Methods and system to dynamically tune an application operating on a cloud computing infrastructure. The dynamic tuning generates a dependency graph based on the applications configurable parameters while operating on cloud computing infrastructure. One or more subgraphs are selected from the dependency graph which include a subset of the parameters of the dependency graph, and corresponding subgraph variations are created. The subgraphs and subgraph variants undergo a simulation used to identify a subset of the subgraphs with viable operation. This subset of subgraphs undergoes probabilistic modelling to generate a ranked set of configurations, which are then implemented in test replicas of the application and evaluated against the production replica. Once the test replicas are determined to outperform the production replica, they are implemented within the production replica.
Disclosed are a data processing method and apparatus, which relate to the technical field of cloud computing. The method is applied to a data backup system. The data backup system pre-stores a data processing policy for target data, wherein the data processing policy is used for indicating a target performance for processing the target data, and the target performance is the performance promised to a user by the data backup system. On this basis, after detecting a target processing task that indicates the processing of the target data, the data backup system executes the processing task by means of M execution units; during the execution of the processing task, the performance of the data backup system for processing the target data is acquired; and when the performance of the data backup system for processing the target data does not reach the target performance promised to the user, the number of execution units for executing the processing task in parallel is increased, so as to improve the performance of the data backup system for processing the target data, such that the data backup system provides for the target data the target performance promised to the user, and the data backup system thus provides for the user the performance that can be promised.
G06F 11/14 - Détection ou correction d'erreur dans les données par redondance dans les opérations, p. ex. en utilisant différentes séquences d'opérations aboutissant au même résultat
A parallel-strategy search method and apparatus, which are used for improving the flexibility of parallel-strategy search. The method comprises: a parallel-strategy search system acquiring a plurality of parallel strategies, each of the plurality of parallel strategies being used for indicating a parallel manner for training an AI model; providing an input interface, the input interface being used for acquiring input content, the input content being used for indicating at least one strategy screening rule, and the at least one strategy screening rule being used for screening the plurality of parallel strategies; on the basis of at least one strategy screening rule, screening the plurality of parallel strategies, so as to determine a plurality of candidate parallel strategies among the plurality of parallel strategies; and evaluating the plurality of candidate parallel strategies to determine a target parallel strategy, the target parallel strategy being a parallel strategy to be used for training the AI model.
G06F 18/214 - Génération de motifs d'entraînementProcédés de Bootstrapping, p. ex. ”bagging” ou ”boosting”
G06F 30/27 - Optimisation, vérification ou simulation de l’objet conçu utilisant l’apprentissage automatique, p. ex. l’intelligence artificielle, les réseaux neuronaux, les machines à support de vecteur [MSV] ou l’apprentissage d’un modèle
4.
DATA TRANSMISSION METHOD BASED ON REMOTE DIRECT MEMORY ACCESS, AND RELATED DEVICE
Disclosed in the present application are an RDMA-based data transmission method and a related device, which can reduce, to a certain extent, the time cost and performance overhead required in the overall data transmission process. The method in the present application comprises: a first processor providing a memory block to a first RDMA network interface card, wherein the memory block comprises a plurality of sub-memories, including a first sub-memory and a second sub-memory; after a second RDMA network interface card sends a packet to the first RDMA network interface card, which packet carries first data generated by a second processor, the first RDMA network interface card generating a first address for the first data; then, the first RDMA network interface card storing the first address in the first sub-memory and storing the first data in the second sub-memory, wherein the first address is used for indicating the location of the first data in the second sub-memory; and subsequently, the first RDMA network interface card notifying the first processor to read the first address from the first sub-memory, so as to read the first data from the second sub-memory on the basis of the first address, thereby realizing data transmission.
A task processing method, applied to an operation and maintenance platform. The operation and maintenance platform comprises an event queue and a task executor. The method comprises: the event queue receives at least one operation and maintenance task of an application, and adds task events corresponding to the at least one operation and maintenance task to the event queue; the task executor fetches a first task event from the event queue, calls a first execution unit in an execution unit resource pool, executes a first subtask, delivers an operation of the first subtask on a resource to the resource, and releases the first execution unit; the event queue receives a first resource event sent by the resource; and on the basis of the state of the resource described by the first resource event, the task executor determines a next task or a next subtask to be processed, and executes the next task or the next subtask. The method is event-centric, and an operation and maintenance task, a subtask, and the state of a resource are processed as an event, so that a task is processed from a smaller granularity. Moreover, an execution unit is not blocked by a waiting state of an operation and maintenance task, so that the resource utilization is improved.
The present application provides a screen generation method and apparatus. The method comprises: acquiring a resource release time interval; on the basis of the resource release time interval, periodically releasing resources in a resource pool, the resources being used for generating a screen; and when a screen generation request is received, using target resources released from the resource pool to generate a target screen requested to be generated by the screen generation request. According to the method, screen stuttering can be prevented or reduced while preventing screen tearing.
The present application discloses an elastic IP address configuration method based on a hybrid cloud scenario, and a public cloud system, which can improve the public network access experience of tenants to a certain extent. The method of the present application comprises: a tenant can provide, to a cloud management platform, access point information of an elastic IP address used by an off-cloud resource, and cloud gateway creation information, wherein the access point information is used for indicating a target cloud data center where an access point of the elastic IP address is located, and the cloud gateway creation information is used for indicating a cloud gateway which is to be created and is bound to the elastic IP address; the cloud management platform can create said cloud gateway in the target data center, and establish a network connection channel between said cloud gateway and the off-cloud resource; and when the remaining tenants in the public network send, to the access point, a first packet of which a destination address is the elastic IP address, the access point can notify said cloud gateway to send the first packet to the off-cloud resource of the tenant by means of the network connection channel, so that the off-cloud resource of the tenant processes the first packet.
The present application provides a cross-cluster application access method, a gateway apparatus, and a related product, capable of realizing communication between cross-cluster applications and reducing the resource occupation of a network proxy component in a cluster. The method is applied to a multi-cluster system, the multi-cluster system comprises an application layer gateway, a first K8s cluster and a second K8s cluster, the first K8s cluster comprises a network proxy component and a first application, and the second K8s cluster comprises a second application. Specifically, the network proxy component intercepts an access request sent by the first application, and forwards the access request on the basis of a routing rule matching the request. When the routing rule instructs to perform application layer processing on the access request, the access request is sent to the application layer gateway, so that the application layer gateway performs application layer processing on the access request, and forwards the processed access request to the second application. When the routing rule instructs to perform transport layer processing on the access request, the access request is sent to the second application.
A method includes: receiving downsampling policies at a plurality of levels that are defined by a user, where a downsampling policy at each level indicates to downsample data in the plurality of time shards based on a defined sampling time interval after raw data is stored in a plurality of time shards and a defined time interval has elapsed; determining a first downsampling policy in the downsampling policies at the plurality of levels based on a first time interval that has elapsed after first raw data is stored in a first time shard, where the longer first time interval indicates a longer sampling time interval of the first downsampling policy; obtaining first data currently stored in the first time shard; and downsampling the first data according to the first downsampling policy, to obtain second data.
A method comprises: obtaining, by a source terminal, first media data of a real person and comprising a first video or a first audio of the real person; extracting, by the source terminal and from the first media data, feature data for describing the real person and comprising a facial feature, a morphological feature, or a voice feature of the real person, wherein the facial feature and the morphological feature are from the first video, and wherein the voice feature is from the first audio; sending, by the source terminal and to a cloud server, the feature data; generating, by the cloud server and based on the feature data and a multimodal AI-driven model, second media data of a digital human and comprising a second video or a second audio of the digital human; and sending, by the cloud server and to a destination terminal, the second media data.
This application relates to the field of cloud computing technologies and provides a cloud disk service change recommendation method and apparatus. An example method is applied to a cloud management platform configured to manage infrastructure that provides a cloud service. The infrastructure includes a storage pool including storage space of a plurality of cloud servers. The storage pool is virtually divided into a plurality of cloud disks. The method includes obtaining running data of a first cloud disk in the plurality of cloud disks. The running data of the first cloud disk includes a flow control parameter and capacity usage. The method further includes determining a change policy for the first cloud disk based on the running data of the first cloud disk and providing a recommendation interface for a tenant of the first cloud disk. The recommendation interface includes the change policy.
Disclosed are a media data processing method and system, and a cluster, a storage medium and a program product, which relate to the technical field of cloud computing. During media data processing, a cloud management platform creates a media processing template on the basis of an AI model and transcoding parameters comprised in an algorithm configuration request; and when receiving a first request sent by a client, the cloud platform processes first media data indicated by the first request on the basis of the media processing template, so as to obtain second media data. A user configures an AI model in a media task processing process by themselves, such that a media processing template has higher flexibility and applicability. In addition, AI models available to a user comprise AI models preset in a cloud platform and AI models customized by the user, such that by increasing the data volume of available AI models, the degree of adaptation between the media processing template and a media processing scenario can be improved, thereby ensuring the reliability of a media processing result.
H04L 65/752 - Gestion des paquets du réseau multimédia en adaptant les médias aux capacités du réseau
H04N 21/2662 - Contrôle de la complexité du flux vidéo, p. ex. en mettant à l'échelle la résolution ou le débit binaire du flux vidéo en fonction des capacités du client
H04N 21/637 - Signaux de commande émis par le client et dirigés vers les éléments du serveur ou du réseau
13.
RESOURCE ISOLATION METHOD AND APPARATUS, AND COMPUTING DEVICE CLUSTER
A resource isolation method, comprising: acquiring the usage amount of computing resources of each tenant among a plurality of tenants; on the basis of the usage amount of computing resources of each tenant, obtaining the total usage amount of computing resources of the plurality of tenants; and when the total usage amount of computing resources of the plurality of tenants is less than a first lower resource threshold value, increasing the number of threads for the plurality of tenants, or keeping the number of threads for the plurality of tenants unchanged. In the present application, on the basis of the method described in the present application, the usage amounts of computing resources of different tenants can be continuously acquired, and the number of threads invoked by different tenants can be dynamically adjusted on the basis of the usage amounts of computing resources of the tenants, so as to adjust the usage amounts of computing resources of different tenants. Thus, a plurality of tenants can fully use computing resources in a system, thereby improving the utilization rate of computing resources of a processor.
G06F 21/53 - Contrôle des utilisateurs, des programmes ou des dispositifs de préservation de l’intégrité des plates-formes, p. ex. des processeurs, des micrologiciels ou des systèmes d’exploitation au stade de l’exécution du programme, p. ex. intégrité de la pile, débordement de tampon ou prévention d'effacement involontaire de données par exécution dans un environnement restreint, p. ex. "boîte à sable" ou machine virtuelle sécurisée
14.
LOG DETECTION METHOD AND APPARATUS, COMPUTING DEVICE CLUSTER, MEDIUM AND PROGRAM PRODUCT
The present application provides a log detection method, a log detection apparatus, a computing device cluster, a computer-readable storage medium, and a computer program product. The method comprises: identifying anomaly sample logs from among sample logs; on the basis of the anomaly sample logs, generating an anomaly log template, the anomaly log template containing common features of the multiple anomaly sample logs; converting the anomaly log template into a regular expression; and, by means of the regular expression, determining whether a log under test is an anomaly log. The present application can comprehensively detect anomaly logs, and consume fewer computing resources.
Disclosed in the embodiments of the present application is a method for determining an initial solution. In the method, preferable preset solutions of a preset mathematical programming model in one or more historical service scenarios can be pre-collected by means of, for example, an uncertainty algorithm, so as to provide deterministic preferable initial solutions corresponding to the mathematical programming model in different historical service scenarios, thereby preventing the uncertainty of solving results caused by the uncertainty algorithm, etc., from being exposed to a user. In this way, in an actual application process, the preset solutions corresponding to the preset mathematical programming model, the similarity between which and a first mathematical programming model meets a specified condition, can be efficiently and quickly obtained and can be used as one or more target initial solutions of the first mathematical programming model, such that the user can be provided with a reproducible and relatively appropriate initial solution required by the first mathematical programming model corresponding to a current service, thereby effectively improving the solving efficiency of the mathematical programming model and meeting the reproducibility requirements of the user.
A cloud service-based data forwarding method applying to a cloud platform, the cloud platform includes a first cloud system, the first cloud system includes a cloud computing device and a forwarding device, and the first cloud system and an extranet system are connected to each other through an extranet gateway. The method includes: The forwarding device obtains a target packet sent by the cloud computing device to the extranet system, and obtains a forwarding attribute of the cloud computing device. When the forwarding attribute of the cloud computing device is a first attribute, the forwarding device forwards the target packet by using the extranet gateway as a next hop of the target packet. When the forwarding attribute of the cloud computing device is a second attribute, the forwarding device forwards the target packet to the extranet gateway based on a routing relationship of the forwarding device.
An SQL defense method includes an SQL defense rule configuration unit configuring corresponding SQL defense rules for a plurality of SQL engines through a unified entrance, where each SQL defense rule includes an effective SQL engine and a corresponding execution action, and the SQL defense rules corresponding to the plurality of SQL engines have a same format; and the SQL defense rule configuration unit sending each configured SQL defense rule to a corresponding SQL engine based on the effective SQL engine in each SQL defense rule, so that the SQL engine performs defense on a received SQL statement according to the obtained SQL defense rule.
Provided in the present invention are a data forwarding method and apparatus based on application network technology. The method is applied to a first acceleration node in an application network deployed in a physical network, and comprises: acquiring a service attribute table sent by a controller, which service attribute table comprises tenant access identifiers and forwarding planes corresponding to the tenant access identifiers, wherein a forwarding plane is determined on the basis of a quality of service requirement configured by a tenant corresponding to each tenant access identifier, and comprises at least a portion of acceleration nodes in the application network; acquiring a data request from a terminal, wherein the data request comprises a target tenant access identifier; on the basis of the target tenant access identifier and the service attribute table, determining a target forwarding plane; and sending the data request to a second acceleration node in the target forwarding plane. In summary, by means of dividing a physical network into virtual networks of different quality of service requirements, network resources of the physical network can be fully utilized, thereby improving the utilization rate of the network resources; and adaptation to diverse quality of service requirements is realized, thereby improving the user experience.
H04L 67/61 - Ordonnancement ou organisation du service des demandes d'application, p. ex. demandes de transmission de données d'application en utilisant l'analyse et l'optimisation des ressources réseau requises en tenant compte de la qualité de service [QoS] ou des exigences de priorité
The present application provides a code problem processing method and device. The method comprises: acquiring a code problem and a code repository, wherein the code problem indicates a code problem existing in the code repository, and the code repository comprises N code files; on the basis of the code problem and the code repository, determining a target intelligent agent network, wherein the target intelligent agent network comprises M intelligent agents and connection edges, the M intelligent agents correspond to P code files and are used for processing the P code files, the connection edges are used for connecting intelligent agents having an association relationship among the M agents, the P code files are code files among the N code files, and P is a positive integer less than or equal to N and greater than or equal to M; updating the code repository on the basis of a result obtained by the M intelligent agents collaboratively processing the P code files, so as to obtain a target code repository. In the present application, M intelligent agents collaboratively process P code files; thus, cross-file editing is achieved, and the processing capability for complex code problems can be improved.
The present application relates to the field of communications, and discloses a tenant management method and apparatus, and a storage medium. The method is applied to a database system, and the database system comprises a master node and at least one first standby node. The method comprises: receiving a creation request, the creation request comprising a target number of resources required by a tenant to be created and service attribute information of the tenant; acquiring a first value on the basis of the service attribute information, the first value being the minimum number of resources that the tenant needs to occupy on the master node, and the first value being less than or equal to the target number of resources; and creating a tenant on the basis of the first value, resources occupied by the tenant including first resources on the master node and second resources on the at least one first standby node, the number of first resources being greater than or equal to the first value and less than or equal to the target number of resources, and the number of second resources being determined on the basis of a difference between the target number of resources and the number of first resources. The present application can improve the flexibility of allocating and managing resources for a tenant.
The present disclosure relates to the technical field of databases. Provided are a data query method and apparatus, and a storage medium and a computer program product. The method is applied to a database, wherein the database comprises an SQL sub-database and a graph sub-database, the database is configured with a plurality of operator APIs, and the plurality of operator APIs include an operator API that can be invoked by an SQL operator, and an operator API that can be invoked by a graph operator. The method comprises: upon receiving a data query instruction, generating an execution plan, the execution plan comprising at least one operator to be executed; and if the operator to be executed is a graph operator, directly invoking an operator API corresponding to the operator to be executed, in order to execute an operation on a graph sub-database, thereby eliminating the need to indirectly implement queries on the graph sub-database by means of invoking an SQL operator, and thus effectively improving the data query efficiency.
The present disclosure relates to the technical field of computers, and provides a traffic playback method and apparatus, a storage medium, and a computer program product. In the present disclosure, during traffic playback, for recorded statements, determinations in terms of a parallel execution condition are performed in the order of end times, so that different statements operating on the same data can only be executed serially, and statements having earlier end times are executed first. In this way, during target database playback, it can be ensured that the execution order of a plurality of operations on the same data is the same as the actual execution order during source database recording, thereby ensuring data consistency during target database playback and source database recording as much as possible, further ensuring as much as possible that the complexity of executing transactions in the target database is consistent with the actual complexity of executing transactions in the source database, and thus reducing test load distortion.
The present disclosure relates to the technical field of computers. Provided are a method and apparatus for creating an AI model, and a database system. In the present disclosure, filtering processing is performed on a first attribute set carried in a model creation instruction. In this way, since attributes used for creating an AI model are reduced by means of filtering processing, during the creation of the AI model, the amount of data that a database server needs to transmit to an AI server is reduced, such that the data transmission efficiency is improved, thereby improving the efficiency of AI model creation.
The present application provides a video parameter adjustment method and apparatus. The method comprises: acquiring a first video segment, the first video segment comprising a digital person, and the digital person being generated by a neural network model; determining a video quality monitoring index on the basis of the first video segment, the video quality monitoring index comprising at least one of the following: the accuracy of the mouth shape of the digital person, the degree of synchronization between the picture and sound of the digital person, and the degree of consistency of the figure of the digital person; and determining a parameter of the neural network model on the basis of the video quality monitoring index, the parameter being used for generating the digital person in a second video segment. In the described method, a quality monitoring index related to a digital person in a livestreaming video can be monitored and corresponding feedback adjustment settings are executed, thereby improving the quality of the livestreaming video of the digital person.
A data storage method and a related device. The method is applied to a cloud management platform. The cloud management platform is configured to manage an infrastructure that provides a cloud service, and the infrastructure includes a plurality of distributed storage devices. The data storage method includes: receiving first request information of a tenant, where the first request information is for obtaining a first file; determining, based on the first request information, a first data block set in which the first file is located, where the first data block set includes one or more files, attributes of the one or more files are the same, and the first data block set is stored in one storage device; determining, based on the first data block set, a first storage device in which the first data block set is located; and reading the first file from the first storage device.
G06F 11/14 - Détection ou correction d'erreur dans les données par redondance dans les opérations, p. ex. en utilisant différentes séquences d'opérations aboutissant au même résultat
G06F 11/10 - Détection ou correction d'erreur par introduction de redondance dans la représentation des données, p. ex. en utilisant des codes de contrôle en ajoutant des chiffres binaires ou des symboles particuliers aux données exprimées suivant un code, p. ex. contrôle de parité, exclusion des 9 ou des 11
A data access method includes receiving a file creation request, where the file creation request carries a first file name and a first file identifier of a parent directory of a first file; obtaining metadata of the parent directory that is of the first file and that corresponds to the first file identifier; creating the first file in the parent directory based on the metadata of the parent directory of the first file, and allocating a second file identifier to the first file, where the first file is a file or a directory in the directory tree; and recording a correspondence between the second file identifier and the first file name into a file index table.
A container management method includes a cloud management platform that obtains, from a container configuration interface, configuration information that is of a to-be-created container, where the configuration information indicates a requirement of the tenant on the container. The cloud management platform obtains status information of a plurality of resource pools, where the status information indicates performance of the plurality of resource pools. The cloud management platform selects, from the plurality of resource pools based on the configuration information and the status information, a first resource pool whose performance meets the requirement. The cloud management platform determines, from a plurality of prewarming levels, a first prewarming level that matches the tenant, and if a first cloud instance corresponding to the first prewarming level exists in a plurality of cloud instances in the first resource pool, the cloud management platform creates the container in the first cloud instance.
G06F 9/455 - ÉmulationInterprétationSimulation de logiciel, p. ex. virtualisation ou émulation des moteurs d’exécution d’applications ou de systèmes d’exploitation
H04L 47/70 - Contrôle d'admissionAllocation des ressources
28.
Method for Solving System of Linear Equations and Related Device
In a method for solving a system of linear equations, a computing device or a computing device cluster may obtain description information input by a user, where the description information includes information about a system of linear equations; perform inference on the description information using an artificial intelligence (AI) model to obtain a first initial solution corresponding to the system of linear equations; determine a target initial solution based on the first initial solution; and iteratively solve the system of linear equations based on the target initial solution. In the method, the AI model can identify features of the system of linear equations based on the input description information.
A reasoning request scheduling method, comprising: a reasoning routing apparatus acquires a first reasoning request, and when the number of reasoning requests to be processed is greater than or equal to a threshold defined in a flow control policy, the reasoning routing apparatus checks the remaining resources of at least one reasoning server in a reasoning resource pool; the reasoning routing apparatus predicts consumed resources of the first reasoning request; and when the consumed resources are less than the remaining resources, the reasoning routing apparatus determines a target reasoning server corresponding to the first reasoning request, and schedules the first reasoning request to the target reasoning server, wherein the target reasoning server is one or more reasoning servers in the reasoning resource pool of which the remaining resources satisfy the first reasoning request, and a language model deployed in the target reasoning server matches the first reasoning request. In the method, when the number of requests is greater than a flow control threshold, reasoning requests are dynamically scheduled on the basis of the remaining resources of reasoning servers dynamically fed back by a reasoning resource pool side, so as to use reasoning resources to the greatest extent, thereby improving the utilization rate of the reasoning resources.
The present application provides a retrieval-augmented generation method. The method is applied to an integrated development environment (IDE), the IDE comprises a plurality of program entities, and the IDE is used for executing at least one operation. The method comprises: acquiring a natural language instruction, the natural language instruction being used for instructing to execute a target operation on a first program entity; determining a plurality of first syntax units of the natural language instruction on the basis of the natural language instruction, wherein the plurality of first syntax units are used for indicating the type of the target operation and the first program entity; determining an identifier of the target operation on the basis of the plurality of first syntax units and a plurality of second syntax units of metadata of the operation, wherein a program entity indicated by the second syntax units corresponding to the target operation is the first program entity, and an identifier of an operation indicated by the second syntax units corresponding to the target operation is the identifier of the target operation; and executing the target operation on the first program entity on the basis of the identifier of the target operation.
Example data transmission methods and apparatuses are provided. In one example method, a media service apparatus determines first media data displayed on a screen of a first terminal, where the first media data includes text content in a first language. If a second terminal selects to use a second language, the media service apparatus determines second media data corresponding to the first media data, where the second media data includes text content in the second language, and the first terminal and the second terminal access a same meeting. The media service apparatus controls screen display content of the second terminal based on the second media data.
H04L 12/18 - Dispositions pour la fourniture de services particuliers aux abonnés pour la diffusion ou les conférences
G06F 40/58 - Utilisation de traduction automatisée, p. ex. pour recherches multilingues, pour fournir aux dispositifs clients une traduction effectuée par le serveur ou pour la traduction en temps réel
32.
Metadata Processing Method and System, and Computing Device
A metadata processing method includes: A data production cluster generates metadata of shared data. The data production cluster stores the shared data into a shared storage. The data production cluster receives a metadata operation instruction, and determines target metadata based on the metadata operation instruction, where the target metadata is metadata of target shared data. The data production cluster sends the target metadata to a data consumption cluster. The data consumption cluster reads the target shared data from the shared storage based on the target metadata.
A validation method includes that a plurality of decentralized application nodes aggregate, using a key aggregation algorithm, public keys held by the plurality of decentralized application nodes, to generate a public address of a blockchain network, and separately transfer tokens to the public address. The plurality of decentralized application nodes transfers the tokens at the public address to a staking account, generates a plurality of staking private keys, and allocates the plurality of staking private keys to a distributed validator technology (DVT) network. A plurality of DVT nodes signs a transaction using the staking private keys respectively held by the plurality of DVT nodes, to obtain an aggregated signature, and then provides the aggregated signature for a consensus node, to participate in validation of the transaction.
A method a for generating content by using a model including: A client receives a content generation instruction sent by a user. The client inputs the content generation instruction into a first portion of the model, to enable the first portion of the model to output an intermediate result of generating the target content by the model. A serving end receives the intermediate result sent by the client, and inputs the intermediate result into a second portion of the model, to enable the second portion of the model to output the target content. The serving end sends the target content to the client.
H04L 41/16 - Dispositions pour la maintenance, l’administration ou la gestion des réseaux de commutation de données, p. ex. des réseaux de commutation de paquets en utilisant l'apprentissage automatique ou l'intelligence artificielle
H04L 47/19 - Commande de fluxCommande de la congestion au niveau des couches au-dessus de la couche réseau
35.
Distributed Training Method, System, and Apparatus
A distributed training method includes a management node that obtains a network topology, where the network topology includes a connection relationship between a core switch and a computing node in a computing cluster. Then, the management node determines a communication plan between N computing nodes based on the network topology, where the N computing nodes are computing nodes that are in the computing cluster and that are configured to train a target model in a distributed manner.
University of Science and Technology of China (Chine)
Inventeur(s)
Li, Xijun
Li, Jianshu
Wang, Zhihai
Zeng, Jia
Wang, Jie
Abrégé
A method and an apparatus for solving an objective function based on a cloud computing technology are disclosed. The method is applied to a cloud management platform. The method includes: obtaining user data uploaded by a tenant; determining, based on a artificial intelligence model, a first cutting plane set corresponding to the user data; performing solving based on the first cutting plane set and an objective function included in the user data to obtain a solving result of the objective function; and providing the solving result for the tenant or saves the solving result in a storage module of the cloud management platform. In the method, a cutting plane set for the objective function in the user data is selected from the M candidate cutting planes based on an artificial intelligence (AI) model to improve solving performance of the objective function.
There is provided a method for selecting an access plan in a cost-based bottom-up optimizer, where the access plan may comprise Bloom filters. In one embodiment, partial access plans with Bloom filter apply nodes are not joined with other partial plans unless the other partial plan can resolve the partial plan with the Bloom filter apply node. This reduces the search space for an optimal access plan to a manageable size and improves query completion run-times.
A test case generation method. The method comprises: acquiring an object under test, said object being software under test or a code snippet within said software; performing software analysis on said object to obtain engineering dependency information and context of said object, the context comprising at least one of a file-level context or a project-level context; then generating a test header of a test case of said object on the basis of the engineering dependency information and the context, the test header being an initial part of the test case, and the initial part comprising information other than a test method; next, performing continuation writing on the test header of the test case by means of a language model to obtain the test case; and presenting the test case to a user. In the method, by combining software analysis and prompt engineering, a test header is provided for a language model to generate a test case, and continuation writing is performed on the test header by means of the language model instead of generating a test case by the language model from scratch, thereby improving the quality and effectiveness of test cases generated by the language model.
Disclosed in the embodiments of the present application is a cloud service-based design code generation method, the method being applied to a cloud management platform. The method can not only acquire from a configuration interface an explicit design intention input by a user, but also query a knowledge graph for target design knowledge associated with the current explicit design intention of the user so as to capture an implicit design intention of a target design object. Thus, the explicit design intention and the target design knowledge from the knowledge graph can be combined to guide a large language model in efficiently and accurately generating target design code that meets a design requirement of the user, thereby intelligently generating the target design code so as to accurately and efficiently implement a design task on the basis of the target design code to obtain the target design object.
The present application belongs to the field of communications. Disclosed are a data center and a communication method for a data center. The method comprises: a first computing device sending M first access requests to a second computing device by means of a first path; when a bandwidth utilization rate of the first path exceeds a threshold, the first computing device sending N second access requests to the second computing device by means of the first path, and sending H third access requests to a third computing device by means of a second path; and the third computing device sending the H third access requests to the second computing device by means of a third path. The present application can increase a data sending rate.
Provided are a method for identifying a lineage relationship between parameters, and a related device. The method comprises: collecting a plurality of parameters of interfaces invoked by traffic; acquiring information of each parameter, wherein the information of each parameter comprises the value of each parameter and/or the memory address of each parameter; and on the basis of the value of each parameter and/or the memory address of each parameter, identifying parameters among the plurality of parameters that have a lineage relationship. Therefore, the accuracy of identifying a lineage relationship between parameters is improved, and the cost of identifying a lineage relationship between parameters is reduced.
A storage node and a request processing method. The storage node comprises a network module, a data plane module, a cache, and a storage medium. The network module is used for acquiring an input/output (IO) write request and sending the IO write request to the data plane module, wherein the IO write request comprises first data to be written. An IO module in the data plane module is used for writing the first data comprised in the IO write request into the cache by means of a data transmission bus in the storage node; the IO module in the data plane module is further used for writing the first data stored in the cache into the storage medium; and the IO module in the data plane module is further used for generating a first response message corresponding to the IO write request and sending the first response message to the network module. The storage node can reduce a storage access delay and improve storage performance.
The present application discloses a data processing method and apparatus, a computing device cluster, and a storage medium. The method comprises: performing feature extraction on data inputted by a user, to obtain query (Q) data, key (K) data, and value (V) data; on the basis of the number N of processing devices, respectively segmenting the Q data, the K data, and the V data to obtain 2N Q data blocks, 2N K data blocks, and 2N V data blocks; and controlling the N processing devices to perform 2N attention calculation operations on the basis of the 2N Q data blocks, the 2N K data blocks, and the 2N V data blocks, so as to obtain attention calculation results of the N processing devices, wherein each processing device performs two sets of attention calculations on the basis of two Q data blocks, two K data blocks, and two V data blocks respectively during each attention calculation operation. In the present application, attention-based long sequence processing can be implemented on a plurality of processing devices by means of two data streams, thereby improving the efficiency of sequence parallel processing.
A metadata processing method includes a data production cluster that generates metadata of shared data. The data production cluster stores the shared data and the metadata into a shared storage. The shared storage generates path information of the metadata. A data consumption cluster obtains the path information of the metadata from the shared storage, obtains the metadata from the shared storage based on the path information of the metadata, and obtains the shared data from the shared storage based on the metadata.
A workflow debugging method is applied to a cloud platform. The cloud platform is configured to provide a workflow design service. The method includes: obtaining, from a design interface of a workflow, a debugging requirement related to a first workflow, where the debugging requirement indicates a to-be-debugged flow fragment in the first workflow; sending related information about the flow fragment to a sandbox runtime environment; and debugging the flow fragment in the sandbox runtime environment.
Provided are a data processing method, a data processing apparatus, and a computing device cluster. The method comprises: acquiring first task information of a data processing platform, wherein the first task information comprises a plurality of job identifiers and upstream and downstream information corresponding to each of the plurality of job identifiers; on the basis of table lineage information of the data processing platform, updating upstream and downstream information corresponding to a first job identifier, wherein the first job identifier belongs to the plurality of job identifiers; updating the first task information on the basis of the updated upstream and downstream information corresponding to the first job identifier, so as to obtain second task information; and performing data processing on the basis of the second task information. By means of the method, unnecessary upstream and downstream relationships can be removed from first task information, thereby improving the efficiency of data processing.
This application discloses a key processing method. The key processing method is applied to a management system. The management system is connected to a service system, and manages cipher devices by using key indexes. The management system obtains a first key operation instruction sent by a first service system in the service system, and sends a second key operation instruction to a first cipher device based on first key information. The second key operation instruction includes a first key index and a first operation type. The first key index is an index of a first key corresponding to the first key information. The second key operation instruction instructs the first cipher device to perform an operation of the first operation type on the first key. The management system obtains a first key operation result fed back by the first cipher device.
The present application belongs to the technical field of databases. Provided are a data synchronization method and a related device. The method implements data synchronization between a source database cluster and a target database cluster on the basis of logical replication, wherein during the execution of an existing transaction, the source database cluster individually starts an autonomous transaction to submit a new generated version of metadata, and after the autonomous transaction is submitted, the target database cluster performs logical decoding on a physical log of the transaction on the basis of the new version of metadata and then performs playback, so as to achieve data synchronization. Compared with the prior art, the method realizes decoupling between physical layers of two database clusters, and supports real-time logical decoding of transaction logs during the execution of hybrid transactions.
G06F 16/27 - Réplication, distribution ou synchronisation de données entre bases de données ou dans un système de bases de données distribuéesArchitectures de systèmes de bases de données distribuées à cet effet
Disclosed in the present application is an application management method. The method comprises: an interaction apparatus receiving a user query; an agent routing and orchestration apparatus identifying, on the basis of the query, an agent requested by a user in the query, and when no agent is requested in the query, retrieving a target agent matching the query, assigning tasks to the target agent, and determining an execution order or a dependency relationship of the tasks; the agent routing and orchestration apparatus generating a standard operating procedure (SOP) by means of a language model on the basis of the execution order or dependency relationship of the tasks and the target agent; and an agent execution apparatus loading the target agent on the basis of the SOP, and executing an operation node in the target agent to obtain a response. By means of the method, intent recognition can be performed on a user query, a target agent matching the query is retrieved, tasks are assigned to the target agent, and static or dynamic SOPs can be generated by means of task orchestration, thereby meeting diverse business requirements and complex business environments, and thus the method has higher flexibility and adaptability.
A phase-change liquid-cooling architecture for a data center, and a machine room. The architecture comprises a machine room, a cabinet, and a condensing unit, a water supply pipe and a water return pipe which are fixed to the machine room, wherein the cabinet and the condensing unit are located in a sealed device cavity, a liquid-accommodating region capable of accommodating a phase-change working medium is provided in a cabinet body thereof, and the cabinet further comprises an exhaust opening and a liquid return interface which are respectively in communication with the liquid-accommodating region; the condensing unit is fixedly arranged on the side of the cabinet, a liquid collecting tank thereof is located below a condensing pipe, the liquid collecting tank is provided with a top opening and a liquid discharge interface, and the projections of the top opening and the condensing pipe on the floor of the machine room at least partially overlap; and a vapor-phase working medium in the cabinet is discharged into the device cavity of the machine room and then comes into contact with the surface of the condensing pipe on the side for heat exchange, and a liquid return channel is provided between the liquid discharge interface of the liquid collecting tank and the liquid return interface of the cabinet. On the basis of the architecture design in which the condensing unit is decoupled from the cabinet, the space of the machine room can be fully utilized for the layout of the condensing unit, thereby realizing the independent evolution of a CDU and the cabinet and effectively reducing the overall sealing cost.
The present application provides a resource allocation method and a related device. The method is applied to a cloud management platform. The method comprises: acquiring historical resource usage of a tenant; on the basis of the historical resource usage, determining a first resource utilization rate, wherein the first resource utilization rate is used for indicating the resource utilization rate of a first resource pool of the tenant within a first historical time period; on the basis of N container groups to be run and the first resource utilization rate, determining M computing nodes, wherein the M computing nodes are used for running the N container groups, and M is less than N; and providing the M computing nodes to the tenant. According to the method, on the basis of the resource utilization rate of a resource pool of a tenant within a historical time period and N container groups to be run, one computing node can be allocated to a plurality of container groups of the tenant, thereby determining M computing nodes for running the N container groups and providing the M computing nodes to the tenant, thus reducing resource waste in computing nodes and reducing the difficulty in managing the computing nodes.
Provided in the present application are a vulnerability analysis method and apparatus. The method comprises: acquiring issue ticket data related to code of a software project; extracting vulnerability information of potential vulnerabilities from the issue ticket data, the vulnerability information of each potential vulnerability indicating at least one of the following: a vulnerability description of the potential vulnerability, the influence range of the potential vulnerability, or a trigger condition for the potential vulnerability; classifying the potential vulnerabilities by means of a first artificial intelligence (AI) model on the basis of the vulnerability information of the potential vulnerabilities, so as to obtain the types of the potential vulnerabilities, the types of the potential vulnerabilities being used for generating early warning information; and outputting the early warning information by means of an interaction client. The solution of the embodiments of the present application helps to improve the efficiency of vulnerability analysis, thereby also helping to reduce security risks.
G06F 21/57 - Certification ou préservation de plates-formes informatiques fiables, p. ex. démarrages ou arrêts sécurisés, suivis de version, contrôles de logiciel système, mises à jour sécurisées ou évaluation de vulnérabilité
53.
AUTONOMOUS DRIVING DEVELOPMENT METHOD AND AUTONOMOUS DRIVING DEVELOPMENT PLATFORM
Provided in the present application are an autonomous driving development method and an autonomous driving development platform. The method comprises: providing a first interface and a second interface, wherein the first interface is used for connecting to sensor systems and a chassis drive-by-wire system of a vehicle, and the second interface is used for connecting to a development and application system of the vehicle; on the basis of the first interface, receiving operating state information of a vehicle and drive control information thereof, wherein the operating state information of the vehicle is collected by means of various sensor systems mounted on the vehicle, and the drive control information of the vehicle is collected by means of the chassis drive-by-wire system mounted on the vehicle; and parsing the operating state information of the vehicle and the drive control information thereof, so as to realize unified information conversion between the first interface and the second interface, wherein the parsed operating state information of the vehicle and the parsed drive control information thereof are used by a user to develop an autonomous driving path of the vehicle on the development and application system of the vehicle. The autonomous driving development platform provided in the present application uses a standardized development method, thereby eliminating the need to take into account the issues regarding service scenarios and vehicle model capabilities.
Embodiments of the present application provide a method and an apparatus for data processing. The method includes: obtaining input parameters related to a PDE to be solved, where the PDE is used to describe a simulation task; and generating an initial guess for a PDE solver according to an initial guess generation model, where the initial guess generation model is an AI model, an input of the initial guess generation model is determined based on the input parameters, and the initial guess is used to calculate a solution of the PDE through the PDE solver. The technical solutions may be beneficial for facilitating the acceleration of PDE solvers while ensuring the accuracy of the solution.
A model inference method and apparatus are disclosed, and relates to the field of machine learning technologies. A client and a server use respective deployed models to process different parts of user data, to obtain respective output results. In addition, the client obtains the output result of the server, and obtains an inference result based on the output results of the server and the client. Compared with a case in which the server needs to obtain all the user data in an inference process, in this application, the server obtains only a part of the user data. As the server cannot obtain, based on the part of the user data, all content included in the user data, security of the user data is ensured.
A hot code replace method includes obtaining a first bytecode obtained by compiling source code and a second bytecode obtained by performing incremental compilation on the source code after modification; transferring, to a third bytecode, a declaration of first-type modification content in modification content in the second bytecode relative to the first bytecode, where the first-type modification content includes a modification other than a modification of a method body; patching at least one instruction in the second bytecode after transfer or the third bytecode, where the at least one instruction invokes at least one piece of content in the second bytecode, and the at least one patched instruction invokes content that is in the third bytecode and that corresponds to the at least one piece of content; and loading a target second bytecode and a target third bytecode.
A method for performing video encoding comprises obtaining a first slicing solution of a first video frame; determining encoding performance prediction values of N second slices obtained when the first slicing solution is used for a second video frame; determining a second slicing solution of the second video frame based on the encoding performance prediction values of the N second slices; segmenting the second video frame into N third slices according to the second slicing solution; and separately encoding the N third slices.
H04N 19/119 - Aspects de subdivision adaptative, p. ex. subdivision d’une image en blocs de codage rectangulaires ou non
H04N 19/172 - Procédés ou dispositions pour le codage, le décodage, la compression ou la décompression de signaux vidéo numériques utilisant le codage adaptatif caractérisés par l’unité de codage, c.-à-d. la partie structurelle ou sémantique du signal vidéo étant l’objet ou le sujet du codage adaptatif l’unité étant une zone de l'image, p. ex. un objet la zone étant une image, une trame ou un champ
H04N 19/174 - Procédés ou dispositions pour le codage, le décodage, la compression ou la décompression de signaux vidéo numériques utilisant le codage adaptatif caractérisés par l’unité de codage, c.-à-d. la partie structurelle ou sémantique du signal vidéo étant l’objet ou le sujet du codage adaptatif l’unité étant une zone de l'image, p. ex. un objet la zone étant une tranche, p. ex. une ligne de blocs ou un groupe de blocs
58.
Cloud Service Providing Method, Cloud Platform, and Related Apparatus
A cloud service providing method including: providing a configuration interface; obtaining attribute information of a target cloud service from the configuration interface; creating, based on the attribute information, a target elastic memory accelerator EMA super instance in a target resource node included in a plurality of resource nodes; and setting a management interface of the target EMA super instance based on a providing manner of the target cloud service, and providing the management interface for a tenant.
H04L 67/10 - Protocoles dans lesquels une application est distribuée parmi les nœuds du réseau
H04L 41/0823 - Réglages de configuration caractérisés par les objectifs d’un changement de paramètres, p. ex. l’optimisation de la configuration pour améliorer la fiabilité
59.
Transaction Processing Method and Cloud-Native Database System
A transaction processing method includes: a first database computing node that responds to a transaction request of a user and starts to process a target transaction; the first database computing node stores state information of the target transaction into a database state node in a process of processing the target transaction; in the process in which the first database computing node processes the target transaction, the first database computing node suspends processing the target transaction; and a second database computing node obtains the state information of the target transaction from the database state node, and continues to process the target transaction based on the state information of the target transaction.
Example cloud game processing methods are provided. In one example method, a cloud game server performs rendering on a background in a game scene of the cloud game to obtain a video stream of a background image of the cloud game. The cloud game server crops an image in the video stream of the background image of the cloud game to obtain a target video stream of the cloud game. After performing compression and encoding processing on the target video stream, the cloud game server sends the processed target video stream to the terminal device. The terminal device locally generates at least one of the game character or an item of the game character. The at least one of the game character or the item of the game character is superimposed on the background image in the processed target video stream.
A63F 13/358 - Adaptation du déroulement du jeu en fonction de la charge du réseau ou du serveur, p. ex. pour diminuer la latence due aux différents débits de connexion entre clients
A63F 13/335 - Dispositions d’interconnexion entre des serveurs et des dispositifs de jeuDispositions d’interconnexion entre des dispositifs de jeuDispositions d’interconnexion entre des serveurs de jeu utilisant des connexions de réseau étendu [WAN] utilisant l’Internet
A63F 13/355 - Réalisation d’opérations pour le compte de clients ayant des capacités de traitement restreintes, p. ex. serveurs transformant une scène de jeu qui évolue en flux vidéo codé à transmettre à un téléphone portable ou à un client léger
A63F 13/92 - Dispositifs de jeu vidéo spécialement adaptés à une prise manuelle pendant le jeu
61.
DATA SYNCHRONIZATION METHOD AND APPARATUS, DEVICE, AND STORAGE MEDIUM
This application provides a data synchronization method and apparatus, and relates to the field of data storage technologies. The method includes: obtaining, from an incremental log in a source database, N incremental operation record sets arranged in a chronological order, where N is an integer greater than 1, and an Nth incremental operation record set is a current to-be-synchronized incremental operation record set; determining a target incremental operation record set based on a validation code of an incremental operation record set that is before the Nth incremental operation record set, where the target incremental operation record set is an incremental operation record set that is successfully synchronized to a last time of a target database; and synchronizing an incremental operation record set in the N incremental operation record sets to the target database, where the incremental operation record set is after the target incremental operation record set.
G06F 16/27 - Réplication, distribution ou synchronisation de données entre bases de données ou dans un système de bases de données distribuéesArchitectures de systèmes de bases de données distribuées à cet effet
An image recognition model training method may be applied to the field of cloud computing. The method includes: A first training apparatus on a user local side inputs, into an encoding module, a first image dataset stored on the user local side, to train the encoding module to obtain a trained encoding module. A second training apparatus on a cloud obtains the trained encoding module from the first training apparatus; and inputs a labeled second image dataset stored on the cloud into an image recognition model that includes the recognition module and the trained encoding module, to train the recognition module to obtain a trained recognition module. According to the method, an image recognition model can be trained using image data of a user while privacy leakage of the user is avoided.
G06V 10/82 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant les réseaux neuronaux
G06V 10/77 - Traitement des caractéristiques d’images ou de vidéos dans les espaces de caractéristiquesDispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant l’intégration et la réduction de données, p. ex. analyse en composantes principales [PCA] ou analyse en composantes indépendantes [ ICA] ou cartes auto-organisatrices [SOM]Séparation aveugle de source
63.
METHOD AND DEVICES FOR SEAMLESS CROSS-GAME-ENGINE PLAYER MIGRATION
There is provided methods and devices for migrating a player in a cloud gaming environment from a first game engine instance to a second game engine instance. The method comprises streaming, during a migration window, an audio-video stream from each of the first game engine instance and the second game engine instance, and blending, either in the cloud or at the remote client, both streams into a single output stream. The blending of both streams uses a weighted average with weights changing over time to favour the source engine instance at the beginning of the migration window, and to favour the destination engine at the end of the migration window. Techniques such as blurring may be used to decrease artifacts in the output stream.
A code check method, comprising: acquiring code to be checked; receiving DSL code check rules written by a user on the basis of a domain-specific language (DSL); when the DSL code check rules comprise a first type of DSL code check rule indicating whether said code has a characteristic described by a natural language, constructing a prompt of a language model on the basis of the first type of DSL code check rule; on the basis of the prompt, invoking the language model to infer whether said code has the characteristic described by the natural language, so as to obtain an invoking result; and presenting to the user a check result generated on the basis of the invoking result, the check result being used for indicating whether said code has the characteristic described by the natural language. When users write DSL code check rules, the method supports using natural languages to describe characteristics that need to be checked, and then invoking language models to infer whether checked code has the characteristics described by the natural languages, so as to solve the problem that code checkers cannot cover subjective and fuzzy descriptions or flexible and changeable scenarios.
An application management method, which is executed by means of an application management platform, wherein the application management platform comprises an interaction apparatus, an agent extension apparatus, and a service agent for at least one service. The method comprises: an interaction apparatus acquiring a query input by a user, acquiring an agent identifier and conditions, and collecting the context of the query; an agent expansion apparatus determining, on the basis of the agent identifier and the conditions, policy configuration information of a target agent corresponding to the agent identifier; and the target agent expanding the context of the query on the basis of the policy configuration information, assembling the context and the expanded context on the basis of the policy configuration information to obtain a prompt, and inputting the prompt into a generative model for inference, so as to obtain an answer. In the method, policy configurations are dynamically selected and adjusted by means of a pluggable configuration mechanism, such that the problem of insufficient flexibility caused by fixed configurations is avoided, and a large number of policy combinations generated under different conditions are effectively managed and optimized. Moreover, a routing policy can be rapidly adjusted on the basis of customized requirements, thereby effectively coping with the problem of combinatorial explosion.
A phase-change liquid cooling architecture for a data center, and a machine room. The architecture comprises a cabinet located in a machine room, and a condensation unit, a water supply pipe, and a water return pipe which are fixed to the machine room. The cabinet comprises a sealed area capable of accommodating a phase-change working medium, and an exhaust port and a liquid return port separately communicated with the sealed area. The condensation unit comprises a housing and a condensation pipe located in an inner cavity of the housing. A water inlet of the condensation pipe is communicated with the water supply pipe, and a water outlet of the condensation pipe is communicated with the water return pipe. The housing is provided with a liquid outlet port and an air inlet port. The liquid outlet port and the air inlet port are separately communicated with the inner cavity of the housing. The liquid outlet port of the condensation unit can mate with the liquid return port of the cabinet to form a liquid return channel. The air inlet port of the condensation unit can mate with the exhaust port of the cabinet to form an exhaust channel. On the basis of the architecture design in which the condensation unit is decoupled from the cabinet, the condensation unit can be arranged by fully utilizing the space in the machine room, thereby achieving independent evolution of a CDU and the cabinet.
The present application relates to the technical field of computers, and discloses an instance scheduling method, a system, a cluster, and a computer-readable storage medium. In embodiments of the present application, a cloud management platform runs a first-type instance having a fixed specification. During implementation of elastic scaling, the cloud management platform obtains a first running parameter for running the first-type instance of a target service. When the first running parameter satisfies a scaling-down requirement and satisfies an instance type adjustment condition, the first-type instance is changed to a second-type instance, and the target service is run by the second-type instance. In this way, in a scaling-down process, the instance type adjustment condition is added to determine and implement an instance type change, and the first-type instance having a fixed specification is changed to the second-type instance having an adjustable specification, so that the specification of the instance can be adjusted on the basis of an actual service load, thereby improving the resource utilization rate.
The present application belongs to the field of communications. Disclosed are an image rendering method and apparatus, and a storage medium. The method is applied to a cloud service platform. The method comprises: acquiring at least one frame of scene data corresponding to at least one image frame to be rendered; determining at least one image rendering mode on the basis of a rendering pipeline of the at least one frame of scene data, wherein the rendering pipeline is used for describing a rendering process for rendering the at least one frame of scene data; on the basis of the at least one image rendering mode, dividing the at least one frame of scene data into a plurality of rendering parts, wherein the dependency intensity between the plurality of rendering parts is lower than an intensity threshold value; and rendering the plurality of rendering parts by means of a plurality of rendering devices to obtain the at least one image frame. The present application can improve the quality of rendered images.
The present invention relates to the field of data storage, and disclosed are a data reading method and a related apparatus. The method comprises: a server receives from a client a first processing request for instructing to acquire shard boundary information of target data, and returns the shard boundary information of the target data to the client in response to the first processing request and by scanning a target self-balancing tree corresponding to the target data; and the server receives from the client a second processing request for instructing to acquire P shards of the target data, and returns the P shards of the target data to the client in the form of shards in response to the second processing request. The shard boundary information of the target data comprises values of identifiers of the target data stored in a root node or child nodes of a root node of the target self-balancing tree. Values of identifiers of target data stored in a root node or child nodes of a root node are used as shard boundary information, thereby accelerating determination of the shard boundary information, and improving the efficiency of performing data reading on P shards of the target data.
The present application provides an information processing method, an information processing apparatus, a computing device cluster, and a computer-readable storage medium. The method comprises: acquiring a plurality of texts to be stored, wherein said plurality of texts include a first text and a second text; using a first embedding model to process the first text so as to obtain a first result; using a second embedding model to process the second text so as to obtain a second result, wherein the number of dimensions of the first embedding model and the number of dimensions of the second embedding model are different; and storing the first result and the second result into a knowledge base. The present application can save storage resources and avoid the excessively large number of dimensions of vectors generated by embedding models in the process of constructing a knowledge base.
Provided in the present application are a fault processing method and a related device. The method is applied to a cloud management platform. The method comprises: when a fault occurs in at least one of M services and there are a plurality of pieces of alarm information, providing first alarm information for a tenant, wherein the first alarm information is generated by summarizing, on the basis of a first operation and maintenance model, alarm information triggered on the basis of the same root fault among the plurality of pieces of alarm information; receiving first input information of the tenant; and in response to the first input information, providing first output information for the tenant. The method can process at least one piece of alarm information triggered on the basis of the same root fault among the plurality of pieces of alarm information, so as to generate one piece of valuable alarm information corresponding to each root fault, thereby effectively suppressing an alarm storm; and by means of natural language interaction with the tenant, the cause of a fault can be provided for the tenant, thereby improving the efficiency of fault handling.
Provided in the present application are a virtual instance management method based on cloud computing technology, and a cloud management platform. The cloud management platform is used for managing a cloud data center, wherein the cloud data center comprises a plurality of servers that access a high-speed interconnect bus inside the cloud data center. The method comprises: receiving a virtual instance memory creation request that is input by a tenant; in response to the virtual instance memory creation request, selecting a first server from a plurality of servers and sending a first control command to the first server; receiving a first virtual instance memory binding request; and in response to the first virtual instance memory binding request, sending a second control command to the first server to instruct the first server to provide, by means of a high-speed interconnect bus, a virtual instance memory located on the first server to a first virtual instance located on a second server for use. By means of the technical solution, memory sharing across servers for virtual instances can be realized, thereby improving data communication efficiency and memory utilization efficiency.
Provided in the present application are a container scheduling method and a related device, which are used for improving the reliability of services. The method is applied to a cloud platform, wherein the cloud platform is used for managing a server cluster. The method comprises: acquiring a first request, wherein the first request comprises at least one maximum resource parameter value of at least one container and at least one service level corresponding to the at least one container, and the at least one service level comprises a high-priority service; and on the basis of the at least one maximum resource parameter value and the at least one service level, scheduling the at least one container to a target server, wherein the target server is included in a server cluster, and at most some or all allocable resources of the target server are allocated for use by all containers of high-priority services that are scheduled by the target server.
G06F 9/455 - ÉmulationInterprétationSimulation de logiciel, p. ex. virtualisation ou émulation des moteurs d’exécution d’applications ou de systèmes d’exploitation
A data processing method and related apparatus are provided in embodiments of the present disclosure. The target data for target items of a target component model is determined based on input data associated with a target component, and in a case that at least one of the target items has no corresponding target data, the target data is 5 refined based on a target database to obtain refined data for the target items. The target database includes reference data for reference items of one or more reference component models corresponding to one or more reference components, and at least one of the one or more reference components is associated with the target component. In this way, the refined data, i.e., component parameters for generating the target component model can be accurately obtained.
G06F 30/27 - Optimisation, vérification ou simulation de l’objet conçu utilisant l’apprentissage automatique, p. ex. l’intelligence artificielle, les réseaux neuronaux, les machines à support de vecteur [MSV] ou l’apprentissage d’un modèle
G06F 16/21 - Conception, administration ou maintenance des bases de données
The present disclosure provides a locking method for locking a directory tree and related products. The locking method includes: obtaining an operation request; and performing, according to the operation request, a multiple-granularity locking (MGL) operation on multiple MGL lock objects in multiple MGL lock object arrays in a set of MGL lock object arrays, where the multiple MGL lock objects in the multiple MGL lock object arrays are used for tracking locking status of multiple nodes on multiple tree levels of the directory tree, where each tree level of the multiple tree levels corresponds to an MGL lock object array of the multiple MGL lock object arrays, and each node of the multiple nodes corresponds to an MGL lock object of the multiple MGL lock objects. MGL of the directory tree can be implemented, thereby improving concurrency of filesystem tree operations.
This application provides a computing device based on a phase-change liquid cooling technology, including a body, a jet component, and a power component. Space in the body is divided into a first accommodation cavity and a second accommodation cavity through a partition plate. The first accommodation cavity is configured to accommodate a first coolant, the second accommodation cavity is configured to accommodate a heating element and a second coolant, and a temperature of the second coolant is higher than a temperature of the first coolant. The body is provided with a liquid refill opening communicating with the first accommodation cavity and an exhaust vent communicating with the second accommodation cavity. The power component is connected to the jet component, and is configured to pump the first coolant to the heating element through the jet component.
In an example method, a scheduling system receives an access request of a user, where the access request includes a service domain name. The scheduling system determines, based on the access request, a target device from a plurality of access network devices that serve the service domain name, and redirects the access request to a target device in a normal service state. A domain name system receives a domain name resolution request of the user, where the domain name resolution request of the user is generated based on redirection information, and the domain name resolution request of the user includes a self-resolved domain name of the target device. The domain name system determines a service address of the target device based on a manner of mapping between a self-resolved domain name and a service address and the self-resolved domain name of the target device.
H04L 61/4511 - Répertoires de réseauCorrespondance nom-adresse en utilisant des répertoires normalisésRépertoires de réseauCorrespondance nom-adresse en utilisant des protocoles normalisés d'accès aux répertoires en utilisant le système de noms de domaine [DNS]
78.
PHASE-CHANGE LIQUID COOLING ARCHITECTURE FOR DATA CENTER AND SERVER ROOM
A phase-change liquid cooling architecture for a data center, and a server room. The architecture comprises a server room, a cabinet, and a condensation unit fixed in the server room. The server room comprises a sealed device chamber, in which the cabinet and the condensation unit are located, the condensation unit being located above the cabinet. A node chassis on a side of the cabinet comprises an exhaust opening and a liquid inlet, and an inner cavity of the chassis is capable of accommodating a phase-change working fluid. A vaporized working fluid in the node chassis can be discharged into the device chamber through the exhaust opening, and a liquid return pipe in communication with the liquid inlet of the chassis is provided with a liquid return interface. The condensation unit comprises a condensation tube and a liquid collection tray, the liquid collection tray being located below the condensation tube. A liquid outlet interface of the liquid collection tray is connected to the liquid return interface of the liquid return pipe to form a liquid return channel. In this configuration, the condensation unit and the cabinet are arranged in the sealed device chamber of the server room to form a server room-level sealed phase-change liquid cooling architecture. Compared with conventional cabinet-level sealing or node-level sealing, the overall cost of implementing sealing can be effectively reduced.
Disclosed are a virtual-machine state switching method and apparatus, and a device cluster and a readable storage medium, which relate to the technical field of virtualization. The method comprises: during the processing of virtual-machine state switching, after having received a switching request of a first virtual machine, creating a second virtual machine in a host machine that runs the first virtual machine, and on the basis of a second state indicated by the switching request, configuring a message exchange mechanism of the second virtual machine; and live-migrating the first virtual machine to the configured second virtual machine, so as to implement online switching of a message exchange mechanism in the first virtual machine. In this way, a data packet processing method between a first virtual machine and a client is switched without shutting down the first virtual machine. Thus, losses caused by service interruptions are reduced.
G06F 9/455 - ÉmulationInterprétationSimulation de logiciel, p. ex. virtualisation ou émulation des moteurs d’exécution d’applications ou de systèmes d’exploitation
80.
RENDERING METHOD AND APPARATUS, DEVICE CLUSTER, STORAGE MEDIUM, AND PROGRAM PRODUCT
Disclosed are a rendering method and apparatus, a device cluster, a storage medium and a program product, relating to the field of image processing. The method comprises: acquiring a first image on the basis of a first SPP of an image to be rendered, rendering the image to be rendered again to obtain a second image when an image quality metric of the first image is lower than a preset threshold, and then fusing the first image and the second image to generate a third image. This allows users to set different preset thresholds according to different requirements for rendering effect and efficiency under different scenarios. Whenever an image quality metric of an image generated at a given level during hierarchical rendering is lower than the preset threshold, the rendered SPP is then adjusted to perform the next level of hierarchical rendering to generate a new image, and then generate an image with a higher SPP by means of image fusion, thereby satisfying different user requirements with respect to rendering effect and rendering efficiency, preventing wasted computing power resources and time resources due to the use of excessively high SPPs for rendering in scenarios with lower rendering effect requirements, and reducing resource waste.
The present application provides an anomaly detection method and apparatus, and a computing device cluster. The method is applied to a network protection device for screening network traffic on the basis of at least two screening rules. The at least two screening rules include a first screening rule and a second screening rule. The method comprises: receiving a target packet comprising a first element and a second element, the target packet being any one of N (a positive integer greater than or equal to 1) packets comprised in the network traffic; determining whether the target packet satisfies the first screening rule (used for screening on the basis of the frequency of occurrence of the first element); when the target packet satisfies the first screening rule, determining whether the target packet satisfies the second screening rule (used for screening on the basis of the frequency of occurrence of the second element); and when the target packet satisfies the second screening rule, performing anomaly detection on the basis of the target packet. By performing anomaly detection on the basis of traffic that has undergone multi-level screening, resource consumption is reduced, and user experience is improved.
The present application relates to the technical field of cloud services, and discloses a cloud resource scheduling method and apparatus. The method comprises: a cloud management platform receives an instance creation request sent by a tenant, wherein the instance creation request carries a first instance type and a first instance specification of a first instance to be created; the cloud management platform acquires an available resource amount of resources that each of a plurality of hosts can provide to an instance of the first instance type, wherein a first host can provide resources to instances of a plurality of instance types, calculation manners for the available resource amounts of the resources that the first host can provide to the instances of the plurality of instance types are different, and the first host is any one of the plurality of hosts; the cloud management platform selects, from among the plurality of hosts, the first host of which the available resource amount can meet the first instance specification, so as to obtain a target first host; and the cloud management platform creates a first instance on the target first host. The present application effectively improves the resource utilization rate of hosts.
Provided in the embodiments of the present application are a video processing method and a related apparatus, which are used for optimizing the matting effect of a video. The method comprises: acquiring at least one key frame among at least one first video frame; inputting the at least one first video frame into a first neural network model, so as to obtain at least one second video frame and the confidence of each second video frame; on the basis of the confidence of the at least one second video frame, screening the at least one first video frame to select at least one third video frame; in response to an operation for at least one frame among the at least one key frame or the at least one third video frame, correcting the at least one frame to obtain at least one fourth video frame; inputting the at least one fourth video frame into the first neural network model for training, so as to obtain a second neural network model; and inputting the at least one first video frame into the second neural network model, and outputting at least one fifth video frame.
Disclosed in the present application are a data processing method and a related device thereof, which can reduce the computational load of a data processing process, thereby reducing the cost of data processing. The method of the present application comprises: when a certain piece of data of a user needs to be processed, first acquiring a data processing model, which may include a first sub-model and a second sub-model; next, inputting the data into the first sub-model, so as to use the first sub-model to convert the data into a first input vector to be input into the second sub-model; then, acquiring from a preset vector set a second input vector matching the first input vector, and a first output vector corresponding to the second input vector; and subsequently, using the first output vector as an output of the second sub-model, and then using the first output vector to acquire the processing result of the data. At this point, a processing process for the data is completed.
The present application provides a conference communication method, a conference communication apparatus, a computing device cluster, a computer-readable storage medium, and a computer program product. The conference communication method is applied to a conference control server, and comprises: receiving a call request sent by a calling terminal; creating a conference on the basis of the call request, and inviting the calling terminal and a called terminal to join the conference; and after the calling terminal and the called terminal join the conference on the basis of the invitation, starting the conference to perform conference communication. The present application can simplify the conference creation process, facilitating free switching between a two-party conference and a multi-party conference.
The present application belongs to the technical field of communications. Disclosed are a method and apparatus for collaboratively executing a task by multiple apparatuses, and a device and a storage medium. The method comprises: a first apparatus sending a call request and a first task parameter to a second apparatus, wherein the call request instructs the second apparatus to execute a first sub-task, the first sub-task comprises a sub-task with power consumption greater than a first power consumption threshold value in a conference task, the second apparatus is an apparatus matching the first apparatus, the power handling capability of the second apparatus is higher than that of the first apparatus, and the first task parameter is a parameter required for executing the first sub-task; and executing a second sub-task in the conference task other than the first sub-task. In the present application, the first apparatus having a relatively low power handling capability calls an apparatus having a relatively high power handling capability to execute a sub-task with high power consumption, thereby reducing the consumption of computing resources and electric quantity of the first apparatus, prolonging the time during which the first apparatus executes a task, and preventing the temperature of the first apparatus from being too high.
G06F 1/3293 - Économie d’énergie caractérisée par l'action entreprise par transfert vers un processeur plus économe en énergie, p. ex. vers un sous-processeur
87.
PUBLIC CLOUD TECHNOLOGY-BASED NETWORK ACCESS METHOD, AND CLOUD SYSTEM
The present application relates to the technical field of cloud services, and discloses a public cloud technology-based network access method, and a cloud system. The cloud system comprises a cloud management platform and infrastructure. The infrastructure comprises access objects obtained on the basis of the infrastructure, and a data subsystem for data transmission between different access objects. The cloud management platform is used for configuring application endpoints for a plurality of access objects in the cloud system, configuring identifiers for the application endpoints, and acquiring an access request sent by a first access object for a second access object, wherein a source address of the access request is an identifier of a first application endpoint of the first access object, a destination address of the access request is an identifier of a second application endpoint of the second access object, and the first access object and the second access object are two of the plurality of access objects; and then, on the basis of the identifier of the second application endpoint, the access request is sent to the second access object by using the infrastructure. The present application achieves a minimalistic unified network model.
The present application discloses a cloud service-based logic node configuration method and a cloud management platform, capable of improving the resource utilization of network interface cards used to implement tenant logic nodes. The method of the present application comprises: a tenant sends a logic node creation request to a cloud management platform, wherein the logic node creation request comprises specification requirements set by the tenant for a logic node; accordingly, the cloud management platform selects, from among a plurality of physical device pools, idle physical devices that satisfy the specification requirements set by the tenant for the logic node, and creates the logic node on the physical devices; and then, the cloud management platform installs an operating system image on the logic node, so that the logic node can be remotely accessed by the tenant, wherein the logic node on which the operating system image is installed generates, on the basis of a remote access operation of the tenant, a packet to be forwarded, writes said packet into a queue created in real time by a network interface card that is used to implement the logic node, and notifies the network interface card to obtain said packet from the queue, so as to forward said packet.
Disclosed in the present application are a cloud management platform-based network anomaly determination method and a cloud management platform, which can improve tenant experience. The method of the present application comprises: a cloud management platform creates a plurality of computing nodes for a tenant to execute a job task of the tenant, wherein a plurality of communication paths are provided between the plurality of computing nodes, and each communication path can transmit traffic generated by two computing nodes connected thereto during execution of the job task; on this basis, the cloud management platform can perform traffic prediction on the plurality of communication paths by using information of the job task, so as to obtain predicted traffic transmitted over the plurality of communication paths; the cloud management platform can also collect traffic over the plurality of communication paths, so as to obtain actual traffic transmitted over the plurality of communication paths; and if the predicted traffic of the first communication path among the plurality of communication paths does not match the actual traffic of the first communication path, the cloud management platform can alert the tenant to an anomaly occurring on two computing nodes connected to the first communication path.
Disclosed in the present application are a server system, and a job task processing method based on a cloud service, which can provide a tenant with a higher-quality cloud service. The server system in the present application may include a plurality of physical server groups, wherein each physical server in these physical server groups includes a CPU, a memory, a network interface card, a disk and a computing chip. For any physical server, the CPU, memory, network interface card, disk and computing chip that are included in the physical server are interconnected inside the physical server, and network interface cards of all of the plurality of physical servers are all connected to a first network, wherein the first network enables the network interface cards of all of the plurality of physical servers to be interconnected within a group and between groups, and computing chips of all of the plurality of physical servers can all be connected to a second network by means of direct-output channels provided by the computing chips, wherein the second network enables the computing chips of all of the plurality of physical servers to be interconnected within a group and between groups.
H04L 49/109 - Éléments de commutation de paquets caractérisés par la construction de la matrice de commutation intégrés sur micropuce, p. ex. interrupteurs sur puce
H04L 41/12 - Découverte ou gestion des topologies de réseau
G06F 15/173 - Communication entre processeurs utilisant un réseau d'interconnexion, p. ex. matriciel, de réarrangement, pyramidal, en étoile ou ramifié
91.
THREE-DIMENSIONAL MODELING METHOD AND APPARATUS BASED ON PUBLIC CLOUD TECHNOLOGY
The present application belongs to the field of public cloud technology. Disclosed are a three-dimensional modeling method and apparatus based on public cloud technology. The method is applied to a cloud platform. The method comprises: acquiring three-dimensional data of a first model input by a user, wherein the three-dimensional data is an explicit expression of the first model; on the basis of the three-dimensional data, performing model segmentation on the first model, so as to obtain a second model comprising a plurality of model portions, wherein each of one or more model portions of the second model has one or more sharp features, and the sharpness of each sharp feature is greater than a specified threshold value; respectively performing implicit modeling on the plurality of model portions, so as to obtain implicit expressions of the plurality of model portions; and on the basis of the implicit expressions of the plurality of model portions, performing merging to obtain a third model, and performing model rendering on the third model, so as to obtain a rendering result of the third model and display same to a user. The present application improves the modeling quality of a three-dimensional modeling result.
The present application discloses a cloud management platform-based gateway instance configuration method, and a cloud management platform, capable of improving tenant experience. The method of the present application comprises: when a tenant needs to configure a gateway instance group thereof, the tenant may send an anti-affinity deployment request for the gateway instance group to a deployment interface provided by a cloud management platform. Therefore, the cloud management platform may create, on the basis of the anti-affinity deployment request, a first gateway instance and a second gateway instance comprised in the gateway instance group, wherein the first gateway instance and the second gateway instance enable the tenant to access an application of the tenant. The first gateway instance may be constructed from a plurality of first cloud instances deployed in a first fault domain, and the second gateway instance may be constructed from a plurality of second cloud instances deployed in a second fault domain. Because the first fault domain and the second fault domain are different fault domains, an anti-affinity deployment is formed between the first gateway instance and the second gateway instance.
H04L 67/10 - Protocoles dans lesquels une application est distribuée parmi les nœuds du réseau
H04L 41/08 - Gestion de la configuration des réseaux ou des éléments de réseau
G06F 9/455 - ÉmulationInterprétationSimulation de logiciel, p. ex. virtualisation ou émulation des moteurs d’exécution d’applications ou de systèmes d’exploitation
93.
GARMENT TRY-ON METHOD AND APPARATUS, AND COMPUTING DEVICE CLUSTER
Embodiments of the present application provide a garment try-on method and apparatus, and a computing device cluster. The garment try-on method comprises: acquiring at least one image of a real garment; performing matching retrieval in a plurality of virtual garment models on the basis of the at least one image of the real garment to acquire a retrieval result, and obtaining a garment try-on model on the basis of the retrieval result; acquiring a virtual human body model; combining the virtual human body model and the garment try-on model to obtain a virtual wearing model, the virtual wearing model being a model of a virtual human body wearing a virtual garment; and performing image rendering on the basis of the virtual wearing model, to generate a virtual human image wearing the virtual garment. In the garment try-on method according to the embodiments of the present application, a real garment can be identified to achieve try-on of a virtual garment corresponding to the real garment, thereby simplifying a virtual garment try-on process, and improving a user's experience of trying on the virtual garment.
Provided in the present application is a data processing method. The method is applied to a storage system, and the storage system comprises a data processing unit (DPU), a computing node, and a storage node, the storage node and the computing node being arranged in at least one data center that provides cloud services. The method comprises: the DPU acquires from the computing node a data block to be written, and stores said data block in a memory of the DPU; the DPU processes said data block to generate a target sorted string table (SSTable), wherein the target SSTable comprises a data block obtained after the DPU compresses and verifies the data block to be written; and the DPU sends the target SSTable to the storage node. The method can improve the resource utilization rate of servers.
The present application provides a document editing method. The method comprises: a first terminal device acquiring a stored target document from a server; the first terminal device generating a first editing operation on the target document, and caching the first editing operation locally; upon determining that another terminal device has edit the target document stored on the server, the server sending a first notification message to the first terminal device; the first terminal device receiving the first notification message, and, on the basis of the first notification message, synchronizing the locally cached first editing operation to the server; and the server receiving the first editing operation, and, on the basis of the first editing operation, updating the target document stored on the server. The solution can reduce power consumption of terminal devices.
Provided in the present application are a virtual object generation method, an apparatus and a system. The method comprises: acquiring a user requirement; on the basis of the user requirement, determining a generation requirement, the generation requirement being used for generating a virtual object; decomposing the generation requirement into at least one subtask, each subtask among the at least one subtask corresponding to one requirement domain, and each requirement domain comprising at least one requirement function; determining a requirement function corresponding to the virtual object in at least one requirement domain; and invoking a function model of the requirement function corresponding to the virtual object in the at least one requirement domain to generate the virtual object. In the present solution, decomposing the generation requirement into multiple layers on the basis of the requirement domain and the requirement function can refine the generation requirement and ensure accurate invoking of the model, thereby generating more accurate virtual objects of higher quality.
This disclosure provides a microservice management method implemented based on a cloud management platform. The method includes obtaining, from a microservice configuration interface, microservice configuration information input by a tenant. The microservice configuration information indicates a region group and a service level agreement SLA constraint for the region group. The region group includes multiple first microservices. The SLA constraint includes a first SLA parameter that needs to be met by first microservices. The method further includes selecting, from multiple sites based on the microservice configuration information, a first site that meets the first SLA parameter. The method includes creating multiple first virtual instances at the first site. The method further includes deploying the first microservices in the first virtual instances.
H04L 41/5019 - Pratiques de respect de l’accord du niveau de service
G06F 9/455 - ÉmulationInterprétationSimulation de logiciel, p. ex. virtualisation ou émulation des moteurs d’exécution d’applications ou de systèmes d’exploitation
H04L 41/12 - Découverte ou gestion des topologies de réseau
A network device, comprising a front panel, a rear panel, and a fan assembly. The front panel is provided with a plurality of first network components; the rear panel is provided with a plurality of second network components; the fan assembly is arranged on the rear panel; the fan assembly comprises an air duct switching assembly, wherein the air duct switching assembly is configured with a first operating mode and a second operating mode; in the first operating mode, air flows along outer wall surfaces of the first network components and the second network components; and in the second operating mode, air flows along the outer wall surfaces of the first network components. In the present application, the egress interface capacity of a switch is improved without losing online maintenance features of the fan assembly. The fan assembly is adapted to two scenarios: the first network components require cooling, whereas the second network components do not; and the first network components and the second network components all require cooling. During cooling of the first network components, the air resistance is lower, thereby achieving lower fan power consumption.
A time management method, wherein the method comprises: acquiring a communication carrier for a first communication request; and on the basis of a first timestamp, a second timestamp, a first remaining time, and a first execution time, determining whether a first microservice has timed out after completing a computing task corresponding to the first communication request. In the present application, upon receiving a first communication request, tools such as a java agent can be invoked to non-invasively obtain, from the first communication request, communication request sending time, the remaining time, and microservice execution time recorded in the communication carrier, and then detect whether the remaining time carried by the communication carrier is less than the communication time for receiving the first communication request and the time taken by the first microservice to execute a computing task corresponding to the first communication request, so that the first microservice can independently detect whether the execution of the computing task corresponding to the first communication request has timed out.
An image segmentation method, comprising: acquiring a video stream, and acquiring a target image frame from the video stream; obtaining at least one mask on the basis of image information of the target image frame and optical flow information of multiple image frames; and on the basis of the at least one mask, segmenting, from the target image frame, pixel points corresponding to pixel values in the at least one mask. In the present application, according to the method, when image segmentation is performed an image frame in a video, a person and an object that have the same or similar optical flow information can be classified into the same category on the basis of the image and optical flow information of multiple image frames previous to and subsequent to the image. According to the method, when a person is segmented from an image, an object of the same category to which the person belongs is also segmented, so that the person and the object associated with the person are segmented.
