13 results about "Client state" patented technology

The invention relates to the technical field of Internet of Things communication, discloses an Internet of Things multi-Android client state synchronization method, and solves the problems of poor state consistency, high back-end dependence, large delay and insufficient compatibility in the prior art. According to the scheme, the method comprises the steps that after a client completes initialization, MQTT connection with a rear end is established, an MQTT state message pushed by the rear end is monitored, after the message is received, equipment state information is obtained through analysis, local state cache and a UI interface are updated, meanwhile, the client serves as a distribution node, a UDP synchronization message containing the equipment state and the IP of the distribution node is established, and the UDP synchronization message is sent to the client. The UDP synchronization message is broadcasted to the local area network, other clients in the local area network receive the UDP synchronization message and then verify whether the distribution node IP is the client in the local area network, if the verification is passed, the message is analyzed, the state updating timestamp and the local cache timestamp are compared, and the local state cache and the UI interface are updated only when the state updating timestamp is larger.

Systems, methods, and media are provided for localization of capture and processing of spoken audio based on predictive modeling. A system for capturing voice communications in multi-user sessions comprises a user client participating in multi-user sessions with voice communication channels, a memory buffer storing audio data streams transmitted via voice channels, a voice detection component analyzing the buffer and tagging portions containing spoken audio, a compression component compressing tagged portions, a stitcher component compiling compressed portions into audio chunks, a client analysis component monitoring client states, a communication component transmitting audio chunks to remote servers when clients are in non-core states, and a controller component clearing memory buffers after receiving server confirmation of chunk receipt.

The invention provides a self-adaptive decision-making method for participation degree and data use of a client in federated learning. The method comprises the following steps: initializing a global model and triggering local training of the client; the server local model calculates model offset and constructs client state features; outputting a continuous action vector through a strategy generation module, and determining a data use proportion which should be adopted by the client in the next round of training based on the action vector; the client selects a local data set training local model in proportion through a data sampling mechanism and uploads the local data set training local model to the server; the server executes weighted aggregation according to the sample size actually used by each client to form a next round of global model; according to the method, whether the client participates or not and the data use proportion during participation are generated as a uniform continuous decision variable, so that the federal learning system can perform fine-grained adjustment according to the data scale, the distribution difference and the model state in each round of training, and the updating stability and the convergence efficiency are remarkably improved.

The invention provides a long-link-based weak networking data processing method and system, computing equipment and a computer readable storage medium, and the method comprises the steps that a client optimistically executes a transactional Action behavior object locally to update a presentation layer in real time, and asynchronously serializes and stores the transactional Action behavior object to a local hard disk; furthermore, the client asynchronously uploads the Action behavior object to the server for verification according to the network condition, and the whole process is not perceived by a user. And finally, the client executes a differentiated state synchronization strategy according to a verification result: if the verification is passed, the client confirms transaction submission, and synchronously removes corresponding records in the memory and the hard disk; and if the verification fails, performing correction according to authority data returned by the server, and forcibly aligning the local state. According to the scheme, the client state can be kept strictly consistent with the server even in a weak network environment, the user experience and the data security are both considered, the client state can be degraded into a standard long link mode when needed, and the compatibility is good.

The invention relates to the technical field of information, and particularly discloses an intelligent customer management and system docking method for personal numbers, which comprises the following steps: S1, receiving a personal number registration request sent by a user terminal, and establishing a personal number data space which comprises a customer information database, a behavior record database and a system docking configuration library; s2, collecting customer data, and uniformly storing the customer data in a customer information database; s3, performing intelligent analysis on the customer data in the customer information database, and generating a customer value evaluation score and a demand prediction label; s4, based on the customer value evaluation score and the demand prediction label, generating a personalized customer management strategy, calculating a similarity matrix between customers, and generating recommendation; and S5, establishing a connection with an external service system, synchronizing the client management strategy to the external service system, and simultaneously obtaining the latest client state update. According to the invention, accurate customer management and efficient system connection can be realized, a data island is broken, and the service efficiency and customer experience are improved.

The invention discloses a security aggregation method for spatial dependency of spatio-temporal data in federated learning. The security aggregation method mainly solves the problem of privacy disclosure in cross-client spatial dependency modeling. The method comprises the following implementation steps that: a client sets parameters before training, selects an assistant and generates a key; the client extracts time sequence and space features, encrypts local topological information and the like and uploads the information to the server; the server aggregates the information, marks the client state and sends the client state to the assistant; after verification of the assistant, related masks are decrypted for the online client and the offline client respectively, and the server recovers global topological information and broadcasts the global topological information; and the client calculates global spatial dependency features according to the global spatial dependency features, and the operation is repeated until all convolutional layers are processed. According to the method, a security aggregation mechanism is embedded into a graph neural network training process, and security aggregation and utilization of spatial dependence are realized on the premise of ensuring privacy.

The invention discloses a client state detection method and device, a storage medium and computer equipment, and relates to the technical field of network communication. The method comprises the following steps: receiving a network data packet from a client, and determining whether the client is a target client based on the network data packet; after the client is determined to be the target client, communication interval timing is performed on the target client, the communication interval time length corresponding to the target client is calculated in an accumulated manner, and whether the network data packet from the target client is received again is monitored; when the network data packet from the target client is received, the communication interval time length corresponding to the target client is zeroed, and the communication interval time length corresponding to the target client is accumulated and calculated again; and when the communication interval time length corresponding to the target client exceeds a preset time threshold, determining that the target client and the server are in a disconnected state. According to the method, the network bandwidth occupation amount when the connection state of the client is detected can be reduced.

Techniques are disclosed that enable automating user interface input by generating a sequence of actions to perform a task utilizing a multi-agent reinforcement learning framework. Various implementations process an intent associated with received user interface input using a holistic reinforcement policy network to select a software reinforcement learning policy network. The sequence of actions can be generated by processing the intent, as well as a sequence of software client state data, using the selected software reinforcement learning policy network. The sequence of actions are utilized to control the software client corresponding to the selected software reinforcement learning policy network.

The invention provides a multi-layer heterogeneous graph construction method based on a TCP session, and relates to the technical field of network security. The method comprises the following steps: reading original network flow data in a PCAP form and splitting the original network flow data into TCP sessions; sorting the data packets in the TCP session according to the timestamps, and generating data packet nodes; constructing a data packet time sequence edge according to the sequence of the timestamps, and associating each data packet node to construct a data packet sequence diagram; constructing an initial server-side state node and an initial client-side state node; traversing the data packet nodes of the data packet sequence diagram in sequence, and judging whether the states of the server-side state nodes and the client-side state nodes are changed or not; if so, newly establishing a server-side state node and a client-side state node, and constructing a corresponding state transition edge and a corresponding mapping association edge; and if not, constructing a corresponding mapping association edge. According to the method, the key information of the TCP session can be completely described, staged accurate extraction of subsequent features can be conveniently realized, and cross-stage feature dilution can be avoided.

Techniques are disclosed that enable automation of user interface inputs by generating action sequences to perform tasks using a multi-agent reinforcement learning framework. Various implementations process an intent associated with a received user interface input using an overall reinforcement policy network to select a software reinforcement learning policy network. An action sequence can be generated by processing the intent using the selected software reinforcement learning policy network and a sequence of software client state data. The action sequence is used to control a software client corresponding to the selected software reinforcement learning policy network.

The application provides a virtual reality scene position synchronization method and device, the method comprises the following steps: defining a field of a message sent between a client and a server; determining a time interval T1 of sending the message between the client and the server, and setting a client state sampling time interval T2; dynamically adjusting the time interval T1 of sending the message, and each client sends a message comprising client state information to the server based on the respective time interval T1 of sending the message; and other clients update the state information of the client corresponding to the received message based on the update step length set by the other clients and the received message, and simulate a virtual object of a user corresponding to the client corresponding to the received message. The method makes the movement of the client more synchronized on different device terminals of different clients, and can greatly reduce the visual difference when starting to move.

The invention discloses a cross-architecture program simulation execution method and device and a storage medium. The method comprises the following steps: establishing a register resource mapping relationship between a target client register resource and a host register resource; defining a calling rule corresponding to the register resource mapping relationship, wherein the calling rule is used for transmitting target client state information between simulation functions executed on the host machine; performing static translation on the target client executable file based on the register resource mapping relation and the calling rule to generate a corresponding host executable file; loading the target client executable file and the corresponding host machine executable file; and in the process of simulating and executing the executable file of the host machine, simulating and executing the executable file of the target client machine based on the register resource mapping relationship and the calling rule. The method has the technical effect of improving the simulation execution efficiency of the instruction simulator.

This invention discloses a federated training scheduling method based on hierarchical dynamic selection and diversity constraints, belonging to the field of distributed machine learning. This invention aims to address the problems of low training efficiency and unstable convergence in existing federated learning methods under environments with both statistical and system heterogeneity. This invention obtains a set of available clients, maintains latency estimates for each client and hierarchically divides them, determines the sampling quota for each layer based on the average latency, historical participation, and fairness constraints, constructs a client data difference graph within each layer, and uses a greedy decentralized selection strategy with diversity constraints to determine the set of participating clients. Local updates for selected clients are aggregated and the client states are updated. This invention is applicable to distributed training scenarios such as mobile terminal collaborative learning, vehicle networking, and edge intelligent analysis.