456 results about "Client-side" patented technology

The invention discloses a federated learning method based on quantum transfer learning, relates to the cross technical field of quantum computing and federated learning, and aims to solve the problem of performance bottleneck of traditional federated learning under non-independent identically distributed data. The local model of each client is fused with a classic convolutional layer and a quantum convolutional layer; during local training, firstly, classic features are extracted by a classic convolutional layer, and then the classic features are coded and input into a quantum convolutional layer to generate quantum features; and the two features are spliced and then classified, and model parameters are uploaded. And the server side adopts a FedAvg algorithm to aggregate parameters so as to update the global model. According to the method, a transfer learning mechanism is introduced to optimize the initialization of the quantum layer, and an experimental result shows that the classification performance and robustness of the method are remarkably superior to those of a traditional federated CNN model on a non-IID image data set, and the method is particularly suitable for a privacy protection cooperative computing scene.

The invention provides a communication segmentation learning system and method for adaptive channel compression, and a medium, and relates to the technical field of segmentation learning and communication compression. The system comprises a plurality of clients, a server side and a channel compression device arranged between the clients and the server side, the channel compression device comprises a channel sensitivity modeling module, a rate distortion adaptive compression module and a cross-client fair coordination module. In a channel sensitivity modeling module, channel importance is dynamically evaluated through intermediate layer activation value fusion; differentiated quantization and compression strategies are designed on the basis of sensitivity scores in a rate-distortion self-adaptive compression module, so that key information is reserved while communication overhead is reduced; and meanwhile, the cross-client fair coordination module realizes balanced distribution of communication resources among multiple clients through a fairness regularization and dual optimization mechanism, so that the influence of excessive compression on global convergence is avoided. According to the method, the communication efficiency and the training stability of segmentation learning in a complex heterogeneous environment are remarkably improved.

The invention discloses a WebSocket-based instant message task assigning system. The system comprises two parts, i.e., a server and clients, wherein the server is deployed on an application server and configured with ports for providing external services, and the server is then started; the clients comprises an APP client and a PC client; and information data transmission among the server, the APP client and the PC client is achieved based on the WebSocket protocol, that is, the server provides the instant messaging service, and the clients achieve acquisition of data from the server. Compared with the prior art, the system provided by the invention has the advantages that WebSocket-based instant message task assigning is achieved, and information can be synchronized between the information acquisition clients, so that high timeliness, effectiveness and safety of the information are ensured.

The present application relates to the field of federated learning and privacy protection, and particularly relates to a direction-aware local differential privacy federated learning method. The present application effectively solves the problem of serious decline in model performance caused by the destruction of gradient direction information by traditional differential privacy methods by adopting a direction-aware noise selection strategy based on an exponential mechanism locally on the client side. Specifically, the client first performs L2 norm clipping on the gradient to control the sensitivity, then generates K candidate noise vectors from a Gaussian distribution, evaluates the direction consistency of each candidate noise and the true gradient through cosine similarity, and selects the noise closest to the gradient direction in a probabilistic manner using the exponential mechanism, and finally adds the selected noise to the gradient to complete the perturbation. The method provides a strict epsilon-local differential privacy mathematical proof for the entire noise selection process, reduces the calculation time by about 80% compared to existing direction-aware methods, and achieves a triple balance of privacy protection, model utility and computational efficiency.

The invention discloses a multipath load control acquisition device and system and a load control acquisition method, the device adopts an architecture in which a master device and a slave device cooperate, the master device comprises a master controller, a master communication module, a user display interface and a master interface unit, and the slave device comprises a slave interface unit, a slave controller, an acquisition unit and a slave communication module. Flexible marshalling management is achieved by distributing electronic identity numbers to the breakers, load parameters and the states of the breakers are synchronously collected through the collection unit, a two-way data transmission channel is constructed by means of the master-slave communication module, a control signal is generated based on preset logic and real-time data through the master controller, and the state of the breakers is controlled. And the slave controller issues an instruction to drive the circuit breaker electric operation mechanism to realize millisecond-level rapid switching-on and switching-off operation. According to the invention, the problems of insufficient number of loops, lack of state monitoring, slow response and the like of an existing load control system are solved, and accurate, rapid and intelligent management of multi-path loads on a client side is realized.

Systems and methods herein are provided for delivering personalized trivia content in an interactive game using tangible game elements and a digital companion application. Players are enabled to select and scan multiple tangible game elements (e.g., via quick-response codes, ArUco markers, tags, or other identifiers), each representing a game parameter value (e.g., a topic or category), through a client-side user interface of the digital application to create different combinations. The user interface is coupled to a backend host that controls communications between the user interface and a database storing cache records including game content for particular combinations. The backend host queries the database to check for available game content for the selected combination (subject to certain constraints, such as game content not already previously presented to a user). If not, the backend host transmits a query to a generative artificial intelligence model trained to generate new game content.

The invention provides a SaaS service security verification method and system under a zero-trust architecture. The method comprises the following steps: a client sends a login request to a server; the server side verifies the login request and returns a dynamic key, a coding rule table and a shuffling rule to the client side based on the login request and a current time window; the client generates a first password table corresponding to the dynamic key, hashing data of the service request according to a coding rule table, embedding the data into the first password table, performing shuffling processing of various different schemes on the first password table according to a shuffling rule to obtain a client request signature, and sending the service request and the client request signature to the server; and the server side obtains the server side request signature according to the same logic, verifies the client side request signature based on the server side request signature, and returns a response result corresponding to the service request to the client side after the verification is passed. According to the method, the end-to-end data integrity and security in the communication process of the client and the server can be remarkably improved.

The invention discloses a real-time software deformation interaction method based on a physical engine in an environment. The real-time software deformation interaction method comprises the following steps: S1, establishing a virtual environment and a software model; s2, constructing an interactive perception and collision detection network; s3, receiving and analyzing a multi-modal interaction instruction; s4, real-time deformation calculation based on a physical engine; s5, performing multi-user collaborative state synchronization and rendering feedback; and S6, carrying out fault state simulation on the software object. According to the method, real-time and high-precision simulation of software deformation is realized through deep fusion of software dynamics calculation of a physical engine and an interaction instruction stream, and through hierarchical physical attribute binding and client prediction, the network synchronization data volume is greatly reduced while the millimeter-level deformation precision is ensured, so that the network synchronization efficiency is improved. And low-delay and high-consistency experience under multi-person cooperative operation is ensured.

The invention discloses a power distribution line passive real-time fault positioning system, which belongs to the technical field of power distribution line fault detection and comprises a power distribution line fault positioning device management system and a plurality of fault positioning terminals distributed along a power distribution line. The distribution line fault positioning device management system establishes two-way communication with each fault positioning terminal through a 4G / 5G wireless public network; the distribution line fault positioning device management system comprises a data center, a client and a plurality of mobile terminals. The fault positioning terminal comprises an ARM processor, a time synchronization unit, a signal acquisition unit, a signal processing unit, a passive self-powered unit, a communication unit and an auxiliary unit. According to the invention, the distributed fault positioning terminal is combined with an intelligent algorithm, so that high-precision fault recognition and real-time positioning under a complex network are realized, and the fault handling efficiency of the power distribution network is remarkably improved.

The invention provides a KEM-based anti-quantum TLCP protocol design method and system, and belongs to the technical field of information security. The method comprises the steps that a server side applies for a server side signature certificate and an encryption certificate from an issuing mechanism, a long-term server side signature public key is stored in the server side signature certificate, and a long-term server side encryption public key is stored in the encryption certificate; and in the TLCP handshake protocol, identity authentication and key exchange between the server and the client are realized based on the long-term server signature public key and the long-term server encryption public key. According to the method, the anti-quantum KEM is used for replacing signature and encrypted primitives to realize key exchange and identity authentication functions of the protocol, identity authentication does not depend on the signature of a message any more, and meanwhile, a client introduces a temporary public key to replace a long-term static public key to participate in a KEM key negotiation process, so that the forward security of the protocol is ensured, and the security of the protocol is improved. Therefore, the efficiency and the security of the TLCP protocol are improved.

The invention discloses a dynamic space-time diagram traffic flow prediction method based on federated learning and a related device, belongs to the technical field of traffic flow prediction, is applied to a client under a federated learning framework, and comprises the following steps: the client generates a dynamic adjacency matrix through a forgetting mechanism based on historical and current traffic data, and extracts node space features; mapping the node features to a unified semantic space through a shared semantic mapping function, aggregating to generate a client prototype, and uploading the client prototype to a server; and receiving cross-client aggregation features generated by the server according to the client prototype, updating local node features, and performing time sequence modeling to output a prediction result. According to the method, the prototype representation of the client level is introduced under the federated learning framework, collaborative optimization of explicit modeling and time sequence prediction of cross-client space-time dependence is realized, and the accuracy of traffic flow prediction in a distributed traffic scene is remarkably improved.

The invention discloses an internet of things heterogeneous equipment federated learning method based on cross-modal knowledge distillation, belongs to the field of internet of things, and proposes a two-stage federated learning framework, introduces a knowledge distillation mechanism at a client side, uniformly distills local models with different structures and parameter scales into isomorphic models, and finally obtains the isomorphic models. And uploading the distilled model parameters to a server for aggregation. On the premise that original data and local model privacy are not leaked, the method realizes effective support for model heterogeneity, so that feasibility and stability of a single-round federal learning process are guaranteed. According to the method, an auxiliary knowledge distillation strategy based on the CLIP is further provided, and text-level semantic supervision information generated by a pre-trained CLIP model is utilized to guide a client image classification model to perform knowledge distillation and feature alignment, so that the generalization ability and the knowledge migration effect of the model are enhanced. According to the method, the effectiveness of knowledge distillation is remarkably improved in a non-independent identically distributed and long-tail data environment, and the overall performance of single-round federated learning is improved.

The invention relates to a longitudinal federal learning method and device based on feature importance and electronic equipment, and aims to realize a local privacy budget generator of a client through a lightweight neural network by taking local feature importance as drive. Therefore, a cooperative closed loop from'feature-level privacy budget 'to'embedded dimension disturbance' to'suggested privacy budget uploading 'to'server-side gradient noise addition' is realized, and meanwhile, discrete grading and suggested privacy budget disturbance mechanisms are introduced to prevent the server-side from reversely deducing privacy contents, so that the differential privacy security and practicability of the whole system are improved.

Decentralized hast table techniques are described. In an implementation, a hash table input is received including a decentralized identifier and index data configured to locate the decentralized identifier. A hash table entry is generated for inclusion in a decentralized hash table in compliance with a hash table schema. A decentralized identifier search query is received from a client device. The decentralized hash table is searched based on the index data using the decentralized identifier search query. A decentralized search result is generated that includes the decentralized identifier for receipt by the client device responsive to the searching.

The invention discloses an auditing method and device during function execution in reproductive medical service. A server side method comprises the steps that an information operation request sent by a client side is received; the information operation request carries a globally unique identifier and auditing parameters of the patient; checking whether the information operation request has an operation authority or not; when the information operation request has the operation authority and the parameter state of the auditing parameter is to-be-stored, performing data desensitization on original medical information in the auditing parameter to obtain to-be-stored medical information; constructing a first operation log based on a data operation function expression and a service attribute in the audit parameter; and storing the to-be-stored medical information to the medical data table by adopting the global unique identifier, and storing the first operation log to the audit data table. By adopting the embodiment of the invention, the security of data transmission and storage can be improved. Meanwhile, the operation behavior of each link is accurately recorded and monitored through a refined management mode, and the transparency and traceability of data are improved.

The invention discloses a data processing method and a related device, and relates to the technical field of message queues, a target client requests to carry out target operation on business data of a target theme, a server verifies whether a permission mapping relation matched with a permission authentication request exists in first-class permission data, and if yes, the first-class permission data is sent to the server; the server searches target message queue cluster information matched with the target theme in second-class authority data and sends the target message queue cluster information to the target client, the second-class authority data comprises a plurality of routing relationships, and the routing relationships comprise corresponding relationships between the theme and each message queue cluster in the message queue system; and the target client side executes the target operation in the target message queue cluster according to the target message queue cluster information, so that the flexibility of accessing multiple clusters by each client side is improved through centralized maintenance of authority data by the server side.

The invention discloses a teaching large model knowledge graph updating method and system based on federated learning. The method comprises the following steps: training a local problem solving strategy model reflecting a specific teaching strategy by each client, and generating structured learning effect data recording an application effect of the local problem solving strategy model; the central server performs functional clustering on the models by analyzing behavior characteristics of the local problem solving strategy models when the local problem solving strategy models process the preset probe problems so as to identify and divide client clusters representing different teaching genres; and finally, based on a clustering result and the learning effect data, performing weighted updating on a multi-dimensional logic knowledge graph, and storing objective effects of each teaching genre for different student portraits into the graph. According to the method, the diversity of teaching strategies can be effectively identified and reserved, and the quality of a knowledge graph on which a large teaching model depends and the accuracy of personalized teaching recommendation are remarkably improved through intelligent fusion based on objective effects.

The invention provides a large-scale AI model-as-a-service-oriented security reasoning and protection method. The method comprises the following steps that S101, a user initiates an encryption request; s102, the server side carries out verification; s103, decrypting the input data; s104, AI model reasoning is carried out; s105, performing output encryption; and S106, the client decrypts and verifies. The security of the sensitive content can be protected.

The application discloses an individualized federated learning method based on evolutionary clustering, comprising the following steps: distributing the initialized global model parameters to all clients to obtain corresponding local model parameters; obtaining an original clustering cluster set and individualized model parameters of each clustering cluster through original clustering; then performing iterative training of federated learning until a preset iteration termination condition is met, and finally outputting the final clustering cluster set and individualized model parameters of each clustering cluster to realize individualized federated learning based on evolutionary clustering; wherein, in each training of federated learning, the merging score based on snapshot quality and clustering stability between clustering cluster pairs is clustered. The method clusters by introducing the merging score based on snapshot quality and clustering stability, reduces the clustering instability caused by short-term fluctuations, ensures that the clustering structure is more stable in the time dimension, and improves the performance and convergence speed of the individualized model.

A graphical, hierarchical document stream browser and environment for semantic (e.g. framing) and performance data analysis and interactive visualization integrates three scales: entities (competitive), entity (diachronic), and document (linguistic). The document level includes annotation and computational linguistics facilities; the entity level has calendrical and time-series focus. All levels emphasize deep linkage and network (i.e. connective / relational space) view of objects, with user-configurable connectivity. Large language model (LLM) integrations provide synthetic advisories, public opinions, reports, plot insights, comparisons; traditional natural language processing techniques and neural models are also employed. A smart plot system includes a “plot cart” and interpreter with an analysis snippet library. Graph structure may arise via adjustable blending or perceptual optimization of canned attribute-related distance functions or via link-induction query language with deep “semantic stored procedure” subexpressions, or feed into graph neural network-style inference for predictions. Most non-LLM ongoing computational load is client-side, using precomputed hierarchical summary files.

A controller maps a plurality of client-side traffic ports to a plurality of line-side optical ports of a wavelength division multiplexing system, that are operatively coupled to at least one optical long haul link. The controller determines idle traffic overhead associated with at least one mapped client-side traffic port of the plurality of client-side traffic ports. The controller determines that there is underutilized bandwidth capacity in at least a first line-side optical port in the plurality of mapped line-side ports that would reduce a need for use of at least a second line-side optical port. The controller logically remaps at least one client-side traffic port from the second line-side port to the underutilized first line-side optical port in response to determining that there is underutilized bandwidth capacity of the first line-side optical port.

The invention provides a credible traceable privacy enhanced heterogeneous federal learning method and system, and belongs to the technical field of artificial intelligence and information security, the method comprises the following steps: initializing a watermark parameter of a pseudo-random code at a server side, and generating a coding key and a decoding key; the client performs coding and tensor remodeling on the message bit sequence of the metadata through the coding key to generate a potential tensor with a watermark; performing reverse diffusion iteration on the potential tensor with the watermark through a pre-trained diffusion model to obtain a generation feature; and the server side performs multi-layer auditing on the client side according to the generation characteristics and the decoding key, and detects whether the client side has malicious behaviors or not. According to the method, the pseudo-random code watermark technology is applied to federated learning for the first time, an unforgeable watermark traceability mechanism is established, malicious behaviors of the client can be effectively monitored, and the method can be widely applied to heterogeneous federated learning scenes.

The invention discloses an intelligent driving scene-oriented lightweight parameter consensus aggregation method and system, and the method comprises the steps: decomposing a weight matrix of a pre-trained large language model to obtain a first matrix and a second matrix, freezing the first matrix, setting the initial value of the second matrix as zero, and generating a consensus parameter increment and a consensus selection matrix thereof; each vehicle-mounted client receives the consensus parameter increment and the consensus selection matrix returned by the roadside server, and the consensus parameter increment and the consensus selection matrix are fused with the dense matrix which is not screened in the previous round to obtain a mixed parameter; each vehicle-mounted client generates a sparse local parameter increment and a mask selection matrix of a corresponding vehicle end mark by using the mixed parameters, and uploads the sparse local parameter increment and the mask selection matrix to a roadside server; and the roadside server receives the uploaded sparse local parameter increment and the mask selection matrix marked by the corresponding vehicle end, calculates a consensus area, and averages the selected consensus parameter increment in the consensus area to generate a new consensus parameter increment.

The invention discloses a federal invariant graph learning method and system for heterogeneous graph data of an unmanned aerial vehicle. The method comprises the following steps: acquiring heterogeneous graph data acquired by an unmanned aerial vehicle; the method comprises the following steps: splitting a local graph structure, and constructing a double-branch representation structure consisting of an invariant branch and a variable branch; processing a local heterogeneous graph based on a sub-graph generation strategy; extracting structure invariant features and local variable features by using a coding module; and stable aggregation of the global model is realized at the server side through double-gradient regularization. The system comprises an unmanned aerial vehicle client module, a server aggregation module, a model construction module, a structure representation module, a sub-graph processing module, a coding module and a training control module, and all the modules cooperate to complete federal invariant graph learning. According to the method, the difference and consistency of the unmanned aerial vehicle data are considered without uploading the original data, the problem of non-independent identical distribution is effectively relieved, and the stability and generalization ability of the model are improved.

The invention discloses a budget allocation strategy generation method and device, equipment and a storage medium, and the method comprises the steps: carrying out the disassembly of a total budget amount according to a budget period and each time period, and obtaining a budget disassembly result; performing target disassembly according to the delivery target type and the industry to which the enterprise belongs to obtain a target disassembly result; combining the budget disassembling result and the target disassembling result to obtain a preliminary budget allocation framework; embedding historical delivery data of the same industry and the optimal practice of the industry into a rule engine; and based on the preliminary budget allocation framework, the static basic data and the dynamic real-time data, generating at least two types of alternative budget allocation strategies by adopting a mode of fusing a machine learning model and an embedded rule engine. According to the method, the client side data, the platform side data and the industry side data are systematically integrated, and the distribution strategy is generated through fusion of the rule engine and the machine learning model, so that adaptive fusion and real-time optimization of the strategy are realized, and the generation efficiency is remarkably improved.

Network elements and methods of use, including a transmitter comprising a client-side input, signal and clock conditioning blocks, a modulation block, and antennas. The client-side input receives baseband signals having client data. The signal conditioning block adjusts signal characteristics of the baseband signals to generate intermediate signals. The clock conditioning block receives a first clock signal having a first clock frequency and adjusts signal characteristics of the first clock signal to generate a second clock signal having a harmonic frequency of the first clock frequency. The modulation block modulates the intermediate signals onto the second clock signal to generate antenna feed signals. The antennas generate radiated signals based on the antenna feed signals and couple the radiated signals into hollow waveguides. The radiated signals are radiated electromagnetic waves configured for coherent detection with a transmission frequency in a range between 300 Gigahertz (GHz) and 10 Terahertz (THz).

A graphical, hierarchical document stream browser and environment for semantic (e.g. framing) and performance data analysis and interactive visualization integrates three scales: entities (competitive), entity (diachronic), and document (linguistic). The document level includes annotation and computational linguistics facilities; the entity level has calendrical and time-series focus. All levels emphasize deep linkage and network (i.e. connective / relational space) view of objects, with user-configurable connectivity. Large language model (LLM) integrations provide synthetic advisories, public opinions, reports, plot insights, comparisons; traditional natural language processing techniques and neural models are also employed. A smart plot system includes a “plot cart” and interpreter with an analysis snippet library. Graph structure may arise via adjustable blending or perceptual optimization of canned attribute-related distance functions or via link-induction query language with deep “semantic stored procedure” subexpressions, or feed into graph neural network-style inference for predictions. Most non-LLM ongoing computational load is client-side, using precomputed hierarchical summary files.

The invention relates to the technical field of network communication, in particular to an agent interaction method based on a chain of responsibility and multi-terminal cooperative security, and the method comprises the steps: building persistent connection with a client, receiving an interaction request from the client, and determining a session identifier according to the interaction request; determining a routing strategy according to the session identifier, and determining a target server node based on the routing strategy and the interaction request; based on the target server node, processing the interaction request to obtain a processing result; and performing protocol adaptation on the processing result, determining an adaptation result of a communication protocol with the client, and pushing the adaptation result to the client. The method has the advantages that the experience feeling of interaction of a user using the intelligent agent can be improved, and meanwhile the processing efficiency and safety of the interaction system can be improved.

This utility model discloses a contact resistance testing device, comprising: a tooling platform, a clamping and pressure testing mechanism, and a resistance testing mechanism. A product placement rack is provided above the tooling platform, and the product placement rack has at least one product placement position. The clamping and pressure testing mechanism is located on the tooling platform and cooperates with the product placement rack. The contact resistance testing device of this utility model clamps the connection between two products using the clamping and pressure testing mechanism, simulating the clamping process during assembly at the customer's location. This allows for the simulation of the contact resistance after assembly, enabling timely detection of problems. Once a problem is detected, it can be addressed promptly, avoiding significant screening costs at the customer's location.

The invention provides an investment supervision index monitoring method and device and an investment supervision system, and the method comprises the steps: obtaining a contract file of a newly-added investment product, calling a plurality of corresponding preset supervision indexes from a supervision index library according to the product type of the newly-added investment product, and obtaining a plurality of index description texts; performing text extraction on the contract file through an OCR (Optical Character Recognition) technology to obtain a contract text; calculating the similarity between the contract text and each index description text to obtain a plurality of similarity scores; and sorting the plurality of similarity scores according to a descending order to obtain an index sequence, intercepting the index sequence according to a preset step length to obtain a recommendation index, monitoring the change of the recommendation index, and sending the recommendation index to the client. According to the method, the problem of supervision deviation caused by difficulty in guaranteeing coverage of indexes on products in a multi-feature complex scene because supervision index screening depends on man-machine combination in the prior art is solved.