56 results about "Task segmentation" patented technology

The invention discloses a task processing method and apparatus, a task segmentation method and apparatus, a task processing system, and an electronic device, which are used for a distributed task scheduling system. After the distributed task scheduling system starts a to-be-executed task, the task processing method comprises the steps of allocating the to-be-executed task to a first task execution server in a task execution server cluster, and instructing the first task execution server to perform task segmentation processing on the to-be-executed task; and receiving at least one to-be-executed sub-task generated after segmentation of the to-be-executed task and returned by the first task execution server, and taking the at least one to-be-executed sub-task as a task needed to be executed by the task execution server cluster. By adopting the method provided by the invention, the problem that tasks processed by a single task execution server include excessive to-be-processed objects can be solved, so that the effects of improving the task execution efficiency and increasing the success rate are achieved.

The invention relates to a path planning method, and specifically relates to a time window-based task segmentation multi-AGV path planning algorithm. The method comprises steps of first, generating anAGV working environment model; second, generating a task, and decomposing the task into multiple secondary stages according to different task types; and third, assigning the task by a task assigningmodule according to task priorities, picking the task out by a scheduling system from a high-priority queue, then determining which stage the task is in, if the task is not executed, assigning the task to closest AGV by using an improved A* algorithm, and setting the task state in first secondary stage; if the task is in the second or after stages, and planning a non-conflict shortest path for theAGV bound with the task by using the time widow algorithm by the scheduling system. Shortest and fastest paths are planned by using the improved A* algorithm and time window algorithm, so that the working efficiency is improved and calculation errors are reduced.

The invention discloses a mobile collaborative calculation method comprising the following steps: using the features that mass wireless devices exist in a wireless network and each device has certain computing resources and communication resources, allowing a local device to search a surrounding auxiliary device with the same idle period according to preset rules, and using the auxiliary device computing resource to assist the local device for collaborative calculation; allowing the auxiliary device to use the communication resource to transfer the local device to a mobile edge communication system, and realizing collaborative communication; optimizing a transmission time, computing task segmentation and transmission power, thus realizing minimum energy loss. The calculation and communication cooperation with the auxiliary device can solve the mobile edge calculation deep fading and computing/communication resource limitation problems, thus improving the mobile edge calculation performances; the invention also discloses a mobile collaborative calculation device realizing said technical effects.

The present invention relates to a load balance dynamic preemptive task scheduling method and system for a relay satellite. The method comprises a first step of if an antenna has long enough idle time, inserting a task directly in the time to perform direct scheduling; if the task cannot be inserted directly, performing preemptive task switching and sub-task segmentation and insertion so as to performing scheduling, and if scheduling still fails, executing a next step; a second step of scheduling, through indirect scheduling, a new task that is not executed after a direct scheduling module is used; and a third step of establishing three target functions to select an optimal dynamic scheduling scheme. In the method, a direct scheduling module and an indirect scheduling module are used to further optimize the scheduling scheme of the relay satellite task in a dynamic disturbance condition, so that the system performance is better than that using the complete rescheduling and traditional dynamic scheduling method. The method can be widely applied in the field of the space information network relay satellite resource scheduling.

The invention provides a processor and a processing method. The processor comprises a task segmentation device used for carrying out task segmentation according to task segmentation granularity, and ahardware resource division device for dividing the hardware resources of the processor according to the task segmentation result. According to the processor and the processing method, the task is segmented, and different hardware configurations are performed according to the task segmentation, so that the processing performance is improved, and the overhead is reduced.

The invention relates to a navy detection model construction method. The navy detection model construction method comprises the steps of (1) conducting task segmentation on a set of sample data to obtain a plurality tasks, and extracting average features to obtain a training sample set of the tasks; (2) selecting the features of the tasks to obtain a feature weight matrix of the tasks; (3) setting a threshold value delta, judging whether the maximum value of one column vector in the feature weight matrix is larger than the threshold value delta or not, and if yes, executing the step (4); if not, abandoning the column vector, and executing the step (5); (4) adding the column vector into a sharing feature item set; (5) judging whether columns vectors which are not compared with the threshold value delta exit in the feature weight matrix or not, and if yes, executing the step (3); if not, executing the step (6); (6) inputting a new training data set; (7) obtaining a linear classification value through calculation; (8) setting a navy threshold value, and determining that data are from a navy when the linear classification value is larger than the navy threshold value. According to the navy detection model construction method, a navy detection model is built through a multi-task learning method, so that a navy user is conveniently and rapidly recognized.

The invention relates to a multitask unloading method based on edge server cooperation, and belongs to the field of wireless communication. The method comprises the following steps: S1, modeling an edge server variable; S2, modeling user task characteristics; S3, modeling a user task segmentation variable, and unloading the variable and a time slot distribution variable; S4, modeling user task local execution delay; S5, modeling the task execution delay of the edge server; s6, modeling user task scheduling limiting conditions; and S7, determining a user task unloading strategy based on the minimization of the maximum task processing delay of the user. The optimal user task scheduling strategy and the optimal unloading ratio under the condition that the task is effectively executed can be ensured, and the user delay minimization is realized.

The invention is suitable for the technical field of video monitoring, and provides a fast characteristic extraction system based on mass videos. The fast characteristic extraction system based on the mass videos comprises an analysis depth application module A of the mass videos, an analysis application service module B of the mass videos, a source module C of the mass videos, a video big-data processing module D and a basic resource module E, wherein the video big-data processing module D comprises an algorithm control module F, a task segmentation module G, an algorithm injection module H, a task scheduling module I and a data storage module K; after the video big-data processing module D receives an operation task, the task segmentation module G is called according to video source information to segment the operation and package the operation into small tasks; the small tasks are issued to each cluster analysis node through the task scheduling module I; and the analysis node analyzes a video segment through the algorithm control module F and the algorithm injection module H. A plurality of severs with relatively low performance are arranged to carry out distributed processing so as to greatly lower cost and bandwidth requirements.

A task splitting device and method, a task processing device and method, and a multi-core processor are provided. That task splitting device comprises a granular task splitting unit for splitting a task to form a sub-task by at least one granularity; and a task segmentation granularity selecting unit for selecting the adopted granularity.

The invention relates to an intelligent manufacturing-oriented collaborative development system and method. The system comprises a demand analysis module used for obtaining demands of a client on a product before production of the product, performing interpretation and classification on the demands according to knowledge and converting the demands into tasks, a task decomposition module for generating a general task according to the demands of the client on the product, performing task subdivision and function subdivision on the general task and finally decomposing the general task to form identifiable sub-tasks, a task assignment module used for performing assignment on described tasks and problems, selecting decision elements of a task execution process and an optimization algorithm of aprocess model and assigning the tasks to related collaborative design modules, and a collaborative development module used for receiving and executing the assigned tasks and performing assessment andanalysis on executive results. Compared with the prior art, the collaborative development system and method has the advantages of organization collaboration, intuitiveness and high efficiency, dynamic change, production flexibility and the like.

The invention provides a cooperative computing unloading method based on a splittable task in a vehicle-mounted edge computing environment, and the method comprises the steps: sending a to-be-processed task unloading request when the calculation amount of a to-be-processed task cannot be borne; acquiring relevant information of callable computing resources according to feedback of the to-be-processed task unloading request; and obtaining an optimal splitting ratio for task segmentation according to the related information of the callable computing resources. Therefore, the optimal unloading decision is selected by comprehensively considering the computing power of the vehicle, and the minimum task completion time delay can be conveniently obtained. After the request vehicle receives the relevant information of the callable computing resources, a task splitting model is constructed to perform construction, the optimal splitting ratio is obtained, a distributed task splitting and unloading scheme is practiced, it is guaranteed that each request vehicle autonomously decides splitting and unloading decisions of a task only based on received information, and frequent information interaction in a centralized scheme is avoided.

The invention discloses a search engine evaluation system and management thereof. The search engine evaluation system comprises a test management module, a fetching module, a parsing module, a Mola database, a task segmentation module, a MySql database, an evaluation module, a task distribution module, a task pool, an answer monitoring module and a result export module. The invention further discloses search engine evaluation system management. According to the invention, the authenticity of network transaction evaluation, test processes and results and learning condition inspection, occurrence of a cheating phenomenon is prevented, the number of the difficulty of tasks are screened and controlled timely, the progress of transaction evaluators, examinees and learners is controlled, occurrence of phenomena that the transaction evaluators evaluate repeatedly and that the examinees and the learners fetch more examination questions and grab examination questions maliciously is prevented, and the quality of evaluation, examination and learning is ensured.

The invention discloses a cooperative unloading method based on a Markov decision process in a mobile cloud computing system, which comprises the following steps: (1) combining a task queue state, anedge cloud state, a mobile self-organizing cloud state and a center cloud state into a state space of the Markov decision process, and calculating a state transition probability matrix; (2) defining action space; (3) defining the immediate return function of Markov decision process by time delay and energy consumption; (4) taking a series of input tasks as statistical samples to calculate task segmentation threshold; (5) adaptive task segmentation being realized by task segmentation algorithm according to the obtained threshold; (6) according to the size of the subtask, the state transition probability matrix and the immediate return function, the unloading decision result being obtained by the value iteration algorithm. The invention is based on a Markov decision process and meets the requirements of minimizing time delay and energy consumption; the adaptive task segmentation algorithm is used to achieve the full utilization of cloud computing resources and load balancing.

The invention provides a data cleaning method and system. The method comprises the steps of conducting task segmentation on data which needs to be manually cleaned; publishing segmented tasks to a crowdsourcing platform; receiving manual cleaning result data returned by task receivers through the crowdsourcing platform, and integrating the manual cleaning result data and machine cleaning result data. According to the data cleaning method and system, data which needs to be manually cleaned is subjected to task segmentation, the segmented tasks are sent to a large number of unfixed task receivers through the crowdsourcing platform, and thus cleaning of the data is completed, and the data cleaning efficiency can be improved, and the cost can be lowered.

The invention provides an aviation product production process execution management system. The system comprises a data obtaining unit 1, a task sending unit 2, a task segmentation unit 3, a task importing unit 4, a permission management unit 5, a task assigning unit 6, a material input unit 7 and a quality inspection management unit 8. The system can automatically collect production data; the datasuch as various test data, material information, bad information and the like of avionics instruments is subjected to systematic collection, so that a large amount of labor costs can be reduced; real-time tracking can be performed in a production process, and the process from work order creation to production workshop execution, the work order completion situation and the like are subjected to accurate statistics, so that production schedule and position can be mastered in real time. For a manufacturing department, a work order schedule can be mastered in real time, and a production completion event and shipping time are mastered, for anytime frame skipping internationalization and delivery timeliness rate.

The invention provides a task execution method and device and a task execution system, and belongs to the technical field of computer application. The embodiment of the invention provides a task execution method and device and a task execution system. The method comprises the following steps: receiving a starting notification sent by a computer in a cluster computer; sending a competition instruction to each computer in the cluster computer according to the starting notification, each computer is enabled to execute the competition instruction; obtaining a competition result returned by each computer; according to the competition results, sending a task segmentation instruction and a standby task instruction to the computer, so that each computer in the cluster computer completes the competition, the main computer extracts data related to the task segmentation instruction from the database and segments the data into a plurality of task segments, and then the main computer and the slavecomputers execute the plurality of task segments together, so that data analysis can be completed quickly, and the real-time performance of data analysis is guaranteed.

The embodiment of the invention relates to the technical field of distributed computing, and discloses a distributed computing system and a distributed computing method. The distributed computing system comprises a task segmentation module and a resource scheduling module. The task segmentation module is used for segmenting the received computing task into a plurality of computing stages and dividing each computing stage into a plurality of computing partitions. The resource scheduling module is used for allocating resources to the computing partitions in the current computing stage to be started. The calculation stage is used for starting the calculation partition and is used for finishing calculation in any one of the currently started calculation partitions. When the next calculation stage can run at the same time as the currently started calculation stage, resources are distributed to at least one calculation partition of the next calculation stage so as to start the at least one calculation partition. Compared with the prior art, the embodiment of the invention has the advantages that the utilization rate of idle resources is improved, the execution time of the whole calculation task is shortened, and the execution efficiency is improved.

The invention relates to a reliable-time-delay-guarantee-oriented end-to-edge cooperation method, which belongs to the technical field of communication. The method comprises the following steps: S1, establishing a system model, and calculating time delay experienced by a task; S2, calculating task unloading average time delay; S3, calculating task local calculation time delay: firstly, modeling local calculation into an M/D/1 queuing model, and performing calculating to obtain local task average time delay according to the deduced default probability model; S4, searching the optimal sending rate configuration under the configuration of different task segmentation ratios, and further selecting an optimal sending rate and an optimal task segmentation ratio; and S5, selecting an optimal sending rate and task segmentation, and starting task sending and calculation. According to the method, an embedded Markov theory is adopted to model a calculation unloading average time delay model, localcalculation average time delay is derived based on a network calculation theory, an unloading proportion and an unloading rate of a task flow are optimized based on a time delay demand of a task, anda time delay default probability of the task is minimized.

The invention relates to a distributed data synchronization method and system. The method comprises the following steps that: a scheduler inquires whether a synchronization task needs to be carried out from a source database, and elastically adjusts the query interval time of the scheduler according to a query result; the scheduler selects one synchronization component in a synchronization component cluster to execute the task; the synchronization component performs task segmentation and adopts a plurality of concurrent threads for processing; the synchronization component reads a synchronization data set from the source database and carries out data processing; and the synchronization component generates temporary unique identifiers to replace original unique identifiers of the synchronous data set, then writes the synchronous data set into the target database in batches, deletes updated old data records in the target database, and finally restores the original unique identifiers of the synchronous data set in batches. According to the distributed data synchronization method and system, the whole data synchronization performance is greatly improved.

The invention discloses a choroid sublayer and choroid vessel segmentation network model and a training method thereof, and the model comprises a sharing encoder module which is used for carrying out the down-sampling and feature extraction of a to-be-segmented image, and obtaining a first sharing feature; the shared decoder module is used for acquiring a first shared feature and performing up-sampling to obtain a second shared feature; the first decoder module is used for extracting choroid sublayer segmentation specific features in the second shared features; the second decoder module is used for extracting choroidal blood vessel segmentation specific features in the second shared features; and the classification module is used for calculating a segmentation result of the to-be-segmented image according to the second shared feature, the choroid sublayer segmentation specificity feature and the choroid blood vessel segmentation specificity feature. According to the invention, segmentation of choroid sublayers and choroid blood vessels is realized by arranging the five modules, namely, multi-task segmentation is realized through a multi-stream network structure.

The invention discloses a remote sensing image segmentation method based on multi-task semi-convolution. The method comprises the following steps: step 1, carrying out the preprocessing of an originalremote sensing image IO, and obtaining a remote sensing image I1 after the interference factors in the image are removed; step 2, constructing a multi-task segmentation network, performing boundary prediction and segmentation prediction tasks on the remote sensing image and adjusting the structure of the multi-task segmentation network to adapt to a specific application scene; and step 3, addingthe semi-convolution into the multi-task segmentation network so as to further improve the effect of the multi-task segmentation network. According to the method, the purpose of boundary refinement isachieved through targeted extraction of boundary information by multi-task reuse features and semi-convolution. Benefited from the boundary refinement of the method, the remote sensing image segmentation method provided by the invention remarkably improves the overall segmentation accuracy, improves the segmentation accuracy by 0.9% in comparison with the existing optimal method in a public dataset test and reduces 7.9% of detail errors in the optimal method.

The invention discloses a method for generating a starting-price calculation task of a tourist route product. The method comprises the following steps of: S1, obtaining information of a tourist route product according with a condition; S2, obtaining a corresponding route according to the information of the tourist route product, wherein the route is used for identifying a resource, the price of which is to be calculated; S3, aggregating products having the same route, so that route and product corresponding relationship data is generated; and S4, performing task segmentation of the route according to the resource, the price of which is to be calculated, so that a task to be processed is generated. According to the method for generating the starting-price calculation task of the tourist route product provided by the invention, the products having the same route are aggregated; then, task segmentation is carried out according to the route; and thus, the prices of all products corresponding to the route are obtained through price calculation of one route. By means of the method, access pressure to a resource provider is reduced; unnecessary resource waste is avoided; therefore, the access cost is reduced; and horizontal expansion can be realized.

The invention relates to the technical field of industrial defect detection, and discloses an industrial defect detection method based on multi-task learning. A defect classification task is subdivided into two subtasks, a normal/abnormal (ok/ng) dichotomy problem (recorded as task1) and a multi-label classification problem (recorded as task2) of n defect categories are solved by constructing a classification model based on a convolutional neural network (CNN). The classification model is composed of a base model and a head. The base model is responsible for extracting image features of an input image to obtain a corresponding feature image, and the base models of different tasks share network weights by adopting a hardsharing connection mode. The head is an output layer, and two branches are led out from the base model and are respectively used for solving the problems of task1 and task2; the two branches are respectively composed of a full connection layer and a sigmod function, and the probability of the ng category and the category probability of the n defects are output. The method can alleviate the problem that the current industrial defect detection method is easily interfered by imaging conditions, small differences between defects and backgrounds, low image contrast, large scale and appearance changes of defects of the same type and the like, resulting in unstable detection effect.