465 results about "Middle tier" patented technology

The invention belongs to the field of intelligent edge systems and privacy enhancement computing, and particularly relates to a privacy protection-oriented robot large model cloud edge end collaborative reasoning and federated learning system, which comprises a cloud server layer used for deploying a large-scale pre-training model and executing complex reasoning and global federated learning coordination; the edge calculation layer is used for deploying an intermediate layer model and executing local data aggregation, privacy protection processing and intermediate feature calculation; the terminal equipment layer is used for deploying a lightweight model and executing data acquisition, primary processing and lightweight reasoning; the federated learning framework is used for optimizing the model; the privacy protection module is used for integrating data localization, differential privacy, homomorphic encryption, secure multi-party computing and a block chain verification mechanism; the adaptive allocation module is used for dynamically adjusting computing resources. According to the method, the problems of privacy leakage risk, computing resource limitation, network delay, insufficient data isolation and the like of the traditional AI service in a robot scene are solved, and efficient privacy protection and data security isolation are realized.

The invention relates to the technical field of intelligent fusion terminal communication, and discloses an intelligent fusion terminal multi-protocol communication method and system based on edge computing. According to the method, a protocol adaptive engine is deployed at an edge node, an original data stream of a communication link is collected and analyzed in real time, and a current protocol type is dynamically identified in a fuzzy matching mode. And based on an identification result, the system dynamically loads a corresponding protocol analysis module, generates an adaptive instruction set, and realizes standardized data frame encapsulation through a protocol conversion intermediate layer. And meanwhile, the system monitors the link state, triggers incremental updating of the protocol feature library, and realizes seamless protocol switching. According to the invention, the communication compatibility and reliability are improved, and the requirements of high-reliability scenes such as the industrial Internet of Things are met.

The invention discloses a knowledge cross validation question-answering method and system for reducing hallusion of a large language model, and belongs to the technical field of artificial intelligence, and the method is implemented by the following steps: generating results through multiple times of sampling: when a user puts forward a question, controlling the large model to perform multiple times of sampling, and generating a specified number of results; calculating hidden state related indexes: extracting the hidden state of the last token of the middle layer of the large model corresponding to the result, and calculating covariance matrixes and answer discrete feature values of the hidden states; mLP model prediction: inputting the discrete feature value of the answer and the length of the answer into a multilayer perceptron MLP, and outputting a hallucination-free probability; querying and summarizing a knowledge graph; and calculating a final illusion-free score and outputting a result. According to the method, the answer quality and credibility of a large language model can be remarkably improved, and the method is particularly suitable for application scenes with extremely high requirements on the accuracy of single-mode text generation contents.

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 relates to the technical field of artificial intelligence model optimization, and discloses a multi-granularity dynamic pruning method and system for a generative AI model. The system comprises a model state acquisition module which acquires output of a middle layer in real time through a probe, and constructs a feature information set of neuron activation distribution, weight matrix norm and connection topology; the sparseness evaluation module outputs a sparseness risk value based on the feature information set, the initial sparseness parameter and the real-time computing resource state; a pruning planning collaborative analysis screening module generates a multi-granularity pruning scheme when the risk value exceeds a threshold value, calculates a collaborative interference value of precision recovery operation and screens an optimal scheme; and the pruning strategy execution feedback module updates model parameters, generates logs and feeds the logs back to the management terminal. The system realizes adaptive structure optimization and resource scheduling of the generative model in the reasoning process.

The invention discloses a method for improving private domain operation and maintenance knowledge retrieval quality based on RAG, knowledge retrieval quality is improved through multi-stage cooperative processing, in a document preprocessing stage, a basic DOM tree is constructed, visual blocks are identified by using OCR and spatial clustering, a dynamic semantic slicing mechanism is adopted, and the knowledge retrieval quality is improved. A large language model and a sliding window technology are combined to generate fine-grained knowledge blocks, a cross-document context tracking network is innovatively constructed, a knowledge block reference relation graph is established based on semantic vector calculation, the traditional retrieval islanding effect is broken through, header alignment and merging optimization are implemented for table data, retrieval integrity is ensured, and the retrieval efficiency is improved. A lightweight knowledge injection strategy is adopted in a model level, intermediate layer parameters of a pre-training model are migrated, TripletLoss and contrast loss function training are combined, the field adaptability is enhanced while the generalization ability is kept, finally, a structured vector index is output, and timeline normalization processing of multi-version logs is supported.

The invention relates to the technical field of embedded maps, and provides a vector map display method under a VxWorks system, which comprises the following steps: S1, analyzing a vector map source file through an MITAB library, and extracting geometric space topological data, geographic coordinate system parameters, style attributes and projection parameters; s2, constructing a hierarchical storage structure based on SQLite3, and storing vector map information including a spatial data table, a style configuration table and a metadata table into a geographic information database; s3, creating a map display control MapWidget inherited from a QGraphics View class based on a Qt framework, loading the geographic information database through a dynamic configuration interface, and initializing a coordinate system conversion module, a multi-level view window module and a covering management system; and S4, dynamically rendering and outputting an interactive vector map interface through a layer priority scheduling strategy in a VxWorks real-time operating system. Through the design of the cross-platform data middle layer, the extensible general symbology system and the resource efficiency optimization, the data compatibility problem of the embedded system and the commercial GIS software is solved.

The invention discloses a large visual language model vulnerability detection method and device based on security sensitive layer activation guidance and a storage medium, and belongs to the technical field of artificial intelligence security, the method comprises the following steps: S1, constructing a security sensitive layer activation induction sample data set; s2, performing quantitative analysis on activation differences of the model intermediate layer under attack and normal input; s3, on the basis of the recognized security sensitive layer combination and an output layer of the large visual language model, generating an in-process confrontation image with an attack attribute through optimization of a loss function; and S4, recording an optimal security sensitive layer combination and a harmful response path guided by the optimal security sensitive layer combination, and constructing a structured representation of the potential security vulnerabilities of the model. According to the method, the security fragile links of the model are accurately positioned, the attack guiding force is improved in combination with multiple losses, the generated confrontation disturbance is controllable and imperceptible, the good cross-instruction migration capability is achieved, and the method adapts to various security assessment and red team test tasks.

The invention relates to a man-machine cooperation fabricated construction management and control system and method based on a BIM and a Unity model.The BIM is integrated into a man-machine cooperation construction workflow through an interactive closed-loop digital twin framework, the closed-loop digital twin framework comprises a model building module, a man-machine interaction module, a middle layer module and a data collection module, the model building module is used for building a BIM model for the prefabricated part; the man-machine interaction module realizes manual remote control through a graphical user interface, issues an instruction to the robot, intervenes and supervises the construction process, and transmits the instruction to the middle layer module through a programming language; the middle layer module is provided with a data processing module, a machine autonomous control module and a decision information integration module; and the data is transmitted to the BIM through the data processing module to update the BIM model so as to reflect the change between the design construction state and the actual construction state. The method has the advantages that the participation degree of the building robot in site construction work is improved, and a solution is provided for intelligent construction of the fabricated building.

The invention discloses a BeeGFS-based metadata query efficiency improvement method and system. The BeeGFS-based metadata query efficiency improvement method comprises the following steps: constructing a three-dimensional index network consisting of a PMEMKV hot data layer, a RocksDB common data layer and a ZFS cold data layer; dividing all metadata into three independent sets of hot data, common data and cold data according to the heat values, and respectively storing the three independent sets to corresponding layers; after a query request is received, a top layer, a middle layer and a bottom layer are queried in sequence, and corresponding metadata is returned if the query request is hit; monitoring query indexes in real time, wherein the query indexes are used for updating the heat value of the metadata and adjusting the storage configuration of each level; performing dynamic promotion or degradation on the metadata between the three-dimensional metadata index networks according to the latest popularity value of the metadata and a preset popularity threshold value, and performing data synchronization and query path switching in the promotion or degradation process; the method can reduce query delay, improve throughput and balance storage resources.

The invention relates to the technical field of computer networks, in particular to an intelligent agent system and method based on the MCP technology, and the system comprises an MCP intermediate layer protocol engine, an intelligent scheduling module, a channel management system, a security enhancement mechanism and an edge node self-learning mechanism. The method has the beneficial effects that support for various communication protocols (such as HTTP, MQTT, gRPC and the like) is realized through the MCP technology. Through the protocol adapter module, the system can automatically identify and load different protocol adapters, and parallel work of multiple protocols can be supported without manual intervention. The multi-protocol support greatly improves the flexibility of the system, so that the system can be seamlessly connected with equipment of different manufacturers and standards, and the requirements of complex and diverse application scenes are met.

The invention provides an FPGA (Field Programmable Gate Array) multi-modal control method based on dynamic partial reconfiguration, which comprises the following steps that: a main body adopts a layered hardware architecture to realize multi-modal control, firstly, a system module of a whole system is divided, so that FPGA logic resources are divided into a static region and a reconfigurable region; wherein the static area is responsible for global control and a communication interface part, the reconfigurable area realizes a multi-mode function by loading different configurations, and the middle layer is used for deploying the dynamic part to reconfigure a controller and is responsible for time sequence coordination and bit stream management. Compared with the prior art, the method has the advantages that the mapping relation between the modal feature library and the hardware module is established, the priority preemptive reconfiguration strategy and the modal switching cost model are adopted, the real-time matching of the sensor data flow and the hardware architecture is achieved, the dynamic data path reconstruction technology supports multi-modal parallel processing, and the real-time matching of the sensor data flow and the hardware architecture is achieved. Meanwhile, time sequence constraint, resource occupancy rate and power consumption budget are considered.

The invention discloses a cascade speculation inference method and system based on hierarchical decline KV cache compression, and the method comprises the steps: firstly inputting a context prompt text into a target model for coding, generating a KV cache, and calculating an attention score between tokens; secondly, descending sorting is carried out based on the attention scores of the final input token, KV cache blocks corresponding to the first k attention scores are selected as cascading middle layer cache, a lightweight large language model is loaded to serve as a draft model, and a hierarchical decline KV cache compression strategy is adopted to maintain draft model cache. Then, based on all the caches, a double-layer cascade speculation reasoning framework is constructed, a target reasoning path is obtained, and the caches are updated; and finally, repeating the operation until target response data corresponding to the context prompt text is output according to the target reasoning path. According to the method, the KV cache proportion is reduced, meanwhile, the draft token acceptance rate of the target model of the full KV cache is improved, and reduction of precision is reduced.

The invention discloses a hierarchical cloud storage adaptive migration method based on data popularity, a medium and equipment. The method comprises the following steps that a migration management platform composed of a statistics and alarm module, a heat model calculation module, a strategy control center module, a metadata management module and a migration action module is added to a middle layer of service side storage and archiving side storage; when the data written by the client reaches an alarm threshold, the modules cooperate with each other to adaptively migrate the data from the service side storage to the archiving side storage according to the data popularity; and when the client needs to relocate the data, the modules cooperate with each other to store the relocated data from the filing side to the service side. The invention provides a self-adaptive migration method according to data popularity in a hierarchical cloud storage environment, and the overall performance and the resource utilization rate of a storage system are improved.

The invention provides a multi-protocol data interaction system, a data interaction method and a data interaction device. The system comprises a process simulation service and a field device, the process simulation service comprises a computing system and a middle layer system, the computing system is in communication connection with the middle layer system, and the computing system is used for computing control parameters of the field device and sending the control parameters to the middle layer system; the intermediate layer system comprises a plurality of protocol interfaces, is in communication connection with the field devices through the protocol interfaces, sends the control parameters to the corresponding field devices through the protocol interfaces, receives action information of the field devices through the protocol interfaces, and uploads the action information to the computing system; and the field equipment is used for receiving the control parameters, executing response actions according to the control parameters and feeding back action information to the intermediate layer system. The problem that in the prior art, data interaction with different field devices can be achieved only by developing multiple sets of communication systems, and consequently development cost and complexity are high is solved.

The invention discloses a multi-core platform-oriented collaborative scheduling method and system and terminal equipment thereof, and belongs to the technical field of collaborative scheduling of operating systems. The method comprises the steps of obtaining a task initial process based on a multi-core heterogeneous chip platform; a task dependency graph is obtained, loop dependency detection and parameter consistency detection are executed based on the task dependency graph, and an intermediate layer task description file is obtained; acquiring real-time operation data and a master control heartbeat survival table based on the intermediate layer task description file; obtaining a target master control node based on the real-time operation data and a preset election method when it is judged that the preset failure condition is met; and carrying out a breakpoint snapshot action based on the target master control node, and carrying out degradation processing based on the original master control node. According to the collaborative scheduling method and system for the multi-core platform and the terminal equipment thereof, the technical problem that an effective master control hot switching and task migration mechanism is lacked in the prior art can be solved, and the intelligent level and reliability of heterogeneous collaboration of multiple operating systems are remarkably improved.

The invention belongs to the technical field of data processing, and particularly discloses a multi-source heterogeneous data acquisition access method and system based on intelligent mapping and a dynamic interface, and the method comprises the steps: registering an input data source class, generating a data source identifier corresponding to the data type, and storing the data source identifier; respectively executing meta-model analysis operation to generate a standardized meta-model associated with the data source identifier and a container instance template; generating a mapping rule set through a twin neural network architecture with multi-modal feature fusion and a preset mapping rule; a middle layer API interface is constructed through interface logic, and real-time conversion and access between input and output data are achieved; and realizing containerized parallel access based on the container instance template and the containerized interface service instance. According to the method, data source change is dynamically adapted, the real-time performance and accuracy of the mapping rule are ensured, and meanwhile, multi-source heterogeneous data can be accessed efficiently in real time in combination with containerized parallel processing and a data source change monitoring mechanism.

The invention discloses an intelligent design research and development system based on domestic AI and an implementation method thereof, the system adopts a modular loose coupling architecture, and comprises a front end layer, a middle layer, a core layer, a base layer and a data layer; the front-end layer is a Web front-end interface, and the middle layer is an API gateway layer; the core layer comprises a knowledge graph module, a simulation engine module, a decision center module and a digital twin manager module; the basic layer comprises a system management module, the data layer comprises a multi-layer data storage and communication component, and the method comprises the steps of initializing a system manager, constructing a knowledge graph, executing multi-physics field simulation and evaluating simulation confidence; generating a decision strategy based on reinforcement learning and knowledge graph reasoning; according to the method, digital twin bodies are constructed, virtual-real linkage and real-time data interaction are achieved, domestic AI is comprehensively adopted and comprises a MindSpore framework and an ancient big model, the technology is autonomous and controllable, and the method is suitable for the whole process of power equipment design, inspection and detection and supervision and construction.

The invention relates to the technical field of voice processing, in particular to a real-time intention recognition method and system based on streaming incremental reasoning, and the method comprises the following steps: receiving the voice input of a user through a voice collection module, slicing the voice into a plurality of audio frames, and carrying out the recognition of the intention of the user through an incremental large language model module by adopting an Early-Exit reasoning mechanism; side outlets are arranged at multiple levels of the model, incremental reasoning is performed on token streams based on a QLoRA4-bit quantization technology, prediction results of multiple tokens are smoothed by using a stream ASR decoding module and an accumulative fusion module, and a stable final label is generated; the method has the beneficial effects that by combining streaming ASR decoding with incremental large language model reasoning, intention recognition and risk assessment can be immediately performed on each speech token after the speech token is generated. Through an Early-Exit reasoning mechanism, under the condition of high confidence, the system can output a fraud intention in advance in a middle layer in the reasoning process and stop subsequent calculation, and unnecessary calculation overhead is reduced.

The invention provides a model task classification method and device, a model selection method and electronic equipment, and the method comprises the steps: obtaining a to-be-processed data set which comprises task data of a plurality of tasks; processing each piece of task data based on the multi-task processing model to obtain a task processing result corresponding to each piece of task data and an output result of each piece of task data in a middle layer of the multi-task processing model; determining a path vector of each piece of task data based on the output result of each piece of task data; based on the path vectors of the task data of the tasks, classifying the tasks to obtain a plurality of task classification groups; and determining a task category label corresponding to each task classification group according to a task processing result corresponding to the task data of the tasks in each task classification group. According to the embodiment of the invention, the task execution accuracy of the multi-task processing model can be improved on the basis of defining the executable task category of the multi-task processing model.

The invention discloses a personalized federated graph learning method oriented to equipment resource isomerism, and aims to solve the defects in graph data isomerism, equipment resource adaptation and privacy protection in the prior art. According to the method, collaborative optimization is realized through a closed-loop process of local pre-training, embedding aggregation, personalized training, soft label collaboration and classifier distillation. Each client pre-trains a model based on a local graph data set and generates interlayer embedding, and uploads the model to a server after sampling and privacy enhancement; the server performs aggregation to form a public embedded data set and distributes the public embedded data set to the client to support distillation training and soft label generation; the soft labels are filtered and subjected to weighted aggregation to form global soft labels, and the client completes classifier knowledge distillation by combining the public embedded data set and the global soft labels, and iteratively optimizes the performance of the model. According to the method, the generalization ability and the resource utilization efficiency of the model are remarkably improved while the data privacy is guaranteed, and the method is suitable for distributed graph data training tasks of multiple scenes such as social networks.

The invention discloses a method and device for accessing hardware equipment to a model reasoning acceleration framework and electronic equipment, and relates to the technical field of computers, and the method comprises the steps: setting a bidirectional decoupling middle layer between target hardware equipment and the model reasoning acceleration framework, a unified memory operation interface and a unified kernel function execution interface of different hardware devices are arranged in a middle layer, and kernel functions of the different hardware devices are registered to a model reasoning acceleration framework through the middle layer, so that the kernel functions of the model reasoning acceleration framework are separated from a system during hardware running. Different hardware devices can realize access to the model reasoning acceleration framework through the intermediate layer, and the codes, operators and even model structures of the model reasoning acceleration framework do not need to be adjusted to adapt to different hardware devices, so that the problem of poor universality of the model reasoning acceleration framework in related technologies is solved, and the universality of the model reasoning acceleration framework is improved. The technical effect of improving the universality of the model reasoning acceleration framework is achieved.

The invention discloses a heterogeneous user instruction set-oriented signal processing method, which comprises the following steps of: identifying and marking the architecture type of a process instruction set in a process loading stage of a user mode, executing a user mode program by the process, and if the user mode program is interrupted or is abnormally triggered to fall into a kernel mode, executing the process; if yes, the process instruction set architecture type is judged for the first time through a flag bit in the created cross-state privileged service context conversion middle layer; if the process is the RISC-V process, the context of the user program is stored according to the regulation of RISC-V, and an operation site is created; if the process is the AArch64 process, the context of the user program is stored according to the regulation of AArch64, and an operation site is created. According to the method, user-awareness-free cross-instruction-set binary compatible execution and operating system-level execution stream safety guarantee are achieved.

A system for faster object attribute and / or intent classification may include an machine-learned (ML) architecture that processes temporal sensor data (e.g., multiple instances of sensor data received at different times) and includes a cache in an intermediate layer of the ML architecture. The ML architecture may be capable of classifying an object's intent to enter a roadway, idling near a roadway, or active crossing of a roadway. The ML architecture may additionally or alternatively classify indicator states, such as indications to turn, stop, or the like. Other attributes and / or intentions are discussed herein.

The invention discloses an improvement method of a large language model, electronic equipment and a storage medium, and the method comprises the steps: calculating a correlation matrix between middle layers through a public data set, and calculating mutual information redundancy between the layers based on the correlation matrix between the middle layers, an inter-layer redundancy matrix is constructed through mutual information redundancy among the layers; calculating the total redundancy of each layer based on the redundancy matrix, and generating a layer importance factor through function conversion; establishing an objective function based on the layer importance factors and constructing a linear optimization model of sparse rate distribution of each layer; solving the optimization model by adopting a linear programming algorithm to obtain an optimal sparse rate distribution scheme of each layer; differentiated pruning is carried out on each layer of weight matrix according to the distribution scheme, a weight parameter with the maximum amplitude in each layer of weight matrix is reserved, the rest parameters are set to be zero, and a sparse weight matrix is obtained; according to the method, the features can be selectively pruned in the middle layer, so that fine-grained optimization of a large language model is realized.

The invention belongs to the technical field of nuclear power DCS knowledge base construction, and aims to solve the problems of knowledge dispersion, update lag, single expression, lack of intelligent application, difficulty in sharing and inheriting and the like in a traditional knowledge management mode of a nuclear power DCS. The invention provides a nuclear power DCS knowledge base construction system and method, the system comprises an application layer, a middle layer, a large model layer, a data layer and a computing power resource layer, the application layer comprises a knowledge base management module, an intelligent interaction module, a data service module, an office assistant module and an operation and maintenance assistant module; according to the method, a system architecture of a knowledge base is designed and comprises function division of a computing power resource layer, a data layer, a large model layer, a middle layer and an application layer. According to the method, an intelligent, systematized and efficient DCS knowledge base can be constructed, a powerful guarantee is provided for safe and stable operation of a nuclear power plant, and the efficiency, accuracy and practicability of nuclear power DCS knowledge management can be improved.

The invention provides a knowledge distillation-based model compression method and system, and the method comprises the steps: inheriting a soft label of a maturely trained teacher model, defining a distillation loss function of a student model through the soft label, enabling the student model to be compressed, generating a new sample extension training set through random sample extraction and mixing, and achieving the compression of the student model. The problem of insufficient data is relieved, meanwhile, the robustness of the model to noise and distribution offset is improved, through dynamic weight adjustment, a student model is made to quickly fit teacher model knowledge, autonomous optimization is conducted in the later training period, and over-fitting of soft labels is avoided. On a test set in which the proportion of unlabeled data is 30%, on the basis that the accuracy of using a real label is 78.3%, the accuracy is improved by 11.2% by using a pseudo target, deployment is performed by using a maturely trained student model, middle layer feature alignment is omitted, the occupation of a GPU video memory is reduced from 3.2 GB to 1.8 GB, and the deployment requirement of edge equipment is met.

The invention relates to a knowledge sharing and realization system based on a social network, and belongs to the technical field of information. The system adopts a three-layer framework including a bottom layer, a middle layer and a top layer. And the bottom layer comprises a social relation graph construction module, a behavior analysis module and a multi-terminal interaction module. And the middle layer comprises a database module, an intelligent recommendation module, a security verification module and a payment gateway module. And the top layer is used for realizing interaction between a user and the system, and comprises a transaction collection module, a knowledge sharing module and a personal information module. According to the method, the knowledge exchange and realization requirements of the college students are met, the learning efficiency and practical ability of the college students are improved, the development of knowledge economy of the college students is effectively promoted, optimal configuration and efficient utilization of educational resources are promoted, and the method has important social value and wide market prospects. Important practical significance and market prospects are realized.

An architecture is disclosed for submission and recordation of application programmer interfaces (APIs) using non-fungible tokens (NFTs) inserted into a blockchain. The API-NFT pairings are validated by nodes of the network. The system automatically searches for and generate NFTs for APIs and / or intermediate layers on a computer network based on metadata associated with the API and a hash of the API / intermediate layer. The API-NFT pair binding ensures that developers / consumers are ingesting APIs that are secure and do not include malicious rules. Furthermore, performance monitoring of the API is performed using artificial intelligence (AI) and machine learning (ML), including long short term memory (LSTM) neural networks.

The invention discloses a protocol conversion method, which comprises the steps of receiving an original data frame from a charging power supply, extracting a protocol feature vector of the original data frame through a pre-trained machine learning model, and matching a first protocol type of the original data frame from a protocol feature library based on the protocol feature vector; calling an analysis rule base corresponding to the first protocol type to deconstruct an original data frame into intermediate layer data irrelevant to the protocol; and calling a field conversion rule from a protocol mapping table associated with the second protocol type, dynamically recombining the standardized field label in the intermediate layer data into a target data frame structure, and injecting a protocol control character originating from a target protocol specification to generate a target data frame conforming to the second protocol type. The method has the capability of adaptively learning a new protocol, reduces the manual intervention and maintenance cost, and improves the compatibility, communication reliability and overall operation efficiency of a battery replacement system for vehicles of different brands.