33 results about "Knowledge sharing" patented technology

The application belongs to the technical field of collaborative learning, and particularly relates to a parameter efficient collaborative learning method and system suitable for intelligent sensing devices. The application is directed to a collaborative application scene of multiple distributed terminal devices with sensing, computing and communication capabilities. Under the premise of freezing backbone parameters, a shared adapter and a personalized adapter are introduced. Personalized parameters are only updated and saved locally to maintain single-device adaptation capability. Shared parameters realize knowledge sharing through interaction and weighted aggregation of neighboring devices. An adaptive weight matrix is used to fuse shared prediction and personalized prediction, so that global collaboration and personalized adaptation are considered in the distributed sensing scene with significant data distribution difference. Meanwhile, only the lightweight trainable part is updated and exchanged, which significantly reduces the end-side computing burden and communication overhead, and maintains the efficiency and convergence stability of collaborative learning under resource-limited conditions.

The application provides an intelligent substation anomaly detection method based on meta-learning technology, first, a multi-working condition task library is constructed for meta-training. Secondly, a double meta-knowledge learning framework is designed to learn task-level model initialization parameters suitable for rapid adaptation. Subsequently, online reconstruction error driven anomaly detection is deployed, and the adapted model is used to calculate data reconstruction error in real time, and combined with the adaptive threshold to accurately distinguish and alarm grading. Finally, the system supports a dynamic updating mechanism, which continuously collects verified samples and periodically triggers meta-updating. The application significantly reduces the data dependence and deployment cost in new scenarios, improves the model's generalization ability to multiple working conditions and sensitivity to rare anomalies through meta-knowledge sharing, and ensures the long-term effectiveness of the detection system through online learning mechanism, providing an efficient, adaptive and evolutionary security protection solution for intelligent substations.

A blockchain-based knowledge-sharing intelligent customer service AI response accuracy optimization system, belonging to the fields of blockchain and artificial intelligence technology, is disclosed. The system includes a multi-source knowledge acquisition unit, a blockchain knowledge sharing unit, a knowledge quality assessment unit, an AI response optimization unit, and an interactive feedback and iteration unit. The multi-source knowledge acquisition unit collects customer service knowledge data from different departments, business lines, and application scenarios. The blockchain knowledge sharing unit receives pre-processed knowledge data and constructs a decentralized knowledge-sharing network based on distributed ledger technology. The knowledge quality assessment unit performs multi-dimensional quality assessments of the knowledge stored in the blockchain. The AI ​​response optimization unit trains and optimizes the intelligent customer service AI model based on high-quality knowledge samples. The interactive feedback and iteration unit collects user feedback from the intelligent customer service system. All units operate collaboratively to form a closed-loop mechanism.

The application provides a new energy mine truck thrust rod fatigue life prediction method, belonging to the field of federated learning and application technology. First, a multi-mine area multi-source fatigue feature dataset is constructed. Then, a double-layer neural network structure based on Bayesian modeling is designed, including a sparse prior generation module, a thrust rod time series feature encoding module and a thrust rod residual life Bayesian prediction module, to realize model parameter sparsification, time series feature expression and uncertainty quantification. Further, a graph structure based federated training and information aggregation mechanism is adopted, combined with global aggregation and local graph modeling strategy, to realize knowledge sharing and personalized adaptation between different mine areas. Finally, an online fine-tuning mechanism based on uncertainty driving is proposed, realizing rapid adaptive optimization of the global model in the new mine area environment through Bayesian inference. The method can significantly improve the accuracy, robustness and cross-domain generalization ability of thrust rod life prediction while ensuring data privacy.

The application provides a robot cluster knowledge sharing method with body perception driving, belonging to the technical field of computer data processing, and generating a verifiable knowledge unit with a perception fingerprint containing sensor modalities and state parameters according to the perception data and task state of the robot. Through the on-demand accurate allocation of communication and computing resources, the overall operation efficiency of the robot cluster is greatly optimized, the passive global information broadcast is changed into the active knowledge acquisition of the robot individual based on the task demand, and the metabolic fusion mechanism based on the trust chain is combined to ensure that each robot only processes the most critical and most credible information for the current task, avoids information overload, significantly reduces the network bandwidth and individual processor load, and provides support for the deployment and application of large-scale robot clusters.

The application belongs to the technical field of personalized federated learning, and discloses an industrial control system intrusion detection method based on personalized federated learning. The method solves the problem of insufficient model generalization ability under Non-IID data, realizes the balance between local adaptability and global threat perception through similar client cooperative learning, and proposes a similarity calculation mechanism based on JS divergence to realize collaborative modeling between similar clients, thereby ensuring personalization and enhancing global knowledge sharing ability. An industrial network feature extraction tool is designed to support Session, Flow and Stream three-layer modeling, extract high semantic features of industrial protocols, and improve the quality of model training data. The tool has industrial protocol perception ability, can extract various high semantic features from various key industrial control protocols, effectively support the standardized conversion of multi-entity data sources, and enhance the representativeness and difference of model training data.

The application discloses a hotel information management method based on a cloud platform and belongs to the technical field of electric digital data processing. The method comprises the following steps: constructing a cloud-edge-end three-layer collaborative network, collecting data on the end side, performing dynamic weight mixed scheduling and local inference on the edge side edge server, performing federal learning training and knowledge graph construction on the cloud side cloud platform, calculating the comprehensive priority according to the user emergency degree, network load and calculation complexity of the edge server, dividing the task into local real-time, local batch processing and collaborative processing queues, training a global user preference prediction model by using a federal learning framework, constructing a hotel operation knowledge graph, inferring potential service demand by using a graph neural network, fusing the model prediction and graph inference results to generate an individualized strategy and executing the strategy, and realizing closed-loop feedback optimization through real-time monitoring and deviation hierarchical processing. The application realizes resource elastic scheduling, cross-domain knowledge sharing under privacy protection and self-adaptive optimization.

The application discloses a kind of multi-agent collaborative architecture systems based on layered collaborative protocol, including at least two AI agents, layered collaborative protocol module and anti-cycle dependency mechanism.The layered collaborative protocol module includes state synchronization layer, task coordination layer, knowledge sharing layer and supervision feedback layer, for realizing the efficient collaboration between agents;The anti-cycle dependency mechanism includes event bus asynchronous communication module, one-way data flow control module and supervision object separation module, to prevent the cycle dependency between agents.The application solves the technical problems of the existing multi-agent system, such as cycle dependency, low collaboration efficiency, inconsistent state, difficulty in knowledge sharing, and imperfect supervision mechanism, by layered collaborative protocol and anti-cycle dependency mechanism, significantly improves the collaboration efficiency and system stability of multi-agent system.Experimental results show that the cycle dependency elimination rate of the application reaches 100%, the collaboration efficiency is improved by 50%-80%, the state consistency is above 99.9%, and the system availability is above 99.9%, with significant technical effects.

A system (100) for performance enhancement of professionals and participants using artificial intelligence is provided An automated transcription module (130) records and transcribes video and audio data. An analysis module (135) performs real-time analysis using a large language model and a graph retrieval-augmented generation training model to identify performance gaps and areas for improvement. A performance metrics evaluation module (140) extracts a plurality of metrics to evaluate the performance. The personalized training module (145) generates customized training programs with rule-based prompts. A collaboration module (150) aggregates best practices and fosters knowledge sharing among professionals, while the performance tracking module (155) monitors their participation in development activities and tracks progress. A machine learning and adaptive learning module (160) predicts potential challenges and adapts training programs. A rewards module (165) enhances engagement through badges, leaderboards, and progress tracking, and a simulations and scenario-based learning module (170) for interactive practice in virtual environments.

The application relates to the technical field of blockchains, and provides a method for screening, sharing and caching edge LLM trusted knowledge based on a blockchain, which comprises the following steps: a data provider node records the meta-information of to-be-shared data and storage credentials on the blockchain; a verifier node acquires data from the blockchain and performs multidimensional evaluation to generate a verification score; based on the verification score, a preset reputation calculation model is used to update the reputations of data items and the reputation of the verifier node; according to a preset constraint condition, a set of data items meeting the condition is screened out from the updated data item reputations, the multidimensional quality evaluation results of the data items and a compromise function composed of data delay and credibility, and the set of data items is transmitted to a cache node, and a screening result summary is recorded on the chain; and the cache node performs hierarchical storage and dynamic management of the set of data items based on a cache evaluation value. The application realizes the collaborative optimization of data quality, process credibility and acquisition efficiency in the process of sharing edge LLM knowledge.

An inter-satellite fault knowledge sharing method based on federated learning, when a master satellite or a slave satellite in a satellite constellation appears a fault, first, a model is trained with local data of the satellite; after the model training is completed, the model parameters of the satellite are transmitted to other satellites, and other satellites update their own models based on the transmitted model parameters. The present application reduces the amount of information exchanged between satellites by transmitting the parameters of each model between satellites without directly transmitting fault data, and simultaneously updates the multi-satellite fault diagnosis model based on the transmitted model parameters, thereby realizing the continuous updating of the fault diagnosis model on each satellite, and further improving the accuracy of subsequent constellation fault diagnosis.

This invention discloses a heterogeneous federated cross-domain recommendation system based on hypernetworks and its implementation method. This method is designed for heterogeneous federated cross-domain recommendation scenarios, allowing participating clients to flexibly deploy personalized recommendation models based on their local data scale. The method includes performing multi-objective collaborative pre-training within each single domain to obtain high-quality initial embedding representations; and constructing a hypernetwork-based bridging model generator on the server side. By fusing the architecture description of the client's local bridging model with specific user preference information, it dynamically generates weight parameters that match the client's model structure. The client uses these parameters for local fine-tuning and sends gradient updates back to the server for global parameter iteration. This invention effectively overcomes the challenge of directly aggregating model parameters due to heterogeneous embedding layer dimensions, and achieves efficient knowledge sharing and cross-domain collaborative training among heterogeneous clients while strictly protecting user privacy.

Methods, systems, and computer programs are presented for learning robust embeddings from multiple modalities and tasks with a unified architecture. A framework is presented for learning robust embeddings from multiple modalities and tasks with improved performance and generalization. The framework includes modality-specific encoders, a shared transformer backbone, and task-specific heads. It allows for end-to-end training and cross-modality knowledge sharing, and the framework results show improvements over the state-of-the-art results on popular benchmarks. A training strategy is presented for leveraging knowledge from multiple modalities and an iterative training mechanism for self-supervised masked pretraining.

The application discloses a steel structure welding quality evaluation system and method based on distributed computing. The system includes a distributed gradient sensing unit, a physical field reconstruction unit, an entropy calculation unit, a chaotic feature extraction unit, a resource scheduling unit, a physical constraint verification unit, a chaotic synchronization analysis unit, a data aggregation processing unit, a quality comprehensive evaluation unit and a knowledge base management unit. By distributed acquisition of multi-physical field signals in the welding process and calculation of spatial gradient, the data transmission amount is significantly reduced; the complete physical field is reconstructed from the gradient information by using the multi-grid method; the potential defect area is identified and the computing resources are dynamically optimized based on information entropy analysis; the nonlinear characteristics of the sound signal are analyzed by using the chaos theory; multi-dimensional cross verification is carried out through physical constraint verification and chaotic synchronization analysis; and the federal learning is used to realize multi-site knowledge sharing. The application realizes real-time, accurate and comprehensive evaluation of the welding quality, and provides reliable technical support for steel structure welding quality control.

The application provides a personalized federated learning method and system of a shared model, comprising: a server end saving and initializing a shared model, a global model and a global class prototype of a client, and the client initializing a local learning weight vector; the server sending a shared model of other clients to one client, the client setting the local shared model as the global model and training, and uploading the shared model and the local class prototype to the server end; and the server calculating the global class prototype and the global model according to all received shared models and local class prototypes. The application solves the knowledge sharing problem in federated learning, improves the performance of the personalized model, solves the deviation problem in the local model training of the client, and the obtained personalized prototype contains more global information than the personalized prototype, so that the effective fusion of the global knowledge and the local characteristics is realized while reducing the state dependence and the communication overhead, so as to improve the robustness and adaptability of the model in a complex scene.

The application provides a federal learning-based industrial internet 5G-A communication evaluation model training method and device. The method comprises the following steps: a server determines a client identifier set participating in training; the server distributes a first initial artificial intelligence model to the clients; the clients are used for training the first initial artificial intelligence model according to local simulation communication data, and upload the obtained model parameters to the server; the server aggregates the received model parameters, and updates the first initial artificial intelligence model; the server iterates until the parameters converge, obtains a second initial artificial intelligence model, and distributes the second initial artificial intelligence model to all the clients; and the clients are used for training the second initial artificial intelligence model according to local real communication data, and obtain corresponding exclusive artificial intelligence models. The server aggregates the received model parameters, and uploads the parameters instead of the original data, so that multi-client knowledge sharing is realized under the premise of protecting data privacy, and the model training efficiency and generalization ability are improved.

The application discloses a process parameter self-adaptive optimization method and system for battery shell cutting, and relates to the field of manufacturing control.The method comprises process parameter-quality mapping modeling based on a physical information deep neural network, adaptive parameter optimization based on a double-agent reinforcement learning, wear compensation, closed-loop continuous learning and knowledge sedimentation.The application embeds shear mechanics theory into a loss function through a physical information neural network to suppress prediction distortion, realizes real-time adjustment through a planning and regulation double-agent architecture, compresses batch adaptation time through meta-learning and transfer learning, and realizes cross-line knowledge sharing under the premise of protecting data privacy through federated learning.The application effectively improves cutting quality consistency, shortens process debugging time, prolongs die life, and enhances system self-adaptive capability.

The present application relates to a kind of in the case where available sample data is scarce, using multi-source transfer learning method is carried out air conditioner energy consumption prediction method, belong to the application field of data mining technology in energy management system.Its method is: data is preprocessed;According to the data for pre-processing multi-source transfer learning model determines loss function;According to the distribution difference between source domain and target domain dynamic adjustment transfer learning model's parameter;Using transfer learning model is carried out air conditioner energy consumption to target domain, obtains energy consumption prediction value.The present application proposes a kind of multi-source transfer learning method for facing the rare sample air conditioner energy consumption prediction, by increasing model level regular function in objective function, realize the knowledge sharing between multi-source domain and improve the transfer precision and generalization performance of model, provide effective solution for the air conditioner system energy consumption prediction under the sample scarce scene.

The application provides a car fault knowledge sharing method and system, and belongs to the technical field of car maintenance. The method comprises a knowledge base construction and updating process and a retrieval process based on the knowledge base. In the knowledge base construction stage, a system administrator defines multiple car professional fields and assigns user permissions, maintains user submission of fault knowledge update requests, and updates the knowledge base after approval by the user for auditing. In the retrieval stage, the terminal user inputs fault description information, the system pre-processes the query information through a semantic retrieval algorithm, vectorizes the representation, calculates the cosine similarity with the fault nodes in the knowledge base, and generates retrieval results in descending order of matching degree. The application realizes the systematic accumulation and dynamic updating of car fault knowledge, provides an efficient and accurate knowledge retrieval channel, constructs a complete knowledge management closed loop, and significantly improves the car maintenance efficiency and fault diagnosis accuracy.

Some aspects of the present technology relate to technologies for context and knowledge sharing abstraction across entities using a relational hierarchical key mapping system. In accordance with some configurations, a first set of one or more data points for a first set of one or more keys is received at a first domain and stored in a schema-less data store. A second set of one or more data points for a second set of one or more keys is received at a second domain and stored in the schema-less data store. Based on at least one shared key, the first set of one or more data points and the second set of one or more data points are linked as a virtual entity such that a request to retrieve data points of the one or more data points is performed as a single seek and a single API call.

This application discloses a multi-agent knowledge sharing method and computing device, relating to the field of artificial intelligence security technology. The method is applied to a computing device capable of coordinating multiple agents to execute tasks. The method includes: receiving a target task, which requires the cooperation of multiple agents; acquiring target knowledge to be shared by a first agent among the multiple agents; the target knowledge being knowledge generated by a preset model included in the first agent executing the target task; acquiring adjustment parameters; determining the target noise intensity corresponding to the target knowledge based on the adjustment parameters and the basic noise intensity; adding noise to the target knowledge based on the target noise intensity, and using the noise-added target knowledge for a second agent among the multiple agents to execute the target task. This allows for a precise balance between the privacy protection strength and the usability of the knowledge.

A method for personalized knowledge sharing and updating among models is proposed. This method decomposes the incremental adapter parameters uploaded by each device into multiple parameter blocks using an edge server. Based on the response strength of each parameter block to the model output and its consistency with the direction of the parameter increment of the target device, a utility score is calculated for each parameter block. Dynamic programming is used to select the optimal subset of parameter blocks, which are then rescaled using a variance matching strategy to generate a merged adapter and distributed to connected devices. Each connected device continuously monitors class increment offset and tag frequency offset, avoids redundant fine-tuning through a caching and reuse mechanism, and dynamically estimates the optimal compression ratio to optimize uploads. This invention enables efficient knowledge sharing among devices under tag frequency offset conditions, significantly reducing communication and computational overhead while maintaining high model accuracy.

This invention relates to the field of power system fault diagnosis and operation and maintenance technology, specifically to an electrical equipment fault diagnosis and early warning system and method based on federated learning and hybrid intelligence. Through a complete closed-loop architecture of "data acquisition—edge computing—federated learning—hybrid intelligent diagnosis—three-level early warning interaction," it achieves full-process coverage from bottom-level multi-source heterogeneous data acquisition to high-level intelligent decision-making and early warning. Specifically, a CNN-LSTM hybrid model is used to simultaneously extract the spatial and temporal features of faults, improving diagnostic accuracy by approximately 15-25% compared to a single model. The federated learning module allows local terminals to transmit encrypted model parameters without sharing raw data, protecting data privacy while enabling cross-regional knowledge sharing. Dynamic weighted aggregation and local model verification mechanisms solve the problem of model performance degradation caused by differences in data distribution across different regions and equipment, and prevent abnormal models from polluting the global model.

The utility model relates to based on punishment mechanism's big match of forbidding drug technology field, and disclose a kind of based on punishment mechanism's big match of forbidding drug device, including main body mechanism and punishment mechanism, the punishment mechanism is fixedly installed in the inside of main body mechanism, the main body mechanism includes show shelf, pedestal, frame, the pedestal is fixedly installed in the show shelf bottom end, the frame is fixedly installed on the show shelf surface.This has community cohesion effect, in community environment, use this device, can effectively condense community resident. Everyone participates in the big match of forbidding drug together, strengthens neighborhood relationship in competition and cooperation. For example, when carrying out activity in community activity center, residents can improve mutual understanding and trust through team competition mode, simultaneously spread the knowledge of forbidding drug, make the whole community form a forbidding drug knowledge sharing atmosphere, enhance the cohesion of community and social stability.

A computer-implemented digital knowledge management system for institutional learning and decision support, comprising the following: A user interaction interface configured to receive structured institutional information from a variety of users connected to a university. This structured institutional information includes strategic issues, institutional goals, performance indicators, academic activity data, and decision-relevant input. an input processing unit that is operationally connected to the user interaction interface and is configured to capture, classify, and convert the received structured institutional information into machine-readable data sets; a collaborative culture assessment unit consisting of at least one processor and a memory in which executable instructions are stored, wherein the collaborative culture assessment unit is configured to collect and process organizational behavior data representing loyalty, engagement, cohesion, well-being and sense of community among members of the institutional leadership; a knowledge generation unit that is operationally linked to the collaborative culture assessment unit, wherein the knowledge generation unit is configured to analyze received knowledge records and generate new institutional knowledge by exploring alternative ideas, identifying new insights, and combining individual and collective institutional contributions; a knowledge exchange unit that is communicatively linked to the knowledge generation unit, wherein the knowledge exchange unit is configured to distribute the generated institutional knowledge to authorized users via controlled communication channels and structured knowledge dissemination procedures; a knowledge application unit designed to transform distributed institutional knowledge into decision support tools, including proposed strategic actions, operational recommendations, and measurable performance targets; an institutional learning repository consisting of a permanent digital storage system configured to store knowledge records, decision outcomes, institutional experiences, and insights gained from implemented decisions; a decision quality assessment processor that is operationally linked to the institutional learning repository and the knowledge application unit and is configured to evaluate generated decisions against predefined institutional criteria, including problem-solving effectiveness, strategic value contribution, achievement of institutional goals, and contribution to organizational learning; and a coordination processor configured to coordinate the operational interaction between the collaborative culture assessment unit, the knowledge generation unit, the knowledge sharing unit, the knowledge application unit, the institutional learning repository, and the decision quality assessment processor, such that the collaborative organizational conditions influence knowledge generation, knowledge distribution, knowledge utilization, and the subsequent evaluation of strategic decisions.

The present invention relates to a system (100) and a method (300) for providing a networking platform for B2B interaction between one or more users. The system (100) integrates a processor (201) and a memory (202) that stores instructions to execute various tasks. The system (100) receives various types of content from the one or more users. Further, the system (100) analyzes one or more user profiles associated with the one or more users to determine the relevancy of the received content based on a user input. By identifying one or more relevant profiles, the platform ensures that content is shared with appropriate users, fostering collaboration and co-creation. The co-created content is then provided to the one or more relevant users, enhancing engagement and knowledge sharing. Overall, the system (100) enables efficient and personalized content delivery, enhancing effective collaboration and co-creation of ideas, product and innovation.

This invention belongs to the field of artificial intelligence communication technology and discloses an automated large-scale model implementation method for knowledge sharing. It focuses on three novel technical perspectives: intelligent deconstruction and reorganization of knowledge content, cross-circle knowledge adaptation and transmission, and closed-loop iteration of knowledge sharing effects. It constructs three core modules: intelligent deconstruction and reorganization of knowledge content, cross-circle knowledge adaptation and transmission, and closed-loop iteration and optimization of knowledge sharing effects, covering the entire process of knowledge sharing from content processing and circle adaptation to effect optimization. Through core algorithms such as multi-dimensional knowledge deconstruction and reorganization optimization, cross-circle knowledge semantic adaptation and form conversion, and multi-dimensional quantification and strategy iteration of knowledge sharing effects, it breaks through the industry bottlenecks of traditional knowledge sharing, namely "rigid content structure, inefficient cross-circle transmission, and lack of closed-loop effect optimization."

The application discloses a mine key facility intelligent visual detection method and system cooperating with unmanned aerial vehicles, and the method comprises the following steps: through the cooperation of multiple unmanned aerial vehicles and a server, a double-layer prompt parameter structure is adopted, and through the hierarchical optimization of a flight environment adaptation prompt layer and a facility defect semantic prompt layer, cross-region detection knowledge sharing is realized under the premise of protecting the privacy of regional inspection data; through real-time visual processing and local model training at the unmanned aerial vehicle end, dynamic defect category discovery and detection capability expansion mechanism are realized through the global optimization of federated learning coordination and model distribution at the server end; the unmanned aerial vehicle and the server are cooperatively optimized, the complex environment of the mine is effectively adapted, the detection precision and operation efficiency of the unmanned aerial vehicle mine inspection are significantly improved, and a complete unmanned solution is provided for the maintenance of mine facilities.

The application belongs to the technical field of artificial intelligence multi-agent cooperation, and discloses a method for ten-thousand-level industry agent cooperation scheduling, parallel execution and self-evolution of a platform, wherein a ten-thousand-level agent cluster covering thousands of industries is constructed, a single master agent receives and issues a unified task instruction, the instruction is automatically decomposed into sub-tasks that can be executed in parallel, ten-thousand industry agents are scheduled to execute tasks, communicate with each other, learn from each other, and based on the task execution results, the agent model is automatically optimized and upgraded. The application solves the technical pain points of the existing multi-agent system, such as insufficient industry coverage, inability to large-scale parallel, difficulty in cross-agent knowledge sharing and inability to self-evolution, realizes the platform-level intelligent cooperation capability of single instruction triggering, ten-thousand-agent concurrency, cross-industry interconnection and lifelong iteration, greatly improves the processing efficiency and intelligence level of complex business scenarios, and is suitable for large-scale operation and management scenarios of digitalization and intelligentization of thousands of industries.

This application provides a training method, apparatus, medium, and device for a PPT generation model. The method is used to jointly train an outline generation model and a PPT generation model. The outline generation model converts a JSON file into a corresponding PPT outline, and the PPT generation model converts the PPT outline into a corresponding JSON file. The method includes: Step 1, reconstructing the original input: The PPT generation model reconstructs the original JSON file based on the PPT outline output by the outline generation model, and simultaneously, the outline generation model reconstructs the original PPT outline based on the JSON file output by the PPT generation model. Training is performed by alternately using generated samples and original samples, and intermediate vectors are passed to optimize the training; Step 2, sharing encoder parameters: The encoders of the PPT generation model and the outline generation model share parameters to achieve knowledge sharing between the two models.