WO2023231350A1 - 利用整数规划求解器实现的任务处理方法、设备和介质 - Google Patents

利用整数规划求解器实现的任务处理方法、设备和介质 Download PDF

Info

Publication number
WO2023231350A1
WO2023231350A1 PCT/CN2022/136280 CN2022136280W WO2023231350A1 WO 2023231350 A1 WO2023231350 A1 WO 2023231350A1 CN 2022136280 W CN2022136280 W CN 2022136280W WO 2023231350 A1 WO2023231350 A1 WO 2023231350A1
Authority
WO
WIPO (PCT)
Prior art keywords
variable
distribution information
task
variable selection
current problem
Prior art date
Application number
PCT/CN2022/136280
Other languages
English (en)
French (fr)
Inventor
丁建辉
陈珍
Original Assignee
北京百度网讯科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 北京百度网讯科技有限公司 filed Critical 北京百度网讯科技有限公司
Publication of WO2023231350A1 publication Critical patent/WO2023231350A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/48—Program initiating; Program switching, e.g. by interrupt
    • G06F9/4806—Task transfer initiation or dispatching
    • G06F9/4843—Task transfer initiation or dispatching by program, e.g. task dispatcher, supervisor, operating system
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Definitions

  • the present disclosure relates to the field of computer technology, in particular to artificial intelligence, and specifically to a task processing method, device, electronic device, computer-readable storage medium and computer program product implemented using an integer programming solver.
  • Artificial intelligence is the study of using computers to simulate certain human thinking processes and intelligent behaviors (such as learning, reasoning, thinking, planning, etc.). It has both hardware-level technology and software-level technology. Artificial intelligence hardware technology generally includes technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing, etc.; artificial intelligence software technology mainly includes computer vision technology, speech recognition technology, natural language processing technology, and machine learning/depth Learning, big data processing technology, knowledge graph technology and other major directions.
  • Mixed integer programming is a very important problem in the field of operations research and optimization. By solving it, planning problems in various fields can be solved. How to improve the solution process of mixed integer programming problems has become the focus of the industry.
  • the present disclosure provides a task processing method, device, electronic device, computer-readable storage medium and computer program product implemented using an integer programming solver.
  • a task processing method implemented using an integer programming solver includes a branch module, and the task includes multiple levels of problems.
  • the method includes using the branch module to determine the problems of each level in turn.
  • At least one target variable of the problem, wherein using the branch module to determine at least one target variable of the current problem includes: obtaining distribution information of at least one variable that is not currently fixed among multiple variables related to the task, and the distribution information includes the type of at least one variable Distribution information and type distribution information of at least one variable in the objective function of the task; determining a first objective variable selection strategy for the current problem from a plurality of candidate variable selection strategies based on at least the distribution information of the at least one variable; and utilizing the first objective
  • the variable selection strategy determines at least one target variable for the current problem from multiple variables.
  • a task processing device implemented using an integer programming solver includes a branch module, and the task includes multiple levels of problems.
  • the device includes a determination unit configured to utilize The branch module determines at least one target variable of each level of the problem in turn, wherein the determination unit includes: an acquisition subunit configured to acquire distribution information of at least one variable that is currently unfixed among multiple variables related to the task, distribution The information includes type distribution information of at least one variable and type distribution information of at least one variable in the objective function of the task; the first determination subunit is configured to select a strategy from a plurality of candidate variables based at least on the distribution information of the at least one variable. determining a first target variable selection strategy of the current problem; and a second determination subunit configured to determine at least one target variable of the current problem from a plurality of variables using the first target variable selection strategy.
  • an electronic device including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by at least one processor, and the instructions are at least One processor executes, so that at least one processor can execute the method of any of the above aspects.
  • a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause a computer to perform the method of any of the above aspects.
  • a computer program product including a computer program, wherein the computer program implements the method of any of the above aspects when executed by a processor.
  • the solving efficiency of various types of mixed integer programming problems can be improved, the solving effect can be improved, and long-term benefits can be considered.
  • FIG. 1 illustrates a schematic diagram of an exemplary system in which various methods described herein may be implemented in accordance with embodiments of the present disclosure
  • FIG. 2 shows a flow chart of a task processing method implemented using an integer programming solver according to an embodiment of the present disclosure
  • Figure 3 shows a flow chart of a task processing method implemented using an integer programming solver according to an embodiment of the present disclosure
  • Figure 4 shows a flowchart of a task processing method implemented using an integer programming solver according to an embodiment of the present disclosure.
  • Figure 5 shows a structural block diagram of a task processing device implemented using an integer programming solver according to an embodiment of the present disclosure
  • FIG. 6 shows a structural block diagram of an exemplary electronic device that can be used to implement embodiments of the present disclosure.
  • first”, “second”, etc. to describe various elements is not intended to limit the positional relationship, timing relationship, or importance relationship of these elements. Such terms are only used for Distinguish one feature from another.
  • first element and the second element may refer to the same instance of the element, and in some cases, based on contextual description, they may refer to different instances.
  • FIG. 1 shows a schematic diagram of an exemplary system 100 in which various methods and apparatus described herein may be implemented in accordance with embodiments of the present disclosure.
  • the system 100 includes one or more client devices 101 , 102 , 103 , 104 , 105 , and 106 , a server 120 , and one or more communication networks coupling the one or more client devices to the server 120 110.
  • Client devices 101, 102, 103, 104, 105, and 106 may be configured to execute one or more applications.
  • server 120 may run one or more services or software applications that enable execution of task processing methods implemented using integer programming solvers.
  • server 120 may also provide other services or software applications, which may include non-virtual environments and virtual environments.
  • these services may be provided as web-based services or cloud services, such as under a Software as a Service (SaaS) model to users of client devices 101, 102, 103, 104, 105, and/or 106 .
  • SaaS Software as a Service
  • server 120 may include one or more components that implement the functions performed by server 120 . These components may include software components, hardware components, or combinations thereof that are executable by one or more processors. Users operating client devices 101, 102, 103, 104, 105, and/or 106 may, in turn, utilize one or more client applications to interact with server 120 to utilize services provided by these components. It should be understood that a variety of different system configurations are possible, which may differ from system 100 . Accordingly, Figure 1 is one example of a system for implementing the various methods described herein and is not intended to be limiting.
  • client devices 101, 102, 103, 104, 105, and/or 106 Users may use client devices 101, 102, 103, 104, 105, and/or 106 to send task processing requests.
  • the client device may provide an interface that enables a user of the client device to interact with the client device.
  • the client device can also output information to the user via the interface.
  • Figure 1 depicts only six client devices, those skilled in the art will understand that the present disclosure can support any number of client devices.
  • Client devices 101, 102, 103, 104, 105, and/or 106 may include various types of computer devices, such as portable handheld devices, general purpose computers (such as personal computers and laptop computers), workstation computers, wearable devices, Smart screen equipment, self-service terminal equipment, service robots, game systems, thin clients, various messaging equipment, sensors or other sensing equipment, etc.
  • These computer devices can run various types and versions of software applications and operating systems, such as MICROSOFT Windows, APPLE iOS, UNIX-like operating systems, Linux or Linux-like operating systems (such as GOOGLE Chrome OS); or include various mobile operating systems , such as MICROSOFT Windows Mobile OS, iOS, Windows Phone, Android.
  • Portable handheld devices may include cellular phones, smart phones, tablet computers, personal digital assistants (PDAs), and the like.
  • Wearable devices may include head-mounted displays (such as smart glasses) and other devices.
  • Gaming systems may include various handheld gaming devices, Internet-enabled gaming devices, and the like.
  • the client device is capable of executing a variety of different applications, such as various Internet-related applications, communication applications (such as email applications), Short Message Service (SMS) applications, and can use various communication protocols.
  • Network 110 may be any type of network known to those skilled in the art that may support data communications using any of a variety of available protocols (including, but not limited to, TCP/IP, SNA, IPX, etc.).
  • one or more networks 110 may be a local area network (LAN), an Ethernet-based network, a token ring, a wide area network (WAN), the Internet, a virtual network, a virtual private network (VPN), an intranet, an extranet, Blockchain networks, Public Switched Telephone Network (PSTN), infrared networks, wireless networks (e.g. Bluetooth, WIFI) and/or any combination of these and/or other networks.
  • LAN local area network
  • Ethernet-based network a token ring
  • WAN wide area network
  • VPN virtual private network
  • PSTN Public Switched Telephone Network
  • WIFI wireless networks
  • Server 120 may include one or more general purpose computers, special purpose server computers (eg, PC (Personal Computer) servers, UNIX servers, midrange servers), blade servers, mainframe computers, server clusters, or any other suitable arrangement and/or combination .
  • Server 120 may include one or more virtual machines running a virtual operating system, or other computing architecture involving virtualization (eg, one or more flexible pools of logical storage devices that may be virtualized to maintain the server's virtual storage devices).
  • server 120 may run one or more services or software applications that provide the functionality described below.
  • Computing units in server 120 may run one or more operating systems, including any of the operating systems described above, as well as any commercially available server operating system.
  • Server 120 may also run any of a variety of additional server applications and/or middle-tier applications, including HTTP servers, FTP servers, CGI servers, JAVA servers, database servers, and the like.
  • server 120 may include one or more applications to analyze and incorporate data feeds and/or event updates received from users of client devices 101, 102, 103, 104, 105, and/or 106.
  • Server 120 may also include one or more applications to display data feeds and/or real-time events via one or more display devices of client devices 101 , 102 , 103 , 104 , 105 , and/or 106 .
  • the server 120 may be a server of a distributed system, or a server combined with a blockchain.
  • the server 120 may also be a cloud server, or an intelligent cloud computing server or intelligent cloud host with artificial intelligence technology.
  • Cloud server is a host product in the cloud computing service system to solve the shortcomings of difficult management and weak business scalability in traditional physical host and virtual private server (VPS) services.
  • System 100 may also include one or more databases 130.
  • these databases may be used to store data and other information.
  • databases 130 may be used to store information such as audio files and video files.
  • Database 130 may reside in various locations.
  • a database used by server 120 may be local to server 120, or may be remote from server 120 and may communicate with server 120 via a network-based or dedicated connection.
  • Database 130 may be of different types.
  • the database used by server 120 may be, for example, a relational database.
  • One or more of these databases may store, update, and retrieve data to and from the database in response to commands.
  • one or more of databases 130 may also be used by applications to store application data.
  • the database used by the application can be different types of databases such as key-value repositories, object repositories or regular repositories backed by a file system.
  • the system 100 of Figure 1 may be configured and operated in various ways to enable the application of the various methods and apparatus described in accordance with the present disclosure.
  • the collection, storage, use, processing, transmission, provision and disclosure of user personal information are in compliance with relevant laws and regulations and do not violate public order and good customs.
  • the method includes branching modules and delimiting modules. Taking the solution of mixed integer programming problems in the form of a search tree as an example, the main function of the branch module is to select nodes in the search tree and select variables at each node position. The quality of node and variable selection by the branch module will directly affect the quality of solving the entire mixed integer programming problem.
  • the solver corresponding to the mixed integer programming problem can select variables based on heuristic strategies, such as strong branch strategies and pseudo-cost strategies.
  • heuristic strategy has problems such as low decision-making efficiency, poor decision-making effect, and the inability to consider long-term benefits in the decision-making process.
  • a single heuristic strategy cannot efficiently solve many different types of mixed integer programming problems.
  • FIG. 2 shows a flowchart of a task processing method 200 implemented using an integer programming solver, according to an embodiment of the present disclosure.
  • the integer programming solver includes branching modules, and tasks include problems at multiple levels.
  • Method 200 includes using a branch module to sequentially determine at least one target variable for each level of the problem, wherein using the branch module to determine at least one target variable for the current problem includes: step S201, obtaining the currently unfixed variables among the multiple variables related to the task.
  • Distribution information of at least one variable includes type distribution information of at least one variable and type distribution information of at least one variable in the objective function of the task; step S202, based on at least the distribution information of at least one variable, select from a plurality of candidate variables Determine the first target variable selection strategy of the current problem in the strategy; and step S203, use the first target variable selection strategy to determine at least one target variable of the current problem from multiple variables. From this, the most appropriate target variable selection strategy can be determined for each level of problem, so as to be applicable to many different types of mixed integer programming problems, improve decision-making efficiency, and combine the advantages of multiple variable selection strategies to improve decision-making. quality.
  • the task may be, for example, but is not limited to, a packing task or a logistics task. That is, various integer programming problems can be handled.
  • tasks include multiple levels of questions.
  • the hierarchy of multiple questions can be reflected in the form of a search tree.
  • the search tree corresponds to the entire task to be processed, while multiple sub-problems at multiple nodes in the search tree correspond to multiple levels of problems.
  • the method 200 can utilize a branching module to sequentially determine at least one target variable (ie, a decision on the current problem) of each current problem at each node of the search tree, thereby solving the entire to-be-processed task.
  • the method 200 also includes constructing the current node of the search tree based on at least one target variable of the current problem, wherein the problems at multiple levels of the task correspond to the nodes of the corresponding layers of the search tree in turn.
  • the entire task to be processed can be divided into problems at multiple nodes through the search tree, and the entire task to be processed can be solved by solving the problems at each node.
  • the method 200 obtains the distribution information of at least one variable that is not currently fixed among the multiple variables related to the task.
  • the distribution information includes the type distribution information of the at least one variable and the type of the at least one variable in the objective function of the task. Distribution information.
  • the type distribution information of at least one variable in the objective function of the task refers to the ratio between the number of variables of the type to which the variable belongs in the objective function and the number of variables of other types in the objective function. For example, if the objective function includes N variables, including a variables of the first type, b variables of the second type, and c variables of the third type, then the type distribution of the variables in the objective function of the task is The information can be expressed as a:b:c, or a/N:b/N:c/N.
  • each level of the problem may have the same objective function, for example, the same objective function as the entire task to be processed.
  • each level of the problem may have multiple different objective functions, and at least one variable in the multiple objective functions may also have different type distribution information accordingly.
  • the types of the above variables may include: 0-1 integer variables, ordinary integer variables, non-integer variables, etc. Furthermore, it should be understood that a greater number of other types of variables may be present in various embodiments.
  • the method 200 further includes: obtaining distribution information of a plurality of variables related to the task, wherein determining a first target variable selection strategy of the current problem from a plurality of candidate variable selection strategies based on at least the distribution information of at least one variable.
  • the method includes: determining a first target variable selection strategy for the current problem from multiple candidate variable selection strategies based on distribution information of multiple variables and distribution information of at least one variable.
  • the method 200 can also obtain distribution information of a plurality of other variables including the fixed variables.
  • the distribution information of multiple variables related to the task obtained by the method 200 may include fixed and unfixed distribution information of various types of variables among the multiple variables related to the task. For example, there are M variables of the first type in the task to be processed, and when solving the current problem, a total of K variables of the first type have been fixed, then the fixed-unfixed distribution information of the variables of the first type Can be expressed as K/M.
  • the distribution information of multiple variables related to the task obtained by method 200 may also include type distribution information corresponding to the variables of the entire task to be processed, and variables in the objective function of the entire task to be processed. type distribution information. This is different from the type distribution information corresponding to the variables of the current problem and the type distribution information of the variables in the objective function.
  • the type distribution information corresponding to the entire task to be processed will not change as the current problem changes. Therefore, during the process of solving the entire search tree, the type distribution information of the variables corresponding to the entire task to be processed and the type distribution information of the variables in the objective function of the entire task to be processed remain unchanged.
  • the method 200 can further obtain the layer depth of the current problem corresponding to the entire task to be processed, and the number of problems that have been solved in the task to be processed, and determine the target variable selection strategy based on this. That is, according to some embodiments, step S202 in the method 200, at least based on the distribution information of at least one variable, determining the first target variable selection strategy of the current problem from a plurality of candidate variable selection strategies includes: based on the distribution information of the at least one variable , the depth of the current problem and the number of all problems before the current problem.
  • the current variable selection involves not only the distribution information in the current variable, but also the features of the entire search tree corresponding to the problem to be processed.
  • the above content introduces various variable information used to determine the first target variable selection strategy of the current problem from multiple candidate variable selection strategies.
  • the method of creating multiple candidate variable selection strategies will be discussed in detail below.
  • the plurality of candidate variable selection strategies include at least one heuristic strategy and at least one imitation learning model.
  • At least one of the heuristic strategies may include a product-based pseudo-cost strategy, a weighted sum-based pseudo-cost strategy, a confidence-based heuristic strategy, etc.
  • the imitation learning model can be an imitation learning model trained with training data obtained by a heuristic strategy, such as a 64D-GCNN model, an 8D-GCNN model, etc. It should be understood that based on actual usage requirements and the characteristics of different heuristic strategies and imitation learning models, different numbers of heuristic strategies and imitation learning models can be selected to constitute multiple candidate variable selection strategies.
  • the size of multiple candidate variable selection strategies may be 5, including the above three heuristic strategies: the product-based pseudo-cost strategy, the weighted sum-based pseudo-cost strategy, the confidence-based heuristic strategy, and the 64D-GCNN model , 8D-GCNN model and two imitation learning models. From this, multiple candidate variable selection strategies can be constructed, the most appropriate variable selection strategy can be selected for each different problem, and the advantages of multiple variable selection strategies can be combined to improve decision-making quality.
  • the sample data for training the imitation learning model is obtained using a strong branching strategy. Because the decision-making effect of some heuristic strategies such as strong branches is better, but the decision-making efficiency is not high, it requires high computational cost and time. Training the imitation learning model by utilizing sample data obtained by strong branching strategies can improve decision-making efficiency while maintaining high decision-making effectiveness. This improves the processing speed of the overall pending issues.
  • the sample data used to train the imitation learning model is sample data from multiple different types of mixed integer programming problems, including binning sample data and logistics sample data. This forms a sample data training set with good diversity, making the trained imitation learning model better applicable to many different types of mixed integer programming problems.
  • At least distribution information of at least one variable is input into a reinforcement learning model, and the reinforcement learning model is used to determine a target variable selection strategy for the current problem from a plurality of candidate variable selection strategies.
  • the reinforcement learning model can be used to consider the overall situation, such as the long-term benefits of the current decision, etc., thereby improving the quality of decision-making.
  • the distribution information of at least one variable in addition to the distribution information of at least one variable, the distribution information of multiple variables related to the task obtained by the above method 200, the depth of the layer where the current problem is located, and the number of all problems before the current problem can also be input. into a reinforcement learning model to determine the target variable selection strategy for the current problem from multiple candidate variable selection strategies.
  • FIG. 3 shows a flowchart of a task processing method 300 using an integer programming solver according to an embodiment of the present disclosure.
  • the method 300 includes: step S301, based on at least one target variable of the current problem, calculate the global optimal relaxed solution and the global optimal feasible solution; step S302, based on the global optimal relaxed solution and the global optimal feasible solution, calculate the reward function ; Step S303, adjust the parameters of the reinforcement learning model based on the reward function; and step S304, use the adjusted reinforcement learning model to determine the second target variable selection strategy for the next-level problem of the current problem from multiple candidate variable selection strategies.
  • the reinforcement learning model can be updated after each decision to improve the quality of the final solution.
  • the global optimal value after solving the current problem can be calculated based on the at least one target variable.
  • the optimal relaxed solution and the global optimal feasible solution are calculated, and then the reward function brought by the current decision is calculated.
  • the global optimal relaxed solution is the optimal solution of the entire task to be processed after removing corresponding constraints such as integer constraints
  • the global optimal feasible solution is the optimal solution of the entire task to be processed that satisfies corresponding constraints such as integer constraints. Excellent solution.
  • the reward function is the ratio of the global optimal relaxed solution and the global optimal feasible solution.
  • the design of the reward function can match the expected optimal solution of the entire task to be processed to maximize the update effect of the reinforcement learning model.
  • the reinforcement model can be trained using existing sample data to solve, for example, the cold start problem.
  • the sample data for training the reinforcement learning model includes: packing sample data and logistics sample data.
  • sample data corresponding to many different types of integer programming problems can be used to train the reinforcement learning model as needed, so that the trained reinforcement learning model can perform well in various types of mixed integer programming problems. Obtain better solution results.
  • the reinforcement learning model may include a generalized spatial regression (SAC) model, and the SAC model may be trained through the method of "compute-sample results-update model parameters" until the reward function corresponding to the entire task converges to predetermined standards.
  • the actor module in the SAC model can be used to switch the variable selection strategy for the problem at each node position in the search tree.
  • the reinforcement learning model is updated using a temporal difference-based approach. After each decision, the parameters of the reinforcement learning model are adjusted based on the calculated reward function. The adjusted reinforcement learning model is directly used to determine the target variable selection strategy in solving the next problem.
  • FIG. 4 shows a flowchart of a task processing method 400 implemented using an integer programming solver according to some embodiments of the present disclosure.
  • Method 400 includes constructing multiple candidate variable selection strategies, specifically including: step S401, obtaining data based on the strong branch strategy to train multiple imitation learning models; step S402, using multiple imitation learning models and other multiple heuristic strategies to construct multiple imitation learning models.
  • candidate variable selection strategy specifically including: step S401, obtaining data based on the strong branch strategy to train multiple imitation learning models; step S402, using multiple imitation learning models and other multiple heuristic strategies to construct multiple imitation learning models.
  • method 400 includes: step S403, obtaining the characteristic information of the original problem (that is, the entire problem to be processed) and the current problem of node N.
  • the characteristic information may include the above various variable information; step S404.
  • step S405. Select the variable for the node N based on the selected variable selection strategy.
  • the method 400 After selecting the variables of node N, the method 400 also includes the step of updating the reinforcement learning model, which includes: step S406, based on the variables of node N, obtain the global optimal relaxed solution after the decision and the global optimal solution after the decision. Optimum feasible solution; Step S407, calculate the reward function, which can be the ratio between the global optimal relaxed solution and the global optimal feasible solution; Step S408, update the reinforcement learning model based on the reward function, and the reinforcement learning model is further used Repeat the steps in the above method 400 at subsequent nodes (for example, node N+1).
  • FIG. 5 shows a structural block diagram of a task processing device 500 using an integer programming solver according to an embodiment of the present disclosure.
  • the integer programming solver includes a branching module, and the task processing includes problems of multiple levels.
  • the device 500 includes a determining unit 510 configured to use the branching module to determine at least one target variable of the problem of each level in turn, wherein the determining unit 510 includes : Acquisition subunit 511, configured to obtain distribution information of at least one variable that is not currently fixed among multiple variables related to the task.
  • the distribution information includes type distribution information of at least one variable and at least one variable in the objective function of the task.
  • the first determination subunit 512 is configured to determine a first target variable selection strategy for the current problem from a plurality of candidate variable selection strategies based on at least distribution information of at least one variable; and a second determination subunit 513. Be configured to use the first target variable selection strategy to determine at least one target variable of the current problem from a plurality of variables. From this, the most appropriate target variable selection strategy can be determined for each level of problem, so as to be applicable to many different types of mixed integer programming problems, improve decision-making efficiency, and combine the advantages of multiple variable selection strategies to improve decision-making. quality.
  • the task processing device 500 can be used to perform operations similar to the task processing methods 200, 300, and 400 described above, which will not be described again here.
  • an electronic device a readable storage medium, and a computer program product are also provided.
  • an electronic device including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, The instructions are executed by at least one processor, so that the at least one processor can execute the above method.
  • a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause the computer to execute the above method.
  • a computer program product including a computer program, wherein the computer program implements the above method when executed by a processor.
  • Electronic devices are intended to refer to various forms of digital electronic computing equipment, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. Electronic devices may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices, and other similar computing devices.
  • the components shown herein, their connections and relationships, and their functions are examples only and are not intended to limit implementations of the disclosure described and/or claimed herein.
  • the electronic device 600 includes a computing unit 601 that can perform calculations according to a computer program stored in a read-only memory (ROM) 602 or loaded from a storage unit 608 into a random access memory (RAM) 603 . Perform various appropriate actions and processing. In the RAM 603, various programs and data required for the operation of the electronic device 600 can also be stored.
  • Computing unit 601, ROM 602 and RAM 603 are connected to each other via bus 604.
  • An input/output (I/O) interface 605 is also connected to bus 604.
  • the input unit 606 may be any type of device capable of inputting information to the electronic device 600, the input unit 606 may receive input numeric or character information, and generate key signal input related to user settings and/or function control of the electronic device, and This may include, but is not limited to, a mouse, keyboard, touch screen, trackpad, trackball, joystick, microphone, and/or remote control.
  • Output unit 607 may be any type of device capable of presenting information, and may include, but is not limited to, a display, speakers, video/audio output terminal, vibrator, and/or printer.
  • the storage unit 608 may include, but is not limited to, a magnetic disk or an optical disk.
  • the communication unit 609 allows the electronic device 600 to exchange information/data with other devices through a computer network such as the Internet and/or various telecommunications networks, and may include, but is not limited to, a modem, a network card, an infrared communication device, a wireless communication transceiver and/or a chip Groups such as BluetoothTM devices, 802.11 devices, WiFi devices, WiMax devices, cellular communications devices, and/or the like.
  • Computing unit 601 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of the computing unit 601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units that run machine learning model algorithms, digital signal processing processor (DSP), and any appropriate processor, controller, microcontroller, etc.
  • the computing unit 601 performs various methods and processes described above, such as methods 200, 300, and 400.
  • methods 200, 300, 400 may be implemented as a computer software program tangibly embodied in a machine-readable medium, such as storage unit 608.
  • part or all of the computer program may be loaded and/or installed onto the electronic device 600 via the ROM 602 and/or the communication unit 609.
  • the computer program When the computer program is loaded into the RAM 603 and executed by the computing unit 601, one or more steps of the methods 200, 300, 400 described above may be performed.
  • the computing unit 601 may be configured to perform the methods 200, 300, 400 in any other suitable manner (eg, by means of firmware).
  • Various implementations of the systems and techniques described above may be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on a chip implemented in a system (SOC), complex programmable logic device (CPLD), computer hardware, firmware, software, and/or combinations thereof.
  • FPGAs field programmable gate arrays
  • ASICs application specific integrated circuits
  • ASSPs application specific standard products
  • SOC system
  • CPLD complex programmable logic device
  • computer hardware firmware, software, and/or combinations thereof.
  • These various embodiments may include implementation in one or more computer programs executable and/or interpreted on a programmable system including at least one programmable processor, the programmable processor
  • the processor which may be a special purpose or general purpose programmable processor, may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.
  • An output device may be a special purpose or general purpose programmable processor, may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.
  • An output device may be a special purpose or general purpose programmable processor, may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.
  • Program code for implementing the methods of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that the program codes, when executed by the processor or controller, cause the functions specified in the flowcharts and/or block diagrams/ The operation is implemented.
  • the program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.
  • a machine-readable medium may be a tangible medium that may contain or store a program for use by or in connection with an instruction execution system, apparatus, or device.
  • the machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium.
  • Machine-readable media may include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or devices, or any suitable combination of the foregoing.
  • machine-readable storage media would include one or more wire-based electrical connections, laptop disks, hard drives, random access memory (RAM), read only memory (ROM), erasable programmable read only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the above.
  • RAM random access memory
  • ROM read only memory
  • EPROM or flash memory erasable programmable read only memory
  • CD-ROM portable compact disk read-only memory
  • magnetic storage device or any suitable combination of the above.
  • the systems and techniques described herein may be implemented on a computer having a display device (eg, a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user ); and a keyboard and pointing device (eg, a mouse or a trackball) through which a user can provide input to the computer.
  • a display device eg, a CRT (cathode ray tube) or LCD (liquid crystal display) monitor
  • a keyboard and pointing device eg, a mouse or a trackball
  • Other kinds of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and may be provided in any form, including Acoustic input, voice input or tactile input) to receive input from the user.
  • the systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., A user's computer having a graphical user interface or web browser through which the user can interact with implementations of the systems and technologies described herein), or including such backend components, middleware components, or any combination of front-end components in a computing system.
  • the components of the system may be interconnected by any form or medium of digital data communication (eg, a communications network). Examples of communication networks include: local area network (LAN), wide area network (WAN), the Internet, and blockchain networks.
  • Computer systems may include clients and servers.
  • Clients and servers are generally remote from each other and typically interact over a communications network.
  • the relationship of client and server is created by computer programs running on corresponding computers and having a client-server relationship with each other.
  • the server can be a cloud server, a distributed system server, or a server combined with a blockchain.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Software Systems (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Biophysics (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Computational Linguistics (AREA)
  • Artificial Intelligence (AREA)
  • Mathematical Physics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

一种利用整数规划求解器实现的任务处理方法、设备和介质，涉及计算机技术领域，尤其涉及人工智能。实现方案为：获取与任务相关的多个变量中当前未固定的至少一个变量的分布信息（S201），分布信息包括至少一个变量的类型分布信息和至少一个变量在任务的目标函数中的类型分布信息；基于至少一个变量的分布信息，从多个候选变量选择策略中确定当前问题的第一目标变量选择策略（S202）；以及利用第一目标变量选择策略从多个变量中确定当前问题的至少一个目标变量（S203）。

Description

利用整数规划求解器实现的任务处理方法、设备和介质
相关申请的交叉引用
本申请要求于2022年6月2日提交的中国专利申请202210625681.5的优先权，其全部内容通过引用整体结合在本申请中。
技术领域
本公开涉及计算机技术领域，尤其涉及人工智能，具体涉及一种利用整数规划求解器实现的任务处理方法、装置、电子设备、计算机可读存储介质和计算机程序产品。
背景技术
人工智能是研究使计算机来模拟人的某些思维过程和智能行为(如学习、推理、思考、规划等)的学科，既有硬件层面的技术也有软件层面的技术。人工智能硬件技术一般包括如传感器、专用人工智能芯片、云计算、分布式存储、大数据处理等技术；人工智能软件技术主要包括计算机视觉技术、语音识别技术、自然语言处理技术以及机器学习/深度学习、大数据处理技术、知识图谱技术等几大方向。
混合整数规划是运筹优化领域中非常重要的问题，通过对其进行求解可以解决各种领域内的规划问题。如何改善混合整数规划的问题的求解过程，成为了业界关注的焦点。
在此部分中描述的方法不一定是之前已经设想到或采用的方法。除非另有指明，否则不应假定此部分中描述的任何方法仅因其包括在此部分中就被认为是现有技术。类似地，除非另有指明，否则此部分中提及的问题不应认为在任何现有技术中已被公认。
发明内容
本公开提供了一种利用整数规划求解器实现的任务处理方法、装置、电子设备、计算机可读存储介质和计算机程序产品。
根据本公开的一方面，提供了一种利用整数规划求解器实现的任务处理方法，整数规划求解器包括分支模块，任务包括多个层次的问题，该方法包括利用分支模块依次确定每个层次的问题的至少一个目标变量，其中，利用分支模块确定当前问题的至少一个目标变量包括：获取与任务相关的多个变量中当前未固定的至少一个变量的分布信息，分布信息包括至少一个变量的类型分布信息和至少一个变量在任务的目标函数中的类型分布信息；至少基于至少一个变量的分布信息，从多个候选变量选择策略中确定当前问题的第一目标变量选择策略；以及利用第一目标变量选择策略从多个变量中确定当前问题的至少一个目标变量。
根据本公开的另一方面，提供了一种利用整数规划求解器实现的任务处理装置，整数规划求解器包括分支模块，任务包括多个层次的问题，该装置包括确定单元，被配置用于利用分支模块依次确定每个层次的问题的至少一个目标变量，其中，确定单元包括：获取子单元，被配置用于获取与任务相关的多个变量中当前未固定的至少一个变量的分布信息，分布信息包括至少一个变量的类型分布信息和至少一个变量在任务的目标函数中的类型分布信息；第一确定子单元，被配置用于至少基于至少一个变量的分布信息，从多个候选变量选择策略中确定当前问题的第一目标变量选择策略；以及第二确定子单元，被配置用于利用第一目标变量选择策略从多个变量中确定当前问题的至少一个目标变量。
根据本公开的另一方面，提供了一种电子设备，包括：至少一个处理器；以及与至少一个处理器通信连接的存储器；其中存储器存储有可被至少一个处理器执行的指令，指令被至少一个处理器执行，以使至少一个处理器能够执行上述任一方面的方法。
根据本公开的另一方面，提供了一种存储有计算机指令的非瞬时计算机可读存储介质，其中，计算机指令用于使计算机执行上述任一方面的方法。
根据本公开的另一方面，提供了一种计算机程序产品，包括计算机程序，其中，计算机程序在被处理器执行时实现上述任一方面的方法。
根据本公开的一个或多个实施例，可以改善多种类型的混合整数规划问题的求解效率、提高求解效果、并考虑长期收益。
应当理解，本部分所描述的内容并非旨在标识本公开的实施例的关键或重要特征，也不用于限制本公开的范围。本公开的其它特征将通过以下的说明书而变得容易理解。
附图说明
附图示例性地示出了实施例并且构成说明书的一部分，与说明书的文字描述一起用于讲解实施例的示例性实施方式。所示出的实施例仅出于例示的目的，并不限制权利要求的范围。在所有附图中，相同的附图标记指代类似但不一定相同的要素。
图1示出了根据本公开的实施例的可以在其中实施本文描述的各种方法的示例性系统的示意图；
图2示出了根据本公开的实施例的利用整数规划求解器实现的任务处理方法的流程图；
图3示出了根据本公开的实施例的利用整数规划求解器实现的任务处理方法的流程图；
图4示出了根据本公开的实施例的利用整数规划求解器实现的任务处理方法的流程图
图5示出了根据本公开的实施例的利用整数规划求解器实现的任务处理装置的结构框图；
图6示出了能够用于实现本公开的实施例的示例性电子设备的结构框图。
具体实施方式
以下结合附图对本公开的示范性实施例做出说明，其中包括本公开实施例的各种细节以助于理解，应当将它们认为仅仅是示范性的。因此，本领域普通技术人员应当认识到，可以对这里描述的实施例做出各种改变和修改，而不会背离本公开的范围。同样，为了清楚和简明，以下的描述中省略了对公知功能和结构的描述。
在本公开中，除非另有说明，否则使用术语“第一”、“第二”等来描述各种要素不意图限定这些要素的位置关系、时序关系或重要性关系，这种 术语只是用于将一个要素与另一要素区分开。在一些示例中，第一要素和第二要素可以指向该要素的同一实例，而在某些情况下，基于上下文的描述，它们也可以指代不同实例。
在本公开中对各种所述示例的描述中所使用的术语只是为了描述特定示例的目的，而并非旨在进行限制。除非上下文另外明确地表明，如果不特意限定要素的数量，则该要素可以是一个也可以是多个。此外，本公开中所使用的术语“和/或”涵盖所列出的项目中的任何一个以及全部可能的组合方式。
下面将结合附图详细描述本公开的实施例。
图1示出了根据本公开的实施例可以将本文描述的各种方法和装置在其中实施的示例性系统100的示意图。参考图1，该系统100包括一个或多个客户端设备101、102、103、104、105和106、服务器120以及将一个或多个客户端设备耦接到服务器120的一个或多个通信网络110。客户端设备101、102、103、104、105和106可以被配置为执行一个或多个应用程序。
在本公开的实施例中，服务器120可以运行使得能够执行利用整数规划求解器实现的任务处理方法的一个或多个服务或软件应用。
在某些实施例中，服务器120还可以提供其他服务或软件应用，这些服务或软件应用可以包括非虚拟环境和虚拟环境。在某些实施例中，这些服务可以作为基于web的服务或云服务提供，例如在软件即服务(SaaS)模型下提供给客户端设备101、102、103、104、105和/或106的用户。
在图1所示的配置中，服务器120可以包括实现由服务器120执行的功能的一个或多个组件。这些组件可以包括可由一个或多个处理器执行的软件组件、硬件组件或其组合。操作客户端设备101、102、103、104、105和/或106的用户可以依次利用一个或多个客户端应用程序来与服务器120进行交互以利用这些组件提供的服务。应当理解，各种不同的系统配置是可能的，其可以与系统100不同。因此，图1是用于实施本文所描述的各种方法的系统的一个示例，并且不旨在进行限制。
用户可以使用客户端设备101、102、103、104、105和/或106来发送任务处理请求。客户端设备可以提供使客户端设备的用户能够与客户端设备进行交互的接口。客户端设备还可以经由该接口向用户输出信息。尽管图1仅 描绘了六种客户端设备，但是本领域技术人员将能够理解，本公开可以支持任何数量的客户端设备。
客户端设备101、102、103、104、105和/或106可以包括各种类型的计算机设备，例如便携式手持设备、通用计算机(诸如个人计算机和膝上型计算机)、工作站计算机、可穿戴设备、智能屏设备、自助服务终端设备、服务机器人、游戏系统、瘦客户端、各种消息收发设备、传感器或其他感测设备等。这些计算机设备可以运行各种类型和版本的软件应用程序和操作系统，例如MICROSOFT Windows、APPLE iOS、类UNIX操作系统、Linux或类Linux操作系统(例如GOOGLE Chrome OS)；或包括各种移动操作系统，例如MICROSOFT Windows Mobile OS、iOS、Windows Phone、Android。便携式手持设备可以包括蜂窝电话、智能电话、平板电脑、个人数字助理(PDA)等。可穿戴设备可以包括头戴式显示器(诸如智能眼镜)和其他设备。游戏系统可以包括各种手持式游戏设备、支持互联网的游戏设备等。客户端设备能够执行各种不同的应用程序，例如各种与Internet相关的应用程序、通信应用程序(例如电子邮件应用程序)、短消息服务(SMS)应用程序，并且可以使用各种通信协议。
网络110可以是本领域技术人员熟知的任何类型的网络，其可以使用多种可用协议中的任何一种(包括但不限于TCP/IP、SNA、IPX等)来支持数据通信。仅作为示例，一个或多个网络110可以是局域网(LAN)、基于以太网的网络、令牌环、广域网(WAN)、因特网、虚拟网络、虚拟专用网络(VPN)、内部网、外部网、区块链网络、公共交换电话网(PSTN)、红外网络、无线网络(例如蓝牙、WIFI)和/或这些和/或其他网络的任意组合。
服务器120可以包括一个或多个通用计算机、专用服务器计算机(例如PC(个人计算机)服务器、UNIX服务器、中端服务器)、刀片式服务器、大型计算机、服务器群集或任何其他适当的布置和/或组合。服务器120可以包括运行虚拟操作系统的一个或多个虚拟机，或者涉及虚拟化的其他计算架构(例如可以被虚拟化以维护服务器的虚拟存储设备的逻辑存储设备的一个或多个灵活池)。在各种实施例中，服务器120可以运行提供下文所描述的功能的一个或多个服务或软件应用。
服务器120中的计算单元可以运行包括上述任何操作系统以及任何商业上可用的服务器操作系统的一个或多个操作系统。服务器120还可以运行各种附加服务器应用程序和/或中间层应用程序中的任何一个，包括HTTP服务器、FTP服务器、CGI服务器、JAVA服务器、数据库服务器等。
在一些实施方式中，服务器120可以包括一个或多个应用程序，以分析和合并从客户端设备101、102、103、104、105和/或106的用户接收的数据馈送和/或事件更新。服务器120还可以包括一个或多个应用程序，以经由客户端设备101、102、103、104、105和/或106的一个或多个显示设备来显示数据馈送和/或实时事件。
在一些实施方式中，服务器120可以为分布式系统的服务器，或者是结合了区块链的服务器。服务器120也可以是云服务器，或者是带人工智能技术的智能云计算服务器或智能云主机。云服务器是云计算服务体系中的一项主机产品，以解决传统物理主机与虚拟专用服务器(VPS，Virtual Private Server)服务中存在的管理难度大、业务扩展性弱的缺陷。
系统100还可以包括一个或多个数据库130。在某些实施例中，这些数据库可以用于存储数据和其他信息。例如，数据库130中的一个或多个可用于存储诸如音频文件和视频文件的信息。数据库130可以驻留在各种位置。例如，由服务器120使用的数据库可以在服务器120本地，或者可以远离服务器120且可以经由基于网络或专用的连接与服务器120通信。数据库130可以是不同的类型。在某些实施例中，由服务器120使用的数据库例如可以是关系数据库。这些数据库中的一个或多个可以响应于命令而存储、更新和检索到数据库以及来自数据库的数据。
在某些实施例中，数据库130中的一个或多个还可以由应用程序使用来存储应用程序数据。由应用程序使用的数据库可以是不同类型的数据库，例如键值存储库，对象存储库或由文件系统支持的常规存储库。
图1的系统100可以以各种方式配置和操作，以使得能够应用根据本公开所描述的各种方法和装置。
本公开的技术方案中，所涉及的用户个人信息的收集、存储、使用、加工、传输、提供和公开等处理，均符合相关法律法规的规定，且不违背公序良俗。
对于混合整数规划问题，可以使用分支定界法进行求解。该方法包括分支模块和定界模块。以搜索树的形式求解混合整数规划问题为例，分支模块的主要作用为进行搜索树中的节点的选择以及在每个节点位置处变量的选择。分支模块对于节点和变量选择的质量好坏将会直接影响整个混合整数规划问题的求解质量。
混合整数规划问题对应的求解器可以基于启发式策略来进行变量选择，例如强分支策略和伪成本策略等。然而发明人注意到启发式策略存在决策效率低、决策效果差、决策过程无法考虑长期收益等问题。同时，由于不同混合整数规划问题之间的差异，单一的启发式策略也无法高效地解决多种不同类型的混合整数规划问题。
因此，根据本公开的实施例，图2示出了利用整数规划求解器实现的任务处理方法200的流程图。整数规划求解器包括分支模块，任务包括多个层次的问题。方法200包括利用分支模块依次确定每个层次的问题的至少一个目标变量，其中，利用分支模块确定当前问题的至少一个目标变量包括：步骤S201，获取与任务相关的多个变量中当前未固定的至少一个变量的分布信息，分布信息包括至少一个变量的类型分布信息和至少一个变量在任务的目标函数中的类型分布信息；步骤S202，至少基于至少一个变量的分布信息，从多个候选变量选择策略中确定当前问题的第一目标变量选择策略；以及步骤S203，利用第一目标变量选择策略从多个变量中确定当前问题的至少一个目标变量。由此，可以针对每个层次的问题都确定最合适的目标变量选择策略，以适用于多种不同类型的混合整数规划问题，提高决策效率，并且能够结合多个变量选择策略的优点，提升决策质量。任务例如可以但不限于为装箱任务或物流任务。即，可以处理各种整数规划问题。
在一些实施例中，任务中包括的多个层次的问题。多个问题的层次性可以通过搜索树的形式来体现。搜索树对应于整个待处理的任务，而搜索树中的多个节点处的多个子问题对应于多个层次的问题。方法200可以利用分支模块依次确定搜索树的每个节点处的每个当前问题的至少一个目标变量(即，对当前问题的决策)，从而求解整个待处理的任务。具体的，方法200还包括基于当前问题的至少一个目标变量构建搜索树的当前节点，其中，任务的多个层次的问题依次对应所述搜索树相应层的节点。通过搜索树的方式可以 将整个待处理的任务划分为多个节点处问题，通过求解各个节点处的问题来求解整个待处理的任务。
在一些实施例中，随着待处理的任务的求解过程的进行，与任务相关的多个变量中的至少部分变量会被固定，即，被固定的变量在搜索树的后续节点处对后续问题的目标变量的求解过程中不再被改变。方法200在步骤S201中获取与任务相关的多个变量中的当前未固定的至少一个变量的分布信息，该分布信息包括至少一个变量的类型分布信息和至少一个变量在任务的目标函数中的类型分布信息。
至少一个变量的类型分布信息指代该变量所属类型中的变量数量与其他类型中的变量数量之间的比例。示例性地，如果在某一整数规划问题中存在N个变量，其中包括a个第一类型的变量、b个第二类型的变量以及c个第三类型的变量(即，N＝a+b+c)，则第一类型、第二类型、第三类型变量的类型分布信息可以表述为a:b:c，或a/N:b/N:c/N。
至少一个变量在任务的目标函数中的类型分布信息指代该变量所属类型中的变量在目标函数中的数量与其他类型的变量在目标函数中的数量之间的比例。示例性地，如果目标函数中包括N个变量，其中包括a个第一类型的变量、b个第二类型的变量以及c个第三类型的变量，则变量在任务的目标函数中的类型分布信息可以表述为a:b:c，或a/N:b/N:c/N。在一些实施例中，每个层次的问题可以具有相同的目标函数，例如，与整个待处理的任务的目标函数相同。在另一些实施例中，每个层次的问题可以具有各自不同的多个目标函数，至少一个变量在该多个目标函数中也相应地具有不同的类型分布信息。
在一些实施例中，上述变量的类型可以包括：0-1整数变量、普通整数变量、非整数变量等。此外应理解，在各种实施例中可以存在更多数量的其他类型的变量。
根据一些实施例，方法200还包括：获取与任务相关的多个变量的分布信息，其中，至少基于至少一个变量的分布信息，从多个候选变量选择策略确定当前问题的第一目标变量选择策略包括：基于多个变量的分布信息以及至少一个变量的分布信息，从多个候选变量选择策略确定当前问题的第一目 标变量选择策略。由此，可以获得更多种类的分布信息，以更全面地反映各种变量的特征。
除了未固定的至少一个变量的分布信息之外，方法200还可以获得包括已固定的变量的其他多个变量的分布信息。
在一些实施例中，方法200获取的与任务相关的多个变量的分布信息可以包括与任务相关的多个变量中各种类型的变量已固定和未固定的分布信息。例如，对于待处理的任务共有M个第一类型的变量，在求解当前问题时，第一类型的变量中共有K个变量已经被固定，那么第一类型的变量的已固定-未固定分布信息可以被表示为K/M。
在一些实施例中，方法200获取的与任务相关的多个变量的分布信息还可以包括对应于整个待处理的任务的变量的类型分布信息，以及在整个待处理的任务的目标函数中的变量的类型分布信息。这与对应于当前问题的变量的类型分布信息和目标函数中的变量的类型分布信息不同，对应于整个待处理的任务的类型分布信息不会随着当前问题的变化而改变。因而在求解整个搜索树的过程中，对应于整个待处理的任务的变量的类型分布信息以及在整个待处理的任务的目标函数中的变量的类型分布信息保持不变。
在一些实施例中，方法200还可以进一步获取当前问题对应于整个待处理的任务的层深，以及待处理的任务中已经求解的问题数量，并且基于此确定目标变量选择策略。即，根据一些实施例，方法200中的步骤S202，至少基于至少一个变量的分布信息，从多个候选变量选择策略中确定当前问题的第一目标变量选择策略包括：基于至少一个变量的分布信息、当前问题所处的层深和当前问题之前的所有问题的数量。由此，当前变量选择不仅涉及当前变量中的分布信息，还涉及对应于待处理问题的整个搜索树的特征。
上面的内容中介绍了用于从多个候选变量选择策略中确定当前问题的第一目标变量选择策略的各种变量信息，下面将具体讨论多个候选变量选择策略的创建方法。
根据一些实施例，多个候选变量选择策略包括至少一个启发式策略和至少一个模仿学习模型。其中至少一个启发式策略可以包括基于乘积的伪成本策略、基于加权和的伪成本策略、基于置信度的启发式策略等。而模仿学习模型可以是由启发式策略得到的训练数据进行训练而得到的模仿学习模型， 例如64D-GCNN模型、8D-GCNN模型等。应理解，基于实际使用的需求以及不同启发式策略和模仿学习模型的特点，可以选择不同数量的启发式策略和模仿学习模型以构成多个候选变量选择策略。示例性地，多个候选变量选择策略的规模可以为5，包括上述基于乘积的伪成本策略、基于加权和的伪成本策略、基于置信度的启发式策略三种启发式策略以及64D-GCNN模型、8D-GCNN模型两种模仿学习模型。由此，可以构建多个候选变量选择策略，针对每个不同问题选择最合适的变量选择策略，综合多个变量选择策略的优点，提升决策质量。
根据一些实施例，训练模仿学习模型的样本数据为利用强分支策略而获得的。由于例如强分支的部分启发式策略的决策效果较好，但决策效率不高，需要消耗较高的计算成本和时间。通过利用强分支策略获得的样本数据来训练模仿学习模型可以在维持较高决策效果的情况下提高决策效率。从而提高整体待处理的问题的处理速度。
根据一些实施例，用于训练模仿学习模型的样本数据是来自多种不同类型的混合整数规划问题的样本数据，包括装箱样本数据和物流样本数据。由此形成多样性较好的样本数据训练集，使得训练得到的模仿学习模型能够较好地适用于多种不同类型的混合整数规划问题。
在创建了多个候选变量选择策略之后，下面将详细地介绍从多个候选变量选择策略中确定目标变量选择策略的方法。
在一些实施例中，至少将至少一个变量的分布信息输入强化学习模型，利用强化学习模型从多个候选变量选择策略确定当前问题的目标变量选择策略。由此，可以通过强化学习模型来考虑全局情况，例如当前决策的长期收益等，进而提升决策的质量。
应理解地，除了至少一个变量的分布信息之外，上述方法200获取的与任务相关的多个变量的分布信息、当前问题所处的层深、当前问题之前的所有问题的数量也可以被输入到强化学习模型中，以用于从多个候选变量选择策略中确定当前问题的目标变量选择策略。
该强化学习模型使用奖励函数来进行更新。具体地，图3示出了根据本公开的实施例的利用整数规划求解器的任务处理方法300的流程图。其中，方法300包括：步骤S301，基于当前问题的至少一个目标变量，计算全局最 优松弛解和全局最优可行解；步骤S302，基于全局最优松弛解和全局最优可行解，计算奖励函数；步骤S303，基于奖励函数调整强化学习模型的参数；以及步骤S304，利用调参后的强化学习模型从多个候选变量选择策略确定当前问题的下一层次问题的第二目标变量选择策略。由此，可以在每次决策之后对强化学习模型进行更新，提升最终求解质量。
当确定当前问题的目标变量选择策略，并基于该目标变量选择策略选择当前问题的至少一个目标变量之后，基于该至少一个目标变量可以计算在求解当前问题(即，进行当前决策)之后的全局最优松弛解和全局最优可行解，并进而计算当前决策带来的奖励函数。其中，全局最优松弛解为去掉例如整数约束等相应约束之后得到的整个待处理的任务的最优解，而全局最优可行解为满足例如整数约束等相应约束的整个待处理的任务的最优解。
根据一些实施例，奖励函数为全局最优松弛解和全局最优可行解的比值。该奖励函数的设计可以与整个待处理的任务的期望最优解相匹配，以最大化强化学习模型的更新效果。
根据一些实施例，该强化模型可以使用已有的样本数据进行训练以解决例如冷启动问题。该训练强化学习模型的样本数据包括：装箱样本数据和物流样本数据。在实际使用中，根据需要可以使用多种不同类型的整数规划问题对应的样本数据来对强化学习模型进行训练，以使得经训练的强化学习模型能够在各种不同类型的混合整数规划问题中都获得较好的求解结果。
在一些实施例中，强化学习模型可以包括广义空间回归(SAC)模型，可以通过“计算-对结果进行采样-更新模型参数”的方法对SAC模型进行训练，直到对应于整个任务的奖励函数收敛到预定标准。当完成训练之后，可以使用SAC模型中的actor模块来为搜索树中的每个节点位置处的问题切换变量选择策略。
在一些实施例中，使用基于时序差分的方式对强化学习模型进行更新。在每次决策之后基于计算得到的奖励函数对强化学习模型的参数进行调整，被调参之后的强化学习模型被直接用于确定下一问题求解中的目标变量选择策略。
示例性地，图4示出了根据本公开的一些实施例的利用整数规划求解器实现的任务处理方法400的流程图。
方法400包括构建多个候选变量选择策略，具体包括：步骤S401，基于强分支策略得到数据训练多个模仿学习模型；步骤S402，使用多个模仿学习模型和其他多个启发式策略来构建多个候选变量选择策略。
以使用搜索树的情况为例，方法400包括：步骤S403，获取原始问题(即，整个待处理的问题)和节点N的当前问题的特征信息，该特征信息可以包括上述各种变量信息；步骤S404，基于获取的特征信息，通过强化学习模型从已构建的多个候选变量选择策略中，选择当前节点N的变量选择策略；步骤S405，基于所选的变量选择策略，选择节点N的变量。
在选择得到节点N的变量之后，方法400还包括对强化学习模型进行更新的步骤，即包括：步骤S406，基于节点N的变量得到此次决策之后的全局最优松弛解和决策之后的全局最优可行解；步骤S407，计算奖励函数，奖励函数可以是全局最优松弛解和全局最优可行解之间的比值；步骤S408，基于奖励函数，更新强化学习模型，该强化学习模型被进一步用于之后的节点(例如节点N+1)，以重复上述方法400中的步骤。
图5示出了根据本公开的实施例的利用整数规划求解器的任务处理装置500的结构框图。整数规划求解器包括分支模块，任务处理包括多个层次的问题，装置500包括确定单元510，被配置用于利用分支模块依次确定每个层次的问题的至少一个目标变量，其中，确定单元510包括：获取子单元511，被配置用于获取与任务相关的多个变量中当前未固定的至少一个变量的分布信息，分布信息包括至少一个变量的类型分布信息和至少一个变量在任务的目标函数中的类型分布信息；第一确定子单元512，被配置用于至少基于至少一个变量的分布信息，从多个候选变量选择策略中确定当前问题的第一目标变量选择策略；以及第二确定子单元513，被配置用于利用第一目标变量选择策略从多个变量中确定当前问题的至少一个目标变量。由此，可以针对每个层次的问题都确定最合适的目标变量选择策略，以适用于多种不同类型的混合整数规划问题，提高决策效率，并且能够结合多个变量选择策略的优点，提升决策质量。
任务处理装置500可以被配合用于执行前面描述的任务处理方法200、300、400类似的操作，在此不作赘述。
根据本公开的实施例，还提供了一种电子设备、一种可读存储介质和一种计算机程序产品。
具体地，根据本公开的另一方面，提供了一种电子设备，包括：至少一个处理器；以及与至少一个处理器通信连接的存储器；其中存储器存储有可被至少一个处理器执行的指令，指令被至少一个处理器执行，以使至少一个处理器能够执行上述的方法。
根据本公开的另一方面，还提供一种存储有计算机指令的非瞬时计算机可读存储介质，其中，计算机指令用于使计算机执行上述的方法。
根据本公开的另一方面，还提供一种计算机程序产品，包括计算机程序，其中，计算机程序在被处理器执行时实现上述的方法。
参考图6，现将描述可以作为本公开的服务器或客户端的电子设备600的结构框图，其是可以应用于本公开的各方面的硬件设备的示例。电子设备旨在表示各种形式的数字电子的计算机设备，诸如，膝上型计算机、台式计算机、工作台、个人数字助理、服务器、刀片式服务器、大型计算机、和其它适合的计算机。电子设备还可以表示各种形式的移动装置，诸如，个人数字处理、蜂窝电话、智能电话、可穿戴设备和其它类似的计算装置。本文所示的部件、它们的连接和关系、以及它们的功能仅仅作为示例，并且不意在限制本文中描述的和/或者要求的本公开的实现。
如图6所示，电子设备600包括计算单元601，其可以根据存储在只读存储器(ROM)602中的计算机程序或者从存储单元608加载到随机访问存储器(RAM)603中的计算机程序，来执行各种适当的动作和处理。在RAM 603中，还可存储电子设备600操作所需的各种程序和数据。计算单元601、ROM 602以及RAM 603通过总线604彼此相连。输入/输出(I/O)接口605也连接至总线604。
电子设备600中的多个部件连接至I/O接口605，包括：输入单元606、输出单元607、存储单元608以及通信单元609。输入单元606可以是能向电子设备600输入信息的任何类型的设备，输入单元606可以接收输入的数字或字符信息，以及产生与电子设备的用户设置和/或功能控制有关的键信号输入，并且可以包括但不限于鼠标、键盘、触摸屏、轨迹板、轨迹球、操作杆、麦克风和/或遥控器。输出单元607可以是能呈现信息的任何类型的设 备，并且可以包括但不限于显示器、扬声器、视频/音频输出终端、振动器和/或打印机。存储单元608可以包括但不限于磁盘、光盘。通信单元609允许电子设备600通过诸如因特网的计算机网络和/或各种电信网络与其他设备交换信息/数据，并且可以包括但不限于调制解调器、网卡、红外通信设备、无线通信收发机和/或芯片组，例如蓝牙TM设备、802.11设备、WiFi设备、WiMax设备、蜂窝通信设备和/或类似物。
计算单元601可以是各种具有处理和计算能力的通用和/或专用处理组件。计算单元601的一些示例包括但不限于中央处理单元(CPU)、图形处理单元(GPU)、各种专用的人工智能(AI)计算芯片、各种运行机器学习模型算法的计算单元、数字信号处理器(DSP)、以及任何适当的处理器、控制器、微控制器等。计算单元601执行上文所描述的各个方法和处理，例如方法200、300、400。例如，在一些实施例中，方法200、300、400可被实现为计算机软件程序，其被有形地包含于机器可读介质，例如存储单元608。在一些实施例中，计算机程序的部分或者全部可以经由ROM 602和/或通信单元609而被载入和/或安装到电子设备600上。当计算机程序加载到RAM 603并由计算单元601执行时，可以执行上文描述的方法200、300、400的一个或多个步骤。备选地，在其他实施例中，计算单元601可以通过其他任何适当的方式(例如，借助于固件)而被配置为执行方法200、300、400。
本文中以上描述的系统和技术的各种实施方式可以在数字电子电路系统、集成电路系统、场可编程门阵列(FPGA)、专用集成电路(ASIC)、专用标准产品(ASSP)、芯片上系统的系统(SOC)、复杂可编程逻辑设备(CPLD)、计算机硬件、固件、软件、和/或它们的组合中实现。这些各种实施方式可以包括：实施在一个或者多个计算机程序中，该一个或者多个计算机程序可在包括至少一个可编程处理器的可编程系统上执行和/或解释，该可编程处理器可以是专用或者通用可编程处理器，可以从存储系统、至少一个输入装置、和至少一个输出装置接收数据和指令，并且将数据和指令传输至该存储系统、该至少一个输入装置、和该至少一个输出装置。
用于实施本公开的方法的程序代码可以采用一个或多个编程语言的任何组合来编写。这些程序代码可以提供给通用计算机、专用计算机或其他可编程数据处理装置的处理器或控制器，使得程序代码当由处理器或控制器执 行时使流程图和/或框图中所规定的功能/操作被实施。程序代码可以完全在机器上执行、部分地在机器上执行，作为独立软件包部分地在机器上执行且部分地在远程机器上执行或完全在远程机器或服务器上执行。
在本公开的上下文中，机器可读介质可以是有形的介质，其可以包含或存储以供指令执行系统、装置或设备使用或与指令执行系统、装置或设备结合地使用的程序。机器可读介质可以是机器可读信号介质或机器可读储存介质。机器可读介质可以包括但不限于电子的、磁性的、光学的、电磁的、红外的、或半导体系统、装置或设备，或者上述内容的任何合适组合。机器可读存储介质的更具体示例会包括基于一个或多个线的电气连接、便携式计算机盘、硬盘、随机存取存储器(RAM)、只读存储器(ROM)、可擦除可编程只读存储器(EPROM或快闪存储器)、光纤、便捷式紧凑盘只读存储器(CD-ROM)、光学储存设备、磁储存设备、或上述内容的任何合适组合。
为了提供与用户的交互，可以在计算机上实施此处描述的系统和技术，该计算机具有：用于向用户显示信息的显示装置(例如，CRT(阴极射线管)或者LCD(液晶显示器)监视器)；以及键盘和指向装置(例如，鼠标或者轨迹球)，用户可以通过该键盘和该指向装置来将输入提供给计算机。其它种类的装置还可以用于提供与用户的交互；例如，提供给用户的反馈可以是任何形式的传感反馈(例如，视觉反馈、听觉反馈、或者触觉反馈)；并且可以用任何形式(包括声输入、语音输入或者、触觉输入)来接收来自用户的输入。
可以将此处描述的系统和技术实施在包括后台部件的计算系统(例如，作为数据服务器)、或者包括中间件部件的计算系统(例如，应用服务器)、或者包括前端部件的计算系统(例如，具有图形用户界面或者网络浏览器的用户计算机，用户可以通过该图形用户界面或者该网络浏览器来与此处描述的系统和技术的实施方式交互)、或者包括这种后台部件、中间件部件、或者前端部件的任何组合的计算系统中。可以通过任何形式或者介质的数字数据通信(例如，通信网络)来将系统的部件相互连接。通信网络的示例包括：局域网(LAN)、广域网(WAN)、互联网和区块链网络。
计算机系统可以包括客户端和服务器。客户端和服务器一般远离彼此并且通常通过通信网络进行交互。通过在相应的计算机上运行并且彼此具有客 户端-服务器关系的计算机程序来产生客户端和服务器的关系。服务器可以是云服务器，也可以为分布式系统的服务器，或者是结合了区块链的服务器。
应该理解，可以使用上面所示的各种形式的流程，重新排序、增加或删除步骤。例如，本公开中记载的各步骤可以并行地执行、也可以顺序地或以不同的次序执行，只要能够实现本公开公开的技术方案所期望的结果，本文在此不进行限制。
虽然已经参照附图描述了本公开的实施例或示例，但应理解，上述的方法、系统和设备仅仅是示例性的实施例或示例，本发明的范围并不由这些实施例或示例限制，而是仅由授权后的权利要求书及其等同范围来限定。实施例或示例中的各种要素可以被省略或者可由其等同要素替代。此外，可以通过不同于本公开中描述的次序来执行各步骤。进一步地，可以以各种方式组合实施例或示例中的各种要素。重要的是随着技术的演进，在此描述的很多要素可以由本公开之后出现的等同要素进行替换。

Claims (16)

  1. 一种利用整数规划求解器实现的任务处理方法，所述整数规划求解器包括分支模块，所述任务包括多个层次的问题，所述方法包括利用分支模块依次确定每个层次的问题的至少一个目标变量，其中，所述利用分支模块确定当前问题的至少一个目标变量包括：
    获取与所述任务相关的多个变量中当前未固定的至少一个变量的分布信息，所述分布信息包括所述至少一个变量的类型分布信息和所述至少一个变量在所述任务的目标函数中的类型分布信息；
    至少基于所述至少一个变量的分布信息，从多个候选变量选择策略中确定所述当前问题的第一目标变量选择策略；以及
    利用所述第一目标变量选择策略从所述多个变量中确定所述当前问题的至少一个目标变量。
  2. 根据权利要求1所述的方法，还包括：
    获取与所述任务相关的多个变量的分布信息，
    其中，至少基于所述至少一个变量的分布信息，从多个候选变量选择策略确定所述当前问题的第一目标变量选择策略包括：
    基于所述多个变量的分布信息以及所述至少一个变量的分布信息，从多个候选变量选择策略确定所述当前问题的第一目标变量选择策略。
  3. 根据权利要求1或2所述的方法，其中，至少将所述至少一个变量的分布信息输入强化学习模型，利用所述强化学习模型从多个候选变量选择策略确定所述当前问题的目标变量选择策略。
  4. 根据权利要求3所述的方法，还包括：
    基于当前问题的至少一个目标变量，计算全局最优松弛解和全局最优可行解；
    基于所述全局最优松弛解和全局最优可行解，计算奖励函数；
    基于所述奖励函数调整所述强化学习模型的参数；以及
    利用调参后的强化学习模型从多个候选变量选择策略确定所述当前问题的下一层次问题的第二目标变量选择策略。
  5. 根据权利要求4所述的方法，其中，所述奖励函数为所述全局最优松弛解和全局最优可行解的比值。
  6. 根据权利要求3所述的方法，其中，训练所述强化学习模型的样本数据包括：装箱样本数据和物流样本数据。
  7. 根据权利要1所述的方法，其中，所述多个候选变量选择策略包括至少一个启发式策略和至少一个模仿学习模型。
  8. 根据权利要求7所述的方法，其中，训练所述模仿学习模型的样本数据为利用强分支策略而获得的。
  9. 根据权利要求7所述的方法，其中，训练所述模仿学习模型的样本数据包括：装箱样本数据和物流样本数据。
  10. 根据权利要求1-9中任一项所述的方法，其中，至少基于所述至少一个变量的分布信息，从多个候选变量选择策略中确定所述当前问题的第一目标变量选择策略包括：
    基于所述至少一个变量的分布信息、所述当前问题所处的层深和所述当前问题之前的所有问题的数量，从多个候选变量选择策略中确定所述当前问题的第一目标变量选择策略。
  11. 根据权利要求1-10中任一项所述的方法，其中，所述任务为装箱任务或物流任务。
  12. 根据权利要求1-10中任一项所述的方法，还包括：
    基于所述当前问题的至少一个目标变量构建搜索树的当前节点，其中，所述任务的多个层次的问题依次对应所述搜索树相应层的节点。
  13. 一种利用整数规划求解器实现的任务处理装置，所述整数规划求解器包括分支模块，所述任务包括多个层次的问题，所述装置包括确定单元，被配置用于利用分支模块依次确定每个层次的问题的至少一个目标变量，其中，所述确定单元包括：
    获取子单元，被配置用于获取与所述任务相关的多个变量中当前未固定的至少一个变量的分布信息，所述分布信息包括所述至少一个变量的类型分布信息和所述至少一个变量在所述任务的目标函数中的类型分布信息；
    第一确定子单元，被配置用于至少基于所述至少一个变量的分布信息，从多个候选变量选择策略中确定所述当前问题的第一目标变量选择策略；以及
    第二确定子单元，被配置用于利用所述第一目标变量选择策略从所述多个变量中确定所述当前问题的至少一个目标变量。
  14. 一种电子设备，包括：
    至少一个处理器；以及
    与所述至少一个处理器通信连接的存储器；其中
    所述存储器存储有可被所述至少一个处理器执行的指令，所述指令被所述至少一个处理器执行，以使所述至少一个处理器能够执行权利要求1-12中任一项所述的方法。
  15. 一种存储有计算机指令的非瞬时计算机可读存储介质，其中，所述计算机指令用于使所述计算机执行根据权利要求1-12中任一项所述的方法。
  16. 一种计算机程序产品，包括计算机程序，其中，所述计算机程序在被处理器执行时实现权利要求1-12中任一项所述的方法。
PCT/CN2022/136280 2022-06-02 2022-12-02 利用整数规划求解器实现的任务处理方法、设备和介质 WO2023231350A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210625681.5A CN114924862A (zh) 2022-06-02 2022-06-02 利用整数规划求解器实现的任务处理方法、设备和介质
CN202210625681.5 2022-06-02

Publications (1)

Publication Number Publication Date
WO2023231350A1 true WO2023231350A1 (zh) 2023-12-07

Family

ID=82813423

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/136280 WO2023231350A1 (zh) 2022-06-02 2022-12-02 利用整数规划求解器实现的任务处理方法、设备和介质

Country Status (2)

Country Link
CN (1) CN114924862A (zh)
WO (1) WO2023231350A1 (zh)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114924862A (zh) * 2022-06-02 2022-08-19 北京百度网讯科技有限公司 利用整数规划求解器实现的任务处理方法、设备和介质
CN115328638B (zh) * 2022-10-13 2023-01-10 北京航空航天大学 一种基于混合整数规划的多飞行器任务调度方法

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160335223A1 (en) * 2014-06-27 2016-11-17 University Of South Florida Methods and systems for computation of bilevel mixed integer programming problems
CN111611324A (zh) * 2020-05-06 2020-09-01 中国科学院信息工程研究所 一种跨域访问策略优化方法及装置
CN111915060A (zh) * 2020-06-30 2020-11-10 华为技术有限公司 组合优化任务的处理方法以及处理装置
CN112215162A (zh) * 2020-10-13 2021-01-12 北京中电兴发科技有限公司 一种基于mcnn网络的多标签多任务人脸属性预测方法
CN114924862A (zh) * 2022-06-02 2022-08-19 北京百度网讯科技有限公司 利用整数规划求解器实现的任务处理方法、设备和介质

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160335223A1 (en) * 2014-06-27 2016-11-17 University Of South Florida Methods and systems for computation of bilevel mixed integer programming problems
CN111611324A (zh) * 2020-05-06 2020-09-01 中国科学院信息工程研究所 一种跨域访问策略优化方法及装置
CN111915060A (zh) * 2020-06-30 2020-11-10 华为技术有限公司 组合优化任务的处理方法以及处理装置
CN112215162A (zh) * 2020-10-13 2021-01-12 北京中电兴发科技有限公司 一种基于mcnn网络的多标签多任务人脸属性预测方法
CN114924862A (zh) * 2022-06-02 2022-08-19 北京百度网讯科技有限公司 利用整数规划求解器实现的任务处理方法、设备和介质

Also Published As

Publication number Publication date
CN114924862A (zh) 2022-08-19

Similar Documents

Publication Publication Date Title
WO2023231350A1 (zh) 利用整数规划求解器实现的任务处理方法、设备和介质
JP7316453B2 (ja) オブジェクト推薦方法及び装置、コンピュータ機器並びに媒体
WO2023221422A1 (zh) 用于文本识别的神经网络及其训练方法、文本识别的方法
US20240028909A1 (en) Deep learning model based data generation
CN112559721B (zh) 人机对话系统的调整方法、装置、设备、介质和程序产品
WO2023245938A1 (zh) 对象推荐方法和装置
WO2023142406A1 (zh) 排序方法、排序模型的训练方法、装置、电子设备及介质
JP2023075239A (ja) 連合学習方法、装置、機器及び媒体
CN114091672B (zh) 分布式模型推理方法及装置、电子设备和介质
KR20230006601A (ko) 정렬 방법, 정렬 모델의 트레이닝 방법, 장치, 전자 기기 및 매체
CN115114424A (zh) 查询请求的应答方法及装置
CN116541536B (zh) 知识增强的内容生成系统、数据生成方法、设备和介质
CN115879469B (zh) 文本数据处理方法、模型训练方法、装置及介质
CN115600646B (zh) 语言模型的训练方法、装置、介质及设备
WO2023240833A1 (zh) 信息推荐方法及装置、电子设备和介质
WO2023142408A1 (zh) 数据处理方法和用于训练预测模型的方法
CN114676062A (zh) 用于接口的差异数据测试方法及装置、电子设备和介质
CN114881170A (zh) 用于对话任务的神经网络的训练方法和对话任务处理方法
CN112860681A (zh) 数据清洗方法及装置、计算机设备和介质
EP4155983A2 (en) Data processing method, apparatus, storage medium, and computer program product
CN115809364B (zh) 对象推荐方法和模型训练方法
CN116860328A (zh) 生成指令数据的方法、装置、设备和介质
CN117909591A (zh) 内容推荐方法、装置、电子设备、介质
CN116541537A (zh) 一种基于知识图谱的企业贸易信息可视化展示方法
CN116541090A (zh) 数据处理方法、装置、设备和介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22944651

Country of ref document: EP

Kind code of ref document: A1