62 results about "Graph alignment" patented technology

The invention discloses an ST code generation method based on graph alignment coding large model and multi-agent collaboration, and the method comprises the steps: receiving an ST code programming demand inputted by a user through a demand analysis module, refining and analyzing the demand based on multiple rounds of interactive conversations of the user and insight agents, and generating a standardized ST code programming demand; the retrieval module receives a standardized ST code programming requirement, and retrieves and obtains related knowledge through a retrieval agent in combination with an ST code knowledge base; and the double-agent collaborative self-correction code generation module receives standardized ST code programming requirements and retrieved related knowledge, a graph alignment coding large model constructed based on a graph neural network and a cross-modal alignment technology serves as a coding agent to cooperatively work with a review agent, ST code structure information is injected into the large model, and a final ST code is generated. According to the method, high-accuracy and high-reliability ST code automatic generation can be realized, and the development efficiency of a PLC program in the industrial control field is improved.

The invention relates to the technical field of large model knowledge questions and answers, and discloses an enterprise knowledge question and answer system based on a large language model, which comprises a heterogeneous semantic embedding module, a permission mask calculation module, an enterprise term alignment module, a candidate knowledge query module and a question and answer generation output module, in the prior art, an enterprise question and answer mode is realized only depending on keyword matching or a fixed FAQ rule, and particularly under the conditions of enterprise internal knowledge distribution fragmentation, authority differentiation and term ambiguity, accurate, compliant and context-consistent semantic question and answer cannot be realized. Due to the fact that multi-mode unified embedding, the semantic level authority control mechanism, the term graph alignment model and the RAG enhanced question and answer mechanism are introduced, cross-format and multi-department enterprise intelligent question and answer is achieved, the problems of semantic drift and unauthorized leakage are avoided, and the accuracy and safety of the question and answer system are improved.

The invention discloses a cross-modal alignment method and device for a multi-modal large language model, and relates to the technical field of multi-modal alignment, and the method comprises the steps: obtaining an image, carrying out the target detection, and generating a local image fine-grained feature; obtaining a text, extracting text features, extracting a text entity relationship, and obtaining an entity relationship dependency tree as a text entity relationship graph; performing local fine-grained semantic alignment on the local image fine-grained features and the text features; using the scene graph model to obtain a visual relation graph between detection targets in the image; performing graph structure alignment and graph semantic alignment on the text entity relation graph and the visual relation graph; and fusing the semantic alignment features and the graph alignment features to obtain semantic alignment features with rich fine grit, global semantic alignment features and richer relationship alignment features, and inputting the semantic alignment features, the global semantic alignment features and the richer relationship alignment features into a multi-modal large model. According to the method, alignment of different modal data on different feature dimensions is realized, so that the different modal data are accurately associated.

The invention relates to the technical field of image segmentation, in particular to an image anaphora segmentation method based on autoregression vertex generation and language structure guidance, and the method comprises the steps: obtaining an input image and a corresponding anaphora query statement; acquiring multi-scale visual features of the input image by using a pre-trained visual encoder; obtaining language features of the anaphora query statement by using a pre-trained language encoder; constructing an image semantic scene graph based on the multi-scale visual features; constructing a query language dependency graph based on the language features; utilizing a graph alignment mechanism and a feature alignment mechanism to realize structured multi-modal fusion of the image semantic scene graph and the query language dependency graph; based on a structured multi-modal fusion result, generating a contour point sequence of the target instance by adopting a language-guided autoregression instance generation method; generating a segmentation mask of the target instance according to the contour point sequence; and outputting the segmentation mask as an image anaphora segmentation result, thereby realizing remarkable performance improvement.

The invention discloses an enterprise and policy service automatic matching method based on machine learning, and the method comprises the steps: carrying out the processing of enterprise multi-source data, constructing an enterprise portrait graph, and generating an enterprise graph embedding and structuring feature set; performing policy text analysis and condition recognition, constructing a policy condition graph and generating condition graph embedding representation; constructing a condition constraint field based on policy conditions, and generating a cross-graph alignment relationship and fusion features; integrating a multi-source feature input improved model to carry out joint modeling, and outputting a basic matching score; constructing a condition boundary manifold, and generating an anti-fact feature sample and a matching elasticity score; and based on the basic and elastic scores, generating a comprehensive score and outputting matching result information. According to the invention, by introducing graph structure perception alignment, conditional constraint guide interaction and a multi-channel scoring aggregation mechanism, high-precision, high-interpretability and intelligent reachability automatic matching between enterprises and policy services is realized.

The invention aims to provide a multi-modal cancer prognosis evaluation method based on a graph. The method comprises the following steps: S1, pre-processing multi-modal data, segmenting a pathological tissue part into patches, extracting patch features, and performing surrounding type composition on the patches according to cell migration continuity; the omics data is subjected to GSEA enrichment classification, wherein five cancer development processes are included; s2, single-modal features are obtained, and modal corresponding features are obtained by using a modal enhanced and shared graph mask auto-encoder; s3, cross-modal interaction is carried out, and multi-modal data distribution mapping alignment is realized by using a cross-modal alignment module; s4, performing multi-modal fusion, and fusing the cross-modal interaction feature and each single-modal feature by using a Hyperdge-Mixer module to obtain a modal fusion feature; and S5, risk prediction: inputting the final multi-modal representation into a risk prediction module, and finally obtaining a risk prediction result. The cancer prognosis is realized by effectively fusing pathology, genes and clinical data, an end-to-end prognosis analysis framework is realized by utilizing a deep learning technology, the method has good performance and relatively high interpretability, and the clinical application practicability is enhanced.

Systems / techniques that facilitate monolith-to-microservice refactoring via source-code-to-domain-model graph comparison are provided. In various embodiments, a system can access source code of a monolithic application and a target domain model corresponding to the monolithic application. In various aspects, the system can refactor the monolithic application into microservices, by aligning a first graph representing the source code with a second graph representing the target domain model.

The invention provides a molecular-graph-oriented primitive-level representation learning method and system, and the method comprises the steps: carrying out the noise addition and reconstruction of an original molecular graph based on a graph diffusion model, calculating an edge reconstruction error, and dividing the original molecular graph into a plurality of primitive sub-graphs according to the edge reconstruction error; constructing an atomic view and a primitive view, and encoding the atomic view and the primitive view by using a graph neural network encoder to obtain an atomic embedding matrix and a primitive embedding matrix; and constructing a positive and negative sample pair based on the atomic view and the primitive view, and optimizing an atomic embedding matrix and a primitive embedding matrix by adopting a Babontwin double-view contrast loss function to realize cross-granularity expression alignment. According to the method, molecular primitives are extracted in a self-supervised mode through the diffusion model, systematic defects of a traditional method in the aspects of cross-granularity structure modeling, semantic guidance contrast learning and multi-level expression fusion are overcome through atom-primitive double-view alignment, and the semantic completeness and task generalization ability of molecular expression are remarkably improved.

The invention provides a double-alignment part multi-view clustering method based on an anchor point diagram, and belongs to the field of multi-view feature processing, and the method comprises the steps: 1) re-representing an unaligned sample based on the anchor point diagram; 2) re-representing preliminary alignment of unaligned samples; 3) training a graph convolutional neural network; and 4) carrying out secondary alignment and clustering on the potential features which are not aligned. The method mainly solves the problem of data mismatching in multi-view clustering, and due to the fact that when a previous method is used for solving data mismatching, a large number of alignment examples need to be trained, and the alignment examples cannot be accurately recovered due to the fact that the alignment examples reconstruct corresponding view relations respectively. In order to solve the problem, preliminary view alignment is realized by constructing an anchor point diagram relationship, and then graph embedding is learned by using a multi-view graph convolutional network. And performing further alignment and fusion on the multi-view features through comparative learning and a Hungary algorithm, and finally obtaining a clustering result. Experimental results show that the method can effectively recover the alignment relationship of different view samples.

The invention discloses a referring multi-target tracking method based on multi-modal correlation graph alignment and a multi-level cross fusion model, and relates to a computer vision technology. The method comprises the following steps: sampling and enhancing training data containing a video frame sequence and referred sentences, and extracting visual and text features; performing early fusion through a cross attention module, processing through an encoder, and initializing query by using a semantic injection mechanism; in the decoder, the self-attention and semantic clearness intensifier adjusts feature interaction, and multi-level fusion is carried out on output of each layer; a correlation graph is constructed by adopting cosine distance, triple loss learning is carried out after samples are screened, and a fine-grained corresponding relation between region-level vision and word features is established; and finally, performing branch prediction and optimization. A precise relation is captured through a multi-modal correlation graph alignment model, semantic understanding is enhanced by using a multi-stage cross fusion model, and the adaptive learning ability of the model is improved. Experiments show that multiple indexes are remarkably improved, positioning and tracking are more accurate, and robustness is higher.

The invention discloses a chip circuit state self-diagnosis method based on a graph neural network, and the method comprises the following steps: obtaining a chip netlist, layout and process information, collecting multi-source state data to construct a multi-domain coupling graph, and generating a multi-scale calculation tree based on a critical path; standardizing key node noise data, extracting noise fingerprints, constructing a noise consistency graph, and aligning the noise consistency graph with the multi-domain coupling graph to form node features; inputting the multi-scale calculation tree into an improved GFT model, and generating unified graph representation through structure token coding and feature fusion; constructing a multi-scale transition graph neural network, simulating multi-domain state propagation and outputting an abnormal score; anti-fact operation and reversible disturbance are applied to the high abnormal score nodes to generate an anti-fact state diagram, and a fault causal chain is constructed; and performing topological variation inversion based on the abnormal score and the causal chain to obtain a fault position, a fault type and a fault degradation trend. According to the invention, high-precision explainable chip fault self-diagnosis can be realized.

The invention discloses a same and different matching network node classification method and system for graph comparison learning based on macro and micro message passing, and a medium, and the method comprises the steps: firstly constructing a k neighbor graph based on node feature similarity through the node features of an original graph, and carrying out the random connection edge removal of the original graph and the k neighbor graph, generating a corresponding structure view and a feature view, then performing view role definition on the structure view and the feature view according to the same and different matching types of the network, and determining a priority view and an auxiliary view; performing optimal and auxiliary view alignment coding on the priority view and the auxiliary view to obtain node representations under different views; further constructing an InfoNCE comparison loss function, and performing optimization training on node embedding representation of the model; and finally, the optimized node embedding representation is used for a node classification task, and training and evaluation are carried out. According to the method, the node representation quality can be effectively improved, so that the accuracy of node classification in the same and different matching network is improved.

The invention relates to a cross-scale image bridging decoupling target detection method based on dynamic adjacency. Comprising the steps of 1, data preprocessing and segmentation, 2, batch sampling and multi-scale feature extraction, 3, high-level self-attention enhancement, 4, same-layer dynamic adjacency learning and alignment, 5, cross-scale graph bridging, 6, gating residual steady-state injection, 7, decoupling detection head prediction, 8, training and optimization and 9, reasoning and evaluation. According to the method, adjacency of dynamic graphs on the same layer is learned on a multi-scale path, graph alignment is carried out, a two-way bridging relation'directional leveraging 'is established between adjacent scales, gating residual error steady-state injection is carried out to a decoupling detection head, high IoU precision is kept, meanwhile, detection performance of small-size and short-time scenes is improved, and overall resource consumption is reduced.

The application discloses a ST code generation method based on a graph alignment coding large model and multi-agent cooperation, a demand analysis module receives an ST code programming demand input by a user, refines and analyzes the demand based on multi-round interactive dialogues of the user and insight agents, and generates a standardized ST code programming demand; a retrieval module receives the standardized ST code programming demand, retrieves and acquires relevant knowledge by means of a retrieval agent and in combination with an ST code knowledge base; a double-agent cooperative self-correcting code generation module receives the standardized ST code programming demand and the retrieved relevant knowledge, a graph alignment coding large model constructed based on a graph neural network and a cross-modal alignment technology, and the module is used as a coding agent and an inspection agent to work cooperatively, injects ST code structure information into the large model, and generates final ST code. The application can realize high-accuracy and high-reliability ST code automatic generation and improve the development efficiency of PLC programs in the industrial control field.

The invention belongs to the technical field of intelligent traffic, and relates to a vehicle image-text retrieval method and system based on multistage semantic graph alignment and attribute enhancement. According to the method, fusion visual embedding is obtained according to the vehicle video, and fusion text embedding is obtained according to the text data; obtaining an updated visual attribute embedding vector and an updated text attribute embedding vector based on fusion visual embedding and fusion text embedding; performing semantic matching on the updated visual attribute embedding vector and the updated text attribute embedding vector, and aligning vehicle cross-modal semantic attributes from three different levels by utilizing multi-granularity semantics to obtain final visual attribute embedding and final text attribute embedding; and according to the final visual attribute embedding and the final text attribute embedding, obtaining the similarity between the vehicle video and the artificial text description, wherein the vehicle video with the highest similarity is a vehicle video retrieval result. According to the method, the target vehicle can be quickly and accurately positioned, and higher robustness and higher retrieval success rate are shown.

The invention discloses a contrast learning recommendation method based on adaptive view optimization, and the method comprises the following steps: enabling an adaptive view enhancement module to generate an enhanced interaction graph from an original user article interaction graph, and calculating adaptive enhancement loss; inputting the original user item interaction diagram and the enhanced interaction diagram into a diagram message transmission encoder module, and performing neighborhood information propagation and node feature fusion to generate node characterization; executing a view alignment task through a view alignment module, fusing node representations, and outputting an aligned node feature matrix; performing final optimization of the recommendation model; and synchronously calculating Bayesian personalized sorting loss and multi-level contrast learning loss, and continuously iterating until the recommendation model converges. By means of a multi-level contrast learning loss-based contrast learning strategy, local negative samples are screened through feature distances, weights are dynamically adjusted, difficult negative samples similar to anchor point samples are emphatically optimized, and the overall sample quality is improved, so that the accuracy of a recommendation result is improved.

This invention relates to a few-shot image segmentation method, device, and medium based on cross-view consistency prototype learning. The method involves acquiring a support image with foreground mask annotations and a query image to be segmented to construct a few-shot segmentation task. The support and query images are input into a two-stream encoder to extract semantic and structural features. Based on the structural features, cross-view structural alignment is performed through spatial view graph alignment to obtain a cross-view consistent structural semantic representation and construct support and query prototypes, calculating the consistency constraint loss. Foreground and background prototypes are extracted based on the mask annotations of the support image, and a discriminative prior is constructed by combining the query features and performing foreground enhancement. Support and query hints are generated based on the enhanced features, and visual reference hints are obtained interactively and input into a decoder. The predicted mask of the query image is used as the segmentation result. Compared with existing technologies, this invention has advantages such as strong structural consistency, high foreground discriminativity, and strong cross-view generalization.

The present invention proposes an early tracing method for public opinion attacks based on causal reasoning and bi-graph alignment, which extracts event causal relationships from historical public opinion data to construct a hierarchically causal library with dynamic updates; based on the historical public opinion evolution path, uses graph convolutional networks to extract real-time propagation features, combines reinforcement learning and Bayesian abduction to generate internal cause links with low confounding bias, and identifies external cause nodes driving the links by means of a Transformer to identify threshold factors; finally, fuses internal and external cause nodes to construct a causal hypergraph, aligns nodes with the backbone of public opinion facts that aggregates social propagation, event evolution, and motivation logic, decodes abnormal subgraphs, and reversely locks sensitive users and communication structures, thereby improving the accuracy and timeliness of tracing.

A method for a server to provide a personalized space map may comprise the steps of: acquiring a first space map generated in advance and including a first map feature representing space information about a space; obtaining a second spatial diagram generated based on the sensor data of the spatial measurement, the second spatial diagram including a second diagram feature; selecting a graph feature for graph alignment from the first graph feature and the second graph feature; aligning the first spatial diagram and the second spatial diagram based on the selected diagram features; identifying a local region in the space where there is a difference between the first space map and the second space map; and generating a personalized spatial map by determining spatial information about the local region based on the second map feature.

The application provides a kind of molecular graph-oriented element level representation learning method and system, based on graph diffusion model, original molecular graph is added with reconstruction, edge reconstruction error is calculated, and original molecular graph is divided into multiple element subgraphs according to edge reconstruction error;Atom view and element view are constructed, and graph neural network encoder is used to encode atom view and element view respectively, to obtain atom embedding matrix and element embedding matrix;Based on atom view and element view, positive and negative sample pairs are constructed, and barlow twin double view contrast loss function is used to optimize atom embedding matrix and element embedding matrix, to realize cross-granularity representation alignment.The application extracts molecular element through diffusion model self-supervision, and solves the systematic defects of traditional method in cross-granularity structure modeling, semantic guided contrast learning and multi-level representation fusion through atom-element double view alignment, significantly improves the semantic completeness and task generalization ability of molecular representation.

The present invention discloses a method, system and medium for node classification of a homogeneous and heterogeneous matching network based on macro-micro message passing for graph contrastive learning, including: First, construct a k-nearest neighbor graph based on node feature similarity using the node features of the original graph, and randomly remove connected edges from the original graph and the k-nearest neighbor graph respectively to generate corresponding structural views and feature views. Then, define view roles for the structural views and feature views according to the homogeneous and heterogeneous matching types of the network to determine the priority view and the auxiliary view; Next, perform alignment encoding on the priority view and the auxiliary view to obtain node representations under different views; Furthermore, construct an InfoNCE contrastive loss function to optimize and train the node embedding representation of the model; Finally, use the optimized node embedding representation for the node classification task for training and evaluation. The present invention can effectively improve the quality of node representations, thereby improving the accuracy of node classification in homogeneous and heterogeneous matching networks.

The invention provides a cancer single cell clustering method based on semantic perception contrast learning, and the method comprises the steps: screening high-expression genes, randomly shielding the gene expression of each cell, and generating two enhancements of cell data; extracting semantic information from the view by using an automatic encoder, and training the encoder by minimizing reconstruction loss; according to different views, respectively constructing a cell-cell graph and a pseudo-tag graph; a dual contrast alignment module is applied to capture semantically invariant representations in the enhancement by utilizing semantic information in the generated pseudo-tag graph. According to the method, the similarity graph and the high-confidence pseudo-tag graph are aligned to ensure that the enhancement shares a consistent semantic structure, the relationship between units and the structural consistency of reliable class information are kept, effective learning of semantic invariant unit representation is realized, semantic information extraction and clustering of cancer single-cell data are effectively performed, and the accuracy and the reliability of cancer single-cell data are improved. And the robustness and the clustering effect of the model are improved.

The invention provides a sign language recognition system based on cross-modal diagram alignment, and relates to the technical field of intelligent sign language recognition. Respectively inputting the medical emergency sign language video vocabulary set into the medical sign language vocabulary feature extraction module and the medical sign language visual feature extraction module for training; and training the medical sign language visual feature extraction module by using the medical sign language vocabulary modal diagram corresponding to each medical first aid sign language video and the medical sign language video modal diagram corresponding to each medical first aid sign language video. According to the method, the medical sign language vocabulary features and the visual features are combined, cross-modal graph alignment is achieved, key information expressed by the sign language can be rapidly and accurately recognized, the generalization ability of the model is improved, and the requirements for the key information extraction speed and accuracy in first aid are met.

An embodiment of the present application discloses a data processing method. The processing device can further add the node information type corresponding to the graph node as the basis for judging whether the graph node is matched, based on the node information possessed by the graph node, thereby enriching the richness of input data during model training and matching application, thereby improving the accuracy of node matching and making the trained node matching model more suitable for performing graph alignment tasks.

The invention relates to the technical field of deep learning, in particular to a model construction method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining a training data set required by a to-be-generated target model, various spleen and stomach disease task instructions and a preset model; based on a pre-constructed knowledge graph, performing alignment training on the preset model, injecting structured knowledge in the knowledge graph into a parameter space of the preset model to obtain an intermediate model, the knowledge graph being used for indicating a training basis and a logic rule corresponding to a training target model and a professional decision basis and an operation specification of clinical diagnosis and treatment; and based on the various spleen and stomach disease task instructions and the training data set, adjusting the intermediate model which completes map alignment to obtain a target model which is used for generating a diagnosis and treatment scheme corresponding to the spleen and stomach disease patient. The problems that the diagnosis and treatment result of the spleen and stomach disease is inaccurate, and an effective treatment scheme cannot be provided for patients with the spleen and stomach disease can be solved.

The application provides a multi-modal knowledge graph representation learning system and product based on subgraph learning, and relates to the technical field of representation learning, and comprises a multi-modal subgraph construction subsystem, which extracts the multi-modal structural information of the head entity in a target triple, and obtains a multi-modal subgraph, wherein the multi-modal subgraph comprises a visual scene graph and a self-centered graph of the head entity; a neighborhood feature aggregation subsystem, which fuses multiple channel feature components of entity features to obtain multi-modal embedding representation of the entity; and a link prediction subsystem, which predicts missing entities and relationships in the target triple based on the multi-modal embedding representation of the entity. The application uses a novel picture structure information extraction mechanism, introduces structured information in the picture by using an efficient graph alignment mechanism, effectively aggregates the neighborhood topological information of the target entity, lays a foundation for efficient fusion of subsequent multi-modal information, and can efficiently learn multi-modal knowledge graph embedding representation.

The invention provides a multi-view clustering method and system based on adaptive anchor graph learning and tensor compensation, and belongs to the technical field of machine learning, and the method comprises the steps: obtaining a data matrix of multi-view data; the optimal anchor point number of each view is automatically determined through a layered bipartite neighborhood clustering method, and a specific anchor map of the view is constructed; selecting a view with optimal clustering quality as a reference view, unifying a clustering representation space, and introducing an enhanced tensor rank regularization item to compensate structure information possibly lost in a cross-view alignment process; and constructing a unified similarity matrix fused with multi-view representation, and obtaining a final clustering result through spectral clustering. According to the method, under the condition that the number of anchor points and the number of clusters are not preset, the anchor point diagram structure of each view can be learned in a self-adaptive mode, high-order complementary information among multiple views is captured, the robustness and expandability of clustering performance are improved, and the method can be widely applied to analysis and mining tasks of large-scale complex data.

The invention belongs to the technical field of network security, and discloses a method and a device for generating a network security attack path based on weak supervision and graph alignment, which adopt a weak supervision entity annotation method and a multi-source voting mechanism of keyword matching, special phrase matching and model integration to remarkably reduce the annotation requirements of experts and solve the problem of insufficient data. And a full-graph alignment mechanism is designed and a verb enhancement strategy is combined, so that boundary limitation of connected sub-graphs is broken through, a cross-regional attack behavior chain is completely captured, and the path recognition accuracy is improved. A node sequence attribute is introduced to explicitly construct a time sequence attack path, causal logic of attack steps in a report is strictly followed, and an actionable decision basis is provided for automatic threat modeling.

A computing system is disclosed for augmenting semantic retrieval using structured knowledge constraints. The system obtains content segments from one or more source documents and processes the content through a dual-indexing pipeline that generates vector embeddings and constructs a graph representation of inter-segment relationships. A vector index and a graph-based structure are maintained in parallel to support retrieval workflows. Upon receiving a search query, the system initiates vector-based retrieval to identify semantically similar segments and concurrently traverses the graph to identify contextually connected segments. Retrieved segments are cross-referenced using constraint resolution logic that evaluates identifier overlap, relationship proximity, and graph alignment metrics. The system adjusts relevance scores using adaptive weighting and structural validation and produces a unified ranking of content segments. Validated results may be presented for review, integrated into external applications, or refined through feedback signals that update scoring parameters and edge weights over time.