73 results about "Algorithm configuration" patented technology

The invention belongs to the technical field of artificial intelligence, is applied to the field of financial science and technology and the field of medical health, and discloses a rule engine configuration method, device and equipment based on machine learning and a storage medium. The method comprises the steps that historical calculation data and performance indexes in historical calculation tasks are collected, and a training data set is constructed; a machine learning model is trained, and algorithm parameters of the rule engine are dynamically optimized through the training data set; configuring an expression and a user-defined function of a rule engine, and realizing analysis and execution of scripts through JEXL (JavaScript Exchange Library); establishing a dynamic mapping relationship between the algorithm configuration library and the business variables, mapping algorithm variable names and business variable names, and generating configurable algorithm rules; and analyzing task load and resource requirements based on a pre-trained machine learning model, distributing system resources, and repairing abnormal behaviors in a calculation process. According to the invention, the problems of low processing efficiency, high maintenance cost, unreasonable resource allocation and poor expandability in the prior art are solved.

Methods for a machine-learning network that provide efficient, scalable, and granular analyses during validation of a machine learning model are disclosed. Validation of models depends upon many factors, including the real-world application of the model, the type of model being trained, and the types of data samples it is being trained on. In order to provide relevant edge case information to users that pertains to their specific model, data slice finding techniques may be used to identify subsets of the dataset that are particularly problematic. By limiting a length of the slice description that the algorithm searches and by configuring the algorithm to target specific types of errors, users are provided with a more granular analysis that then allows them to determine how or if they need to retrain the model.

The invention discloses a reinforcement learning aided evolutionary algorithm for parameter optimization in algorithm configuration, which comprises the following steps of: performing parameterized definition on four core operators, coding all adjustable parameters, and constructing a unified parameter configuration space; then randomly initializing a configuration population, and inputting the configuration population into an optimization framework assisted by reinforcement learning; the framework is internally provided with an intelligent agent and is executed in each generation of iteration: operating an evolutionary algorithm according to parameter configuration and evaluating performance; the intelligent agent adaptively selects a predefined alignment strategy based on a population state, and guides mutation operation to generate a filial generation; performing environment selection and population updating based on performance evaluation; and finally, calculating rewards according to population performance, and training and updating the agent decision model. And after iteration to a termination condition, outputting configuration with optimal performance and a corresponding algorithm thereof. According to the method, an evolutionary algorithm search space is constructed through comprehensive parameterization, a reinforcement learning agent is introduced to adaptively process variable-length coding, and a double-layer optimization framework is adopted to efficiently search optimal algorithm configuration.

The invention discloses an industrial image anomaly detection method based on a multi-agent arrangement heterogeneous algorithm. The method comprises the following steps: S1, constructing an algorithm component library for shielding data representation differences among different architecture algorithms; s2, constructing a multi-modal large model system based on a multi-agent architecture; the multi-agent architecture comprises the following steps: receiving an image by using a visual expert agent, and outputting structured physical metadata and an unstructured visual suggestion text; the planner agent receives the physical metadata and the visual suggestion text, and generates an algorithm configuration tree based on an algorithm component library; the optimizer agent optimizes the algorithm configuration tree in the closed-loop feedback stage; s3, using the algorithm configuration tree to construct a reference model representing good product distribution according to the normal training sample images; s4, performing anomaly detection on the to-be-detected sample image by using the algorithm configuration tree to generate an anomaly judgment result; and S5, optimizing the algorithm configuration tree.

The invention discloses an automatic algorithm configuration method based on deep reinforcement learning in evolutionary computation. The method comprises the following steps: determining a problem set; obtaining initial data for constructing a Markov decision process; defining a state vector of each individual in a time step in an evolutionary computation algorithm by using an individual state vector according to the initial data; constructing feature vectors of population features, individual features and development features according to the initial data; defining a continuous action space according to the initial data, and jointly controlling the selection of all individuals on hyper-parameters; based on whether a better solution is found or not, a reward mechanism is set, and maximization of expected benefits is achieved; and the deep reinforcement learning agent uses an attention mechanism to control dynamic hyper-parameters of the algorithm, so that a Markov decision process is realized, and automatic algorithm configuration is completed. According to the method, direct mapping from high-dimensional sensing information to continuous action space output is realized, configuration of individual exploration and development tradeoff is dynamically adjusted, and a balance mechanism of exploration and development is optimized.

The invention relates to the technical field of semiconductor manufacturing, in particular to a semiconductor scheduling algorithm model configuration system, method and equipment and a medium. The system comprises a data configuration module, an algorithm configuration module, a parameter analysis module, a task scheduling module, a data acquisition module, a model execution module and a result processing module. The data configuration module receives data model configuration parameters input by a user; the algorithm configuration module receives algorithm model configuration parameters. The parameter analysis module analyzes parameters and generates a data preprocessing rule and an algorithm model instance, the task scheduling module creates a scheduling task and triggers a process, the data acquisition module acquires original data from a source data table and preprocesses the original data, the model execution module executes a scheduling algorithm based on the algorithm instance, and the result processing module outputs standard data. According to the invention, decoupling of the data and the algorithm is realized, and the adaptability and expandability of the system are improved.

The application discloses a cryptographic algorithm debugging data hierarchical grouping compression method for a reconfigurable cryptographic chip. In the debugging work of the reconfigurable cryptographic chip, the debugging data is divided into reconfigurable cryptographic algorithm configuration information and operation intermediate data. The debugging data is processed by hierarchical grouping, the debugging data of each level module is divided into one or more groups, then the debugging data is filled or spliced, and then the reconfigurable debugging data is compressed by using a Huffman coding mode. After the hierarchical grouping compression method of the debugging data is used, the transmission performance of the debugging data can be effectively improved while the storage resource of the debugging data is reduced.

This application provides a method for configuring video stream algorithms, comprising: extracting visual features, spatiotemporal features, and semantic features of the input video stream; fusing these features to generate a first scene feature of the video stream; and generating a configuration algorithm for the video stream based on the first scene feature and a pre-constructed knowledge graph representing the relationship between scene features and algorithms. The technical solution of this application can effectively solve the problems of high configuration cost, poor scene adaptability, knowledge transfer gaps, and low configuration efficiency of video stream algorithms in smart city scenarios in existing technologies.

There is provided a computer implemented method comprising obtaining image data representing a platform, processing the image data using an algorithm, the algorithm configured to determine a property of the platform, obtaining, from the algorithm, data indicating a property of the platform and determining, based on the data indicating the property of the platform, a location to which the platform is to be transported.

The invention provides a performance analysis method and device for source address verification, equipment and a storage medium, and relates to the technical field of network security. The method comprises the following steps: receiving a source address verification processing program written through a description language; loading an algorithm configuration file; wherein the algorithm configuration file at least comprises a performance model configured for a plurality of preset algorithms, and the performance model is used for calculating the consumed time length of the operation statement based on the operand length of each operation statement; based on the source address verification processing program and the performance model, calculating the consumed time length of each operation statement; and generating a performance analysis report of the source address verification processing program according to the time-consuming duration of all the operation statements in the source address verification processing program. According to the embodiment of the invention, automatic and theoretical performance analysis can be carried out on the source address verification processing flow based on the formalized description language and the predefined algorithm configuration file under the condition of separating from the dependence of specific hardware.

The invention discloses an encryption device and system, and the device comprises a global encryptor which is used for obtaining algorithm configuration data which is determined through the joint negotiation of a producer and a consumer based on an agreed target cryptographic algorithm, and constructing a target encryptor instance which supports the target cryptographic algorithm; the encoder conforms to serializer interface specifications of a client library of a producer and is used for encoding first data generated by the producer into a first plaintext byte array; the target encryptor instance is called to encrypt the first plaintext byte array into a first ciphertext byte array through a target cryptographic algorithm, so that a producer stores the first ciphertext byte array to a distributed stream processing platform; and the decoder conforms to a deserializer interface specification of a client library of the consumer and is used for calling the target encryptor instance to decrypt a second ciphertext byte array obtained by the consumer from the distributed stream processing platform into a second plaintext byte array through a target cryptographic algorithm and decode the second plaintext byte array into second data.

The invention relates to a supervision algorithm calling method and device, computer equipment, a medium and a product. The method comprises the following steps: receiving field operation information of a current task sent by terminal power equipment and algorithm configuration information sent by a management platform; determining a target algorithm corresponding to the field operation information; determining a target server corresponding to the field operation information according to the algorithm configuration information and a target algorithm; the algorithm configuration information comprises a corresponding relation between algorithms and servers, and one server corresponds to at least one algorithm; sending the field operation information to a target server; receiving a processing result sent by the target server; the processing result is obtained by processing the field operation information by the target server based on the target algorithm. By adopting the method, real-time and accurate supervision of the field operation process can be realized, and the strict requirements of modern industry on operation safety, high efficiency and compliance are met.

The disclosure provides methods and apparatus for training and executing a reinforcement learning, RL, algorithm configured with multiple reward functions to perform a task in a communication network (1000) The method (100, 500) comprising, the first network node (1100) transmitting (103, 503) a first message, to the second network node (1200), comprising parameters of the RL algorithm. The first network node (1100) receiving (104, 507) a second message, from the second network node (1200), comprising one or more observations. The first network node (1100) updating (105, 508) the parameters of the RL algorithm based on the received one or more observation. The first network node (1100) performing (106, 509) the transmitting, the receiving and the updating iteratively until a stopping criterion exceeds a threshold and storing (107, 510) the updated parameters of the RL algorithm.

The invention provides an old city pattern recognition and updating method and system based on multi-source data fusion, and the method comprises the steps: collecting a multi-source data set, analyzing n-hour activity cycle characteristics, and generating a time period thermodynamic diagram; identifying a community structure by using a graph neural network, outputting social topology, and quantifying a texture evolution track by using a dynamic attitude model; quantizing the traditional spatial pattern by using a landscape sight line algorithm, and outputting a dynamic pattern recognition result; and configuring aging facilities by using a population adaptation algorithm, optimizing industrial layout by using a location entropy algorithm, and generating a micro-update scheme. According to the method, a space-time-society-ecology four-dimensional dynamic pattern recognition system is constructed; spatial pattern quantization and weight adjustment are realized by innovatively using a landscape sight line algorithm, and the traditional limitation is broken through; and finally, a tensor decomposition algorithm is used for fusing the configuration and optimization scheme, a micro-update scheme is generated, an update strategy is formed through iterative optimization, complex requirements are accurately responded, and the old city pattern recognition scientificity and the update scheme implementation efficiency are improved.

The invention discloses a VPN (Virtual Private Network) system encryption algorithm management and control method and device, relates to the technical field of network security, and mainly aims to overcome the limitation of fixed configuration of a VPN system encryption algorithm and realize dynamic configuration of a VPN tunnel encryption algorithm. According to the main technical scheme, the method comprises the steps that the VPN system configures an algorithm calling engine corresponding to at least one algorithm library, and the algorithm calling engine executes encryption operation by calling encryption algorithms in the corresponding algorithm library; if it is determined that the VPN system establishes the tunnel, algorithm description information corresponding to the configured algorithm library is sent to both communication parties of the tunnel, and the algorithm description information is used for describing an encryption algorithm in the algorithm library; and if it is determined that the two communication parties negotiate a target algorithm library required to be used by the tunnel based on the algorithm description information, calling an encryption algorithm in the target algorithm library through an algorithm calling engine corresponding to the target algorithm library to encrypt data transmitted in the tunnel.

The invention discloses a resource optimization method and system for decision making of a main processor and scheduling in an audio processor. The method comprises the following steps: acquiring processor load information when the audio processor performs audio processing on multiple audio input channels and channel feature information of each audio input channel; when the processor load information meets a preset load condition, the main processor determines an adjustment result of audio processing configuration according to a preset resource optimization rule based on the processor load information and the channel feature information; the main processor generates and sends a control instruction to the audio processor according to the adjustment result; and the audio processor updates the sound mixing routing relationship in the internal scheduling according to the control instruction, and carries out switching or intensity adjustment on the processing link of each channel according to the processing algorithm configuration. According to the technical scheme, the continuity and the stability of conference audio processing are improved on the premise of not changing a hardware structure.

The invention discloses an MZI-based space division multiplexing convolution parallel computing method, which comprises the following steps of: configuring an MZI array through an in-situ training algorithm, and enabling the MZI array to respectively present target transmission characteristics corresponding to N preset convolution kernels for incoherent light with N preset wavelengths in a single control state. When calculation is carried out, an input vector to be processed is coded into incident power of light of N wavelengths and injected into the array, the superposition value of the light power of each wavelength is directly detected at an output port of the array, and the superposition value is an N-path parallel convolution result. The system of the invention correspondingly includes an incident layer, a multi-wavelength operating MZI array, and a summing layer. According to the method, the calculation parallelism degree can be improved by N times without increasing hardware overhead, the unitary matrix limitation of a traditional MZI array is broken through, the calculation density and the hardware utilization efficiency are remarkably improved, and the method has wide application prospects in the field of photon calculation.

The embodiment of the invention relates to an information aggregation processing method and device, equipment and a medium, and the method is applied to an aggregation server side and comprises the steps that an event corresponding to an abnormal file is obtained, the event is obtained by analyzing the abnormal file through an analysis server side, and the analysis server side and the aggregation server side are independent of each other and communicate with each other; an aggregation algorithm corresponding to the event type of the event is obtained, and the aggregation algorithm is dynamically configured through an aggregation engine; and performing aggregation processing on the events based on an aggregation algorithm. According to the method, the aggregation server and the analysis server which are mutually independent and communicate with each other are arranged, analysis and aggregation processing are completely separated, the influence of analysis on aggregation during large flow is avoided, the efficiency of event aggregation is effectively improved, an independent aggregation engine is introduced for dynamic configuration of an aggregation algorithm, code modification and release are not needed, and the efficiency of event aggregation is improved. The algorithm configuration cost is greatly reduced, and the algorithm configuration efficiency and flexibility are improved.

The application discloses a general neural network creation method, device, platform, medium and equipment, and belongs to the field of deep learning. The method comprises the following steps: processing parameters according to custom functions of a user, creating an algorithm configuration file, and integrating the algorithm configuration file into a model package; setting data processing logic of the algorithm configuration file by using the processing parameters in one or more algorithm configuration files, obtaining an algorithm inference library matched with the model package; creating a model package loading component according to a path of the model package; running a specified model in the model package through a model loading interface by using the model package loading component, and setting a default running device; running the specified model in the model package through the algorithm inference library in the default running device, processing user-defined data to obtain network processing data, and creating a general neural network. The efficiency of the general neural network in running the algorithm inference library to obtain a conversion result is higher, the effect is better, and the time for actual deployment of a vehicle is saved.

The application discloses an intersatellite link satellite data transmission task scheduling method and system and a storage medium, and comprises the following steps: acquiring a satellite set, a task target set and a ground station set; determining a network connection plan and a data transmission plan in each time slot; acquiring a task constraint condition, establishing an intersatellite link satellite data transmission task scheduling model, and taking the maximum complete task data value transmitted to the ground station after T time slots as the scheduling model; constructing a parallel algorithm group, acquiring an optimal algorithm and corresponding parameter configuration, and matching the optimal algorithm with a scene instance. The method constructs an integer programming model of the intersatellite link satellite data transmission task model, constructs a parallel algorithm group, searches for an optimal algorithm configuration in a machine learning mode, and overcomes the contradiction between the cost and quality of huge computing resources required by parallelism.

The invention discloses an industrial detection method based on a cloud and edge collaborative architecture, which relates to the technical field of cloud platform services and mainly comprises the following steps: creating industrial detection items according to a hierarchical structure on a cloud platform; under the task, defining or quoting at least one operator encapsulated with a single data processing function; under the task, at least one workflow is created, and each workflow defines a specific algorithm execution process for the detection scene through one or more operators selected by combining; on the cloud platform, packaging complete algorithm configuration information which is associated with a task and comprises at least one workflow and an operator combined by the workflow to generate a standardized deployment pipeline file; and receiving and analyzing the deployment pipeline file on the client device of the edge end, connecting the industrial detection device, carrying out industrial detection on the detection data collected by the industrial detection device, and outputting a detection result. According to the method, the problem of repeated development in a traditional scheme is solved, and the algorithm construction efficiency is improved.

The invention discloses a simulation setting system and method based on a distribution network digitization system, and belongs to the technical field of digitization simulation, and the simulation setting system comprises a digitization base module which is used for constructing a power distribution network digital twinborn body; the topology construction module is used for dynamically constructing a simulation subnet topology based on the target node of the simulation task and the electrical sensitivity; the model configuration module is used for configuring a simulation mathematical model for the power grid element model according to the parameters of the power grid element model; the algorithm configuration module configures calculation logic for the simulation task according to probability distribution characteristics of input variables in the simulation request; the task packaging module integrates simulation subnet topology, a simulation mathematical model, computational logic and initial parameters, and generates a simulation task package used for being called by an external simulation system; according to the method, the full-process automation and intelligentization of simulation setting are realized, and the simulation setting efficiency and the model fidelity are greatly improved.

Principal component analysis (PCA) accelerates and increases accuracy of genetic algorithms. In an embodiment, a computer generates many original chromosomes. Each original chromosome contains a sequence of original values. Each position in the sequences in the original chromosomes corresponds to only one respective distinct parameter in a set of parameters to be optimized. Based on the original chromosomes, many virtual chromosomes are generated. Each virtual chromosome contains a sequence of numeric values. Positions in the sequences in the virtual chromosomes do not correspond to only one respective distinct parameter in the set of parameters to be optimized. Based on the virtual chromosomes, many new chromosomes are generated. Each new chromosome contains a sequence of values. Each position in the sequences in the new chromosomes corresponds to only one respective distinct parameter in the set of parameters to be optimized. The computer may be configured based on a best new chromosome.

The invention relates to a supervision algorithm management method and device, computer equipment and a storage medium. The method comprises the following steps: acquiring resource use information of a plurality of servers and resource configuration demand information of a plurality of supervision algorithms; determining supervision algorithm configuration information corresponding to each server according to the resource use information of the plurality of servers and the resource configuration demand information of the plurality of supervision algorithms; the supervision algorithm configuration information corresponding to each server is sent to an operation process management and control platform; the supervision algorithm configuration information is used for enabling the operation process management and control platform to call an algorithm in the corresponding server based on the field operation information. The method provides a systematic algorithm management method, and provides a data basis for a follow-up operation process management and control platform to call algorithm execution corresponding to field operation information based on supervision algorithm configuration information corresponding to each server when processing the field operation information.

The invention discloses a media data processing method and system, a cluster, a storage medium and a program product, and relates to the technical field of cloud computing. In media data processing, the cloud management platform creates a media processing template based on the AI model and the transcoding parameter included in the algorithm configuration request. And under the condition that the cloud platform receives a first request sent by the client, the cloud platform processes first media data indicated by the first request according to the media processing template to obtain second media data. A user configures the AI model in the media task processing process by himself, so that the flexibility and applicability of the media processing template are higher. Moreover, the AI models which can be used by the user comprise the AI model preset in the cloud platform and the AI model customized by the user, so that the adaptation degree between the media processing template and the media processing scene can be improved by increasing the data size of the AI models which can be used, and the reliability of a media processing result can be ensured.

The invention discloses a dynamic negotiation-based self-adaptive secure communication method and system for an internet of things device, the internet of things device and a storage medium, and the method comprises the steps that the internet of things device exchanges a security context through a first packet, and receives and updates a security configuration issued by a cloud in real time; the Internet of Things equipment reads and starts a corresponding encryption algorithm from the local or directly selects a plaintext transmission mode according to the negotiated algorithm configuration; the Internet of Things equipment dynamically manages the secret key, and decides to execute secret key derivation or read the cache secret key for encryption and decryption according to the updating period; the Internet of Things equipment analyzes the security context, obtains corresponding secret keys and algorithm parameters according to configuration requirements, and executes encryption or plaintext transmission; when the security context is changed, the Internet of Things equipment immediately updates an encryption and decryption algorithm adapter of uplink and downlink messages; and after the message is encrypted and decrypted, packaging the message body according to a preset message packaging format and then sending the message body. According to the invention, the communication efficiency and interoperability are improved by optimizing the communication process.

The invention relates to a message rate regulation and control method and device, a chip, equipment, a network interface card, a storage medium and a program product. The method comprises the following steps: a message processing module sends a processing request of a token event to an event conversion module; the event conversion module converts the token event into an algorithm sensing event according to conversion condition information corresponding to the token event and / or a preset configuration mode of the sensing event, and sends the algorithm sensing event to an algorithm configuration execution module; the algorithm configuration execution module executes a message rate lifting algorithm corresponding to the algorithm sensing event to obtain a message sending rate corresponding to the algorithm sensing event, and sends the message sending rate to the event conversion module; and the event conversion module generates a token value corresponding to the algorithm sensing event according to the message sending rate corresponding to the algorithm sensing event, and sends the token value corresponding to the algorithm sensing event to the message processing module. By adopting the method, the message speed regulation of the token event can be realized.

The application discloses a kind of MZI-based space division multiplexing convolution parallel computing method, including, by in situ training algorithm configuration one MZI array, make it in single control state, to N kind of preset wavelength non-coherent light respectively present with N corresponding target transmission characteristics of preset convolution kernel.For calculation, the input vector to be processed is encoded as the incident power of N wavelength light and injected into the array, and the superposition value of the power of each wavelength light is directly detected at the output port, which is the N parallel convolution result.The system of the application correspondingly includes an incident layer, a multi-wavelength working MZI array and a summation layer.The application can improve the calculation parallelism by N times without increasing hardware overhead, and breaks through the unitary matrix limitation of traditional MZI array, significantly improves the calculation density and hardware utilization efficiency, and has wide application prospect in the field of photonic computing.

An embodiment of the present invention provides a method and apparatus for calculating business indicators. The method includes receiving a data query instruction, obtaining a business indicator's operational expression from an algorithm configuration library and obtaining the business values ​​of each operational factor of the operational expression from business data through a dynamic formula calculation function configured in a structured query language (SQL) data operation script, calling an operational expression engine toolkit through the dynamic formula calculation function, and executing the operational expression engine toolkit based on the operational expression and the business values ​​of each operational factor to obtain an indicator result for the business indicator. Based on this, the solution can automatically obtain the operational expression of the business indicator to be calculated by calling the dynamic formula calculation function, without pre-setting a fixed operational expression in the SQL data operation script. This allows for dynamic configuration of the business indicator's operational expression and makes the execution of the SQL data operation script more flexible.

The invention provides a configuration file management method and device, equipment and a medium, a software resource package of a vehicle end algorithm comprises an algorithm program of the vehicle end algorithm and a plurality of vehicle model folders, the plurality of vehicle model folders correspond to a plurality of vehicle model identifiers, and the configuration file management method comprises the steps of obtaining a current vehicle model identifier; a vehicle model folder corresponding to the current vehicle model identifier is searched from a plurality of vehicle model folders, and the vehicle model folder corresponding to the current vehicle model identifier comprises an algorithm configuration file matched with the algorithm program under the current vehicle model identifier; copying a vehicle model folder corresponding to the current vehicle model identifier from an original path where the software resource package is located to a set first target path; wherein the algorithm program reads the algorithm configuration file from the first target path. Therefore, through the same software resource package, the algorithm configuration files adaptive to different vehicle types are provided for the algorithm program of the vehicle end algorithm, the requirement that the same software resource package supports multiple vehicle types is met, and the software testing and maintenance cost is reduced.