtaining a rotational mode prediction model for predicting the rotational mode of the interstellar molecule and the corresponding physical parameters, realizing the prediction of the rotational mode of the interstellar molecule and the explanation of the related physical parameters, improving the accuracy of the prediction through a machine learning algorithm, and providing data support for subsequent interstellar molecule research.

The invention discloses a tunnel structure long-term stability evaluation method based on intelligent detection and monitoring data, and relates to the technical field of tunnel engineering monitoring and evaluation, and the method comprises the steps: S1, extracting tunnel monitoring data from a preset database, and carrying out the preliminary classification of user roles, processing the classified data set by adopting a branch condition setting and feature selection mechanism to obtain a mapped user demand set; s2, according to the mapped user demand set, analyzing demand difference by adopting a node splitting rule, if the demand difference points to a manager, extracting a security level quantitative index, and judging preliminary deconstruction information through information gain calculation; according to the tunnel structure long-term stability evaluation method based on intelligent detection and monitoring data, the accuracy of information expression and the efficiency of report generation are improved, a closed-loop information chain from original data to a final report is established, and the reliability and scientificity of tunnel structure long-term stability evaluation are effectively guaranteed.

The invention discloses an unstructured grid adaptive node splitting method for complex form three-dimensional modeling, which relates to the technical field of three-dimensional modeling and grid generation, and comprises the following steps: constructing an initial unstructured grid, and adopting a Delaunay triangulation generation unit to ensure that a covered modeling area has no overlapping gap; key features such as sharp corners and edges are extracted, geometric parameters are calculated, and feature priorities are divided; establishing a splitting judgment system based on a multi-dimensional index, and marking a to-be-split unit; determining split node coordinates, splitting according to unit types, and updating a topological relation; optimizing the grid quality through a smoothing algorithm, and deleting invalid units; and calculating the modeling precision and the grid quality, and if the convergence standard is not reached, repeating the process until the standard is reached. According to the method, complex morphological characteristics are accurately captured, a splitting strategy is adaptively adjusted, and node coordinates are highly fitted with boundaries; error accumulation is effectively controlled, numerical simulation and visualization requirements are met, and reliable support is provided for complex modeling in multiple fields.

This invention relates to the field of unmanned aerial vehicle (UAV) path planning technology, and more particularly to a distributed UAV local replanning and real-time obstacle avoidance method and system. The method includes real-time obstacle detection; triggering local trajectory replanning when the expanded obstacle region overlaps with the UAV's global guidance trajectory in time; independently generating locally replanned trajectories based on an improved time elastic band algorithm and broadcasting them to a decision set for distributed conflict detection; constructing a constraint tree for each conflicting UAV in the conflict set using a high-level node splitting method based on enhanced conflict search, according to a fixed time window synchronization mechanism; calculating the total cost of each child node in the constraint tree using the same cost function and calculation rules; selecting the optimal child node based on the total cost of each child node; and repeatedly selecting the optimal child node to obtain a conflict-free executable trajectory. This invention solves the deadlock problem caused by short-range UAV replanning trajectory conflicts in dense obstacle environments.

This invention discloses an embedded microkernel iterative generation and graph rewriting method and system based on physical constraint feedback. The method establishes a reverse feedback loop from a deterministic backend to the LLM (Limited Language Management Module): First, the deterministic backend is used as a "physical verifier" to tentatively synthesize the initial task graph; when physical violations such as register overflow or volume exceeding limits are detected, a physical diagnostic report containing bottleneck node information is generated; then, the diagnostic report is converted into natural language prompts using feedback encoding technology; finally, the LLM is driven to execute graph rewriting strategies such as node splitting and input serialization to correct the task graph topology. This invention achieves a closed-loop iteration of "physical problem, semantic solution." Experiments show that this method can increase the generation success rate of complex embedded tasks on extremely constrained hardware from 32% to 94%, significantly enhancing the robustness and adaptability of the system.

Provided in the present application are a distributed computing method and system, and a related device. The method comprises: a management node issuing a fused operator to a compute node, wherein the fused operator comprises two matrices and an instruction for implementing a matrix multiplication operation of the two matrices; on the basis of the fused operator, a plurality of matrix computation units in the compute node separately executing a partial operation in the multiplication operation of the two matrices, so as to obtain a partial computation result of the matrix multiplication operation; and then executing a communication task to synchronize the partial computation result. The management node only needs to send one fused operator to the compute node, and the compute node can complete computation of one matrix multiplication on the basis of the fused operator, without the need for the management node to split one matrix multiplication operation into a plurality of sub-tasks for computation and communication, and then sequentially send the sub-tasks to the compute node. Thus, idle resources of the compute node caused by the management node splitting and issuing tasks can be avoided, thereby improving the resource utilization and the efficiency of the distributed computing system.