2383 results about "Server-side" patented technology

The invention relates to the technical field of artificial intelligence, can be applied to business scenes such as financial science and technology and medical health, and discloses a federal collaborative optimization method, device and equipment and a medium. Each participant terminal executes instruction tuning training and value alignment training based on the initial global model, generates model parameters and uploads the model parameters, and the server side aggregates the received model parameters, generates an iterative global model and repeats the training process until a preset performance index is reached. According to the method, instruction tuning and value alignment training are carried out in parallel at the terminal side of the participant, and the optimization result is fused at the server side, so that the adaptability and training efficiency of the model to multi-source data are improved, the problems that in the prior art, the optimization target is single, and aggregation strategies are lack of collaboration are solved, and the practicability and generalization ability of the model are enhanced.

A method includes receiving, at a server from a user device, a user prompt segment to a large language model (LLM), obtaining an additional prompt segment from a prompt data source, identifying a electronic address in the prompt segment, replacing the electronic address with a placeholder to generate a updated prompt segment, generating a LLM prompt comprising the updated prompt segment and the user prompt segment, and sending the LLM prompt to the LLM. The method further includes receiving a response to the LLM prompt from the LLM, the response comprising the placeholder, replacing the placeholder with the electronic address to generate an updated response, and sending the updated response to the user device.

The invention discloses a user-noninductive cold and hot sensing remote memory exchange system, which belongs to the field of computer storage, and is characterized in that cluster nodes are divided into extensible remote memory service nodes and computing client nodes according to roles; the remote memory node comprises a remote memory exchange server and a memory resource registration module; the computing client node comprises a FrontSwap-based remote memory exchange client, and a remote memory exchange server side is registered as a memory resource which is non-inductive to a user; furthermore, popularity statistics based on a popularity histogram is realized in a kernel mode; the memory exchange behavior is guided through the dynamic decision-making module based on online reinforcement learning, the memory management efficiency and the system performance are remarkably improved, and efficient and non-inductive memory expansion service is provided for user application.

Systems and methods are presented herein for requesting a version of media content from a server that includes haptic feedback rending criteria compatible with the haptics capabilities of a client device. At a server, a request is received for a media asset for interaction on a haptic enabled client device, wherein the media asset comprises haptic feedback rendering criteria. Based on the request, haptic feedback capabilities of the haptic enabled client device associated with the request is determined. The haptic feedback capabilities of the haptic enabled client device are compared to the haptic feedback rendering criteria of one or more versions of the media asset available via the server. In response to the comparing, a version of the media asset comprising haptic feedback rendering criteria compatible with the haptic feedback capabilities of the haptic enabled client device is transmitted from the server to the haptic enabled client device.

The invention provides a five-axis series-parallel numerical control machine tool machining process virtual monitoring simulation method and system based on digital twinning, and the method comprises the steps: building an information interaction mechanism between a physical machine tool and a digital twinning model, enabling a server side to collect sensor data, and transmitting the real-time data to a client side through a communication protocol; the client processes the data and drives the digital twin model. And aiming at potential errors of twin system forecast, tool setting error compensation can be carried out in real time, so that the machining precision is improved. Meanwhile, the system supports an offline simulation function, combines an inverse kinematics algorithm and an NC code, fuses a material removal model, simulates a machining process, evaluates machining performance and errors, and provides an optimization basis for machining strategies of workpieces with different geometrical shapes and materials. According to the method, bidirectional interaction between the physical entity and the virtual model is achieved, the real-time monitoring and visualization capability of the machining process is enhanced, a reliable simulation evaluation and optimization means is provided for machining of complex parts, and the method has wide industrial application prospects.

The invention provides a personalized federal learning method and system for a shared model, and the method comprises the steps: enabling a server to store and initialize the shared model, a global model and a global class prototype of a client, and enabling the client to initialize a local learning weight vector; the server sends sharing models of other clients to one client, and the client sets a local sharing model as a global model and trains the global model, and uploads the sharing model and a local class prototype to the server; and the server calculates and obtains a global class prototype and a global model according to all the received shared models and local class prototypes. According to the method, the knowledge sharing problem in federal learning is solved, and the performance of a personalized model is improved; the problem of offset in local model training of the client is solved, and the obtained personalized prototype contains more abundant global information than the personalized prototype; while state dependence and communication overhead are reduced, effective fusion of global knowledge and local characteristics is realized so as to improve robustness and adaptability of the model in a complex scene.

The invention relates to a large model data protection method and device based on a trusted environment, equipment and a medium. In the scheme, after a server side obtains original large model data, the original large model data needs to be encrypted and signed, and then the original large model data and an encrypted information table are issued to a target application side; when the target application end calls the large model data, performing exception check according to a credible strategy; if the check is passed, performing signature verification on each encrypted data slice; if the signature verification succeeds, decrypting the encrypted slice data by using the encrypted information table to obtain original large model data; wherein the server side and the target application side are both in a trusted environment. It can be seen that the trusted environment and the key system are combined, and a dual protection mechanism for large model data is achieved; moreover, before the large model data is called, the operation environment of the target application end needs to be subjected to exception check, so that an attacker is prevented from acquiring or tampering the sensitive data, and all-around data protection is provided for the large model data.

Disclosed in the present invention are a large language model federated fine-tuning method and apparatus based on gradient compression. The method comprises the following steps: constructing, on the basis of a gradient tensor generated during fine tuning of a large language model, a raw data set having a time series relationship, performing inference by means of an autoencoder to obtain a reconstructed gradient data set, and constructing a reconstruction loss function to optimize the autoencoder; and initializing a base model of the large language model as a global model at a server end, the server end updating the global model to a client, using a pre-trained encoder to obtain a compressed gradient at the client, and at the server end, using a pre-trained decoder to decode and aggregate the compressed gradient, and then updating the global model. The present invention can improve the fine-tuning efficiency of the large language model and reduce computing resource requirements while ensuring data privacy protection, and is suitable for application scenarios such as communication optimization improvement and privacy protection enhancement in the process of scientific computing-oriented large model fine-tuning and training.

The invention provides a multi-mode medical record intelligent generation and interaction system and method based on a wearable terminal and an MOE architecture, and relates to the technical field of AI large models. The terminal comprises a wearable terminal used for carrying out real-time acquisition and interaction on multi-modal data in a medical scene to obtain multi-modal medical data; the communication network is used for realizing data interaction between the wearable terminal and the server-side AI large model processing system; the server side AI large model processing system is used for performing semantic analysis, structured extraction and medical record document generation based on the multi-modal medical data to obtain an editable first draft of a medical record document; and the wearable terminal is also used for receiving and presenting the first draft of the medical record document returned by the server side AI large model processing system, and editing and confirming the first draft of the medical record document to form a returned draft of the medical record document. The problem that in the prior art, a terminal lacks multi-modal real-time fusion and LLM closed loop, and doctor duplication interaction cannot be achieved is solved.

The invention discloses an Internet of Vehicles federated learning excitation method and system based on a Nash game and a coalition game, and the method comprises the steps: initializing a vehicle client set in an Internet of Vehicles environment, and enabling a server to broadcast global model parameters to all clients; a two-stage Nash game model is established, the first stage is used for determining an alliance set in the vehicle client set, and the second stage is used for bargaining to optimize resource contribution; negotiating and determining an optimal data contribution ratio and an optimal resource contribution ratio based on the server side and the alliance set by using the two-stage Nash game model; and the server allocates rewards to the alliance set according to the optimal data contribution proportion and the optimal resource contribution proportion. The method can adapt to high dynamics of the Internet of Vehicles, and fair and efficient resource allocation is realized.

The invention discloses a dynamic scheduling optimization method for a low-altitude logistics distribution network, and the method comprises the steps: a server side builds a multi-source sensing network through satellite remote sensing, an unmanned plane airborne sensor and ground traffic monitoring, fuses meteorological data, airspace control data and order distribution data which are collected in real time, and generates a four-dimensional space-time grid map; based on the four-dimensional space-time grid map, the server side adopts a TD3-GA hybrid intelligent algorithm to carry out path planning of the logistics distribution network; the edge calculation end generates an optimal scheduling scheme of the unmanned aerial vehicle group through a multi-objective optimization function based on the global path; and the server side performs security risk assessment on the optimal scheduling scheme by using a Bayesian network model, and dynamically adjusts a space-time routing strategy of the unmanned aerial vehicle cluster according to an assessment result. According to the method, in a large-scale unmanned aerial vehicle concurrent scheduling scene, the scheduling efficiency can be effectively improved, the response time delay is reduced, the risk prediction accuracy is improved, and the timeliness and safety of a low-altitude distribution network are remarkably improved.

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 discloses a federal large model knowledge collaborative training method supporting multi-modal heterogeneous clients, which comprises the following steps: each client receives a model initialization parameter issued by a central server, and applies adaptive differential privacy noise to independently train a heterogeneous lightweight model based on local private data; updating the model to which the noise is applied and uploading a modal identifier of the model to a central server side; after model updating and modal identification of each client are received, based on a modal perception weighted consensus fusion mechanism, knowledge of each client is fused to update a global large model; and the central server side issues the updated presentation layer parameters of the global large model to the client side for initialization of the next round of local training. According to the method, on the premise that a public data set or specific task setting is not needed, comprehensive compatibility of data isomerism, client dynamic participation, model diversity and privacy protection requirements is achieved, and the adaptability, stability and knowledge utilization efficiency of large model federation training are remarkably improved.

The invention discloses a method and a system for realizing remote ADB debugging of Android equipment based on Frpc, and belongs to the technical field of remote debugging of mobile equipment. Aiming at the problems of complex configuration, low security, poor network penetration capability and the like of a traditional scheme, a system architecture including dynamic configuration management, bidirectional security authentication and intelligent tunnel maintenance is constructed by embedding an Frpc client in an equipment end. An encryption tunnel is established between an equipment end and a server end through TLS1.3, ADB flow reverse proxy transmission is achieved, configuration safety is guaranteed by adopting a dynamic token and HMAC-SHA256 encryption, and hot loading update and certificate two-way verification are supported; and an exponential backoff reconnection strategy is designed, and the network outage recovery capability is improved. According to the scheme, the configuration complexity is reduced by 83% in a public network environment, the connection speed is increased by 5 times, the data leakage risk is reduced by 97%, seamless debugging across NAT networks is supported, and a safe and efficient solution is provided for remote maintenance of Internet of Things equipment and intelligent terminals.

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 cross-institution financial data federated learning modeling system and a privacy compliance verification method, and relates to the technical field of financial data processing, the system comprises a federated aggregation module used for a server to align financial time sequence characteristics of each institution through a timestamp hash alignment strategy, and a privacy verification module used for verifying the privacy compliance of each institution; aggregating the aligned financial time sequence characteristics by using a federated time convolutional network, and generating time sequence model parameters for capturing cross-mechanism global time sequence dependence; the federation reasoning module is used for generating a minimum spanning tree representation of a global enterprise association relationship based on a security multi-party computing strategy by the server side on the basis of a local sub-graph of the enterprise association relationship constructed by each institution client side; the server performs cross-mechanism graph reasoning by using a federated graph attention network and an aggregation graph embedding vector and combining minimum spanning tree representation, and constructs graph model parameters for identifying enterprise associated risks; and the model joint training module is used for the server to fuse the time sequence model parameters and the graph model parameters so as to determine global model parameters.

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 invention discloses a three-dimensional scene reconstruction and monitoring method based on multi-view fusion and deep learning. The method comprises the following steps: acquiring image data from different viewpoints through a plurality of cameras, and carrying out geometric calibration; denoising, correcting and enhancing the image; constructing a multi-scale convolutional neural network and a variational auto-encoder model, and extracting multi-level features; carrying out weighted fusion on the features, and carrying out three-dimensional reconstruction through sparse coding and a graph neural network; performing feature optimization by applying a dynamic graph convolutional network and a double attention mechanism, and performing model updating based on adversarial gradient descent; anomaly detection is carried out through multi-scale analysis and an adversarial variational auto-encoder, preliminary processing is carried out at a camera end by adopting an edge computing technology, and further analysis is carried out at a central server end through a heterogeneous graph neural network and sparse subspace clustering. According to the method, high-precision and intelligent three-dimensional scene reconstruction and monitoring are realized, and the method has a wide application prospect.

The invention provides a configuration parameter processing method, electronic equipment and a storage medium, and relates to the field of distributed technologies. The method is applied to a configuration center, and comprises the following steps: receiving key packaging information, a configuration parameter ciphertext and storage request information sent by an application server; randomly selecting an unused target unique identifier according to the storage request information, and inputting the target unique identifier, the key packaging information and the configuration parameter ciphertext into a trusted execution environment for processing to obtain an unpackaging result; determining a storage key according to the deblocking result; when it is detected that the hash value to be verified is equal to the encrypted hash value, an identifier ciphertext and a backup ciphertext of the target unique identifier are obtained through calculation, and the identifier ciphertext is sent to the application server side; and storing the identification hash value and the configuration parameter ciphertext into a parameter database, and storing the backup ciphertext and the pre-stored application identification code into a backup database. According to the method, the storage security of the configuration parameters is improved.

The invention relates to a base station communication method and system based on a Websocket protocol, an electronic device and a storage medium, and is applied to the field of wireless communication.The base station communication method based on the Websocket protocol comprises the steps that in response to a base station service process data request sent by a client side based on a browser, in a single communication thread corresponding to each client side, a single communication thread corresponding to each client side is obtained; sending a request for establishing a second heartbeat detection connection to the independent service process end; and the client is used for establishing a first heartbeat detection long connection with the server based on a Websocket protocol. And receiving base station service process data sent by the independent service process end under the condition that a second heartbeat detection long connection is established with the independent service process end based on a Websocket protocol. And forwarding the received base station service process data to the client based on the browser. According to the invention, the real-time performance and stability of base station communication are improved.

The invention provides a component configuration method which can be applied to the technical field of cloud computing and the technical field of big data. The method comprises the steps of obtaining configuration information used for configuring a target interface component from a server side based on an interface request; packaging the interface request into an asynchronous data acquisition object, and storing the asynchronous data acquisition object as a global shared instance; registering a callback processing function based on the global shared instance for the target interface component; and responding to the configuration information returned by the server side and triggering the callback processing function, and executing component rendering on the target interface component based on the configuration information.

The invention discloses a violation snapshot method based on image recognition, which relates to the related technical field of traffic video monitoring and comprises the steps of environment perception, scene analysis, scene feature library establishment, multi-modal data acquisition, server model aggregation, model issuing and fine adjustment and violation behavior recognition. The invention further discloses a violation snapshot system based on image recognition. The violation snapshot system comprises an environment sensing module, a data acquisition module, a data processing center, a violation behavior recognition module and a snapshot and recording module. Different from a fixed snapshot strategy of a traditional method, the system can dynamically adjust the snapshot strategy according to real-time understanding of a traffic scene, the system can automatically improve the snapshot frame rate when detecting a high-risk scene in which a traffic accident is about to occur, and the snapshot frame rate can be automatically increased in a road section in which the traffic flow is small and the scene is simple. Snapshot resource consumption can be properly reduced; therefore, a complex and changeable traffic environment can be better dealt with, and the snapshot effectiveness and the resource utilization efficiency are improved.

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

The invention relates to the technical field of driver monitoring, in particular to a dangerous driving behavior detection method based on heterogeneous federal ensemble learning. The client extracts a feature vector of the driving behavior data, and maps the feature vector into a low-dimensional category prototype by using a prototype adapter network; a local dynamic differential privacy technology is adopted to carry out noise injection encryption processing on parameters of the prototype adapter and then upload the parameters to a server; the server performs cross-client prototype alignment and aggregation to generate a global prototype; the client downloads global prototype parameters and updates a local model, incremental learning is carried out based on a dynamic prototype library, and when a novel dangerous driving behavior is detected, a newly-added prototype and a historical prototype are separately stored; and the client performs similarity matching with the global prototype according to the driving behavior characteristics acquired in real time to obtain a detection result. According to the method, an extensible and high-reliability federal learning framework is provided for landing of an intelligent automobile safety system from three dimensions of heterogeneous compatibility, privacy-efficiency balance and personalized dynamic updating.

The invention discloses a task scheduling method, electronic equipment and a storage medium. The method is applied to a configuration center side and comprises the following steps: acquiring a configuration instruction subjected to permission verification; acquiring multi-dimensional service scene data based on the configuration instruction, and inputting the multi-dimensional service scene data into a preset adaptive strategy model to generate a corresponding target adaptive strategy; the preset adaptive strategy model is obtained based on rule learning and transfer learning training; and performing visual simulation verification on the target adaptive strategy, and issuing the target adaptive strategy passing the verification to a server side. According to the scheme, the target self-adaptive strategy adaptive to the current service scene is generated through the multi-dimensional service scene data collected in real time and the preset self-adaptive strategy model, and the target self-adaptive strategy passing the visual simulation verification is issued to the server side for scheduling execution, so that the optimal task scheduling in the dynamic service scene can be realized, and the task scheduling efficiency is improved. And the autonomy and maintainability of the system are improved.

The invention discloses a semi-centralized edge federated segmentation learning method of a federated learning system facing Internet of Things terminal equipment under a wireless network. The method comprises the following steps: establishing a to-be-trained model; splitting the to-be-trained model into a server-side first model and a client-side first model; distributing the client first model to each client in the to-be-trained client group; training each client first model to obtain a client second model and a first gradient corresponding to each client; according to the plurality of first gradients, updating the server-side first model to obtain a server-side second model and server-side second model features; according to the server-side second model features, updating each client-side second model to obtain a client-side third model corresponding to each client-side in the to-be-trained device group; repeating the execution; selecting the models meeting the first condition from the plurality of client third models for aggregation to obtain an aggregated client third model; and combining the aggregated third client model and the aggregated second server model to obtain a trained model. The method has the characteristics of low computing power consumption and short training time delay.

The invention provides a multi-source log fusion driven database monitoring method and system, and relates to the technical field of computers.The method comprises the steps that format analysis processing is conducted on multi-source heterogeneous logs of a to-be-monitored database on the basis of a dynamic analysis template library, and initial database logs are obtained; performing data preprocessing on the initial database log based on a rule engine to obtain a database log; and performing streaming processing on the database log to obtain a log event stream, and sending the log event stream to the server side, so that the server side detects the log event stream based on the log anomaly detection model and the associated alarm analysis engine, and generates a monitoring analysis result of the to-be-monitored database. By means of the mode, the technical bottleneck that heterogeneous database logs are difficult to analyze is broken through, the problem that the terminal is poor in database log format compatibility can be solved, invalid data transmission between the terminal and the server side is effectively reduced, and the database monitoring effect is improved.

According to the internal network and external network synchronous transmission method and system based on the TCP provided by the invention, file type detection and file title detection are realized at a server side, compliance screening is carried out on file contents before a transmission link, and sensitive files are prevented from entering a transmission link; after the detection is passed, segmented transmission is carried out by adopting a preset fragmentation rule, and breakpoint resume is executed by utilizing a fragmentation index when the transmission is interrupted, so that the stability and the efficiency of large file transmission are remarkably improved; furthermore, the server pushes the downloading notice to the extranet client through the TCP after receiving the auditing result of the intranet client with the auditing authority, so that the file is ensured to be subjected to the auditing link before being transmitted to the extranet, and the unification of high safety, controllability and high efficiency is realized.

The invention provides a segmentation federal learning method and system based on aggregation gradient broadcast, and the method comprises the steps: initializing a global model, and segmenting the global model into a server model and a client model according to the capability limitation of a local device; issuing the client model to all local equipment terminals; the multiple local devices execute forward propagation at the same time, and shredded data are obtained through calculation; transmitting the shredded data and the training data label to an edge server; performing forward propagation in parallel according to the shredded data, calculating the loss of each local equipment end in combination with a corresponding training data label, and performing back propagation to obtain a loss function and a gradient of the shredded data; aggregating the gradient of the shredded data and broadcasting to all local equipment; updating the server side model and the client side model, and aggregating the server side model; and the iterative learning loop is repeatedly executed until convergence or the maximum communication round is reached. According to the invention, the communication overhead in segmentation federal learning is saved, and the model training efficiency is effectively improved.

The invention discloses a privacy protection recommendation method based on graph federal learning, and belongs to the technical field of intelligent recommendation. The method mainly comprises the following steps: a client receives an initial weight, and constructs an initial user-article interaction graph based on user local data; modeling interaction between nodes on the initial user-article interaction graph based on the graph neural network and the initial weight; training the local user-article interaction model based on user local data; the server side aggregates the article embedding matrix gradient uploaded by each client side; the server side clusters the updated article embedding matrix to generate a sampling article set; and the client performs multi-task joint training based on self-supervised learning on the local user-article interaction model based on the user local interaction subgraph, and performs article recommendation based on the trained local user-article interaction model. The problem of unbalanced data distribution can be relieved, and meanwhile the risk of user privacy disclosure is reduced.