Task processing method and device

Abstract

The embodiments of the present specification provide a task processing method and device, wherein a task processing method comprises: after obtaining variable information of candidate variables on a task node of a to-be-processed task, calculating variable relationships between the candidate variables by means of the variable information of the candidate variables, and performing information fusion processing by means of associated variable information of associated variables of the candidate variables and the variable relationships between the candidate variables and the associated variables, to obtain fusion information of the candidate variables, determining a target variable in the candidate variables according to the fusion information of the candidate variables, and performing corresponding task processing on a sub-node of the task node.

Description

Technical Field

[0001] This document relates to the field of data processing technology, and in particular to a task processing method and apparatus. Background Technology

[0002] With the continuous development of Internet technology, more and more pending tasks are being processed through the Internet, such as container scheduling tasks, traffic management tasks, and factory site selection tasks. In the process of processing pending tasks, it may be necessary for various service providers to cooperate. For example, pending tasks may be divided into various task nodes. In view of this, higher requirements are placed on the task processing of pending tasks by various service providers. Summary of the Invention

[0003] This specification provides one or more embodiments of a task processing method, comprising: acquiring variable information of candidate variables on a task node of a task to be processed; calculating variable relationships between the candidate variables based on the variable information of the candidate variables; performing information fusion processing based on the correlation variable information of the candidate variables and the variable relationships between the candidate variables and the correlation variables to obtain fusion information of the candidate variables; and determining a target variable among the candidate variables based on the fusion information, so as to perform corresponding task processing on the child nodes of the task node based on the target variable.

[0004] This specification provides one or more embodiments of a task processing apparatus, including: an information acquisition module configured to acquire variable information of candidate variables on a task node of a task to be processed; a relationship calculation module configured to calculate the variable relationships between the candidate variables based on the variable information of the candidate variables; a fusion processing module configured to perform information fusion processing based on the association variable information of the associated variables of the candidate variables and the variable relationships between the candidate variables and the associated variables to obtain fusion information of the candidate variables; and a variable determination module configured to determine a target variable among the candidate variables based on the fusion information, so as to perform corresponding task processing on the child nodes of the task node based on the target variable.

[0005] This specification provides one or more embodiments of a task processing device, including: a processor; and a memory configured to store computer-executable instructions, which, when executed, cause the processor to: acquire variable information of candidate variables on a task node to be processed; calculate variable relationships between the candidate variables based on the variable information of the candidate variables; perform information fusion processing based on the correlation variable information of the candidate variables and the variable relationships between the candidate variables and the correlation variables to obtain fused information of the candidate variables; and determine a target variable among the candidate variables based on the fused information, so as to perform corresponding task processing on the child nodes of the task node based on the target variable.

[0006] This specification provides one or more embodiments of a storage medium for storing computer-executable instructions. When executed by a processor, these instructions implement the following process: obtaining variable information of candidate variables on a task node of a task to be processed; calculating variable relationships between the candidate variables based on the variable information of the candidate variables; performing information fusion processing based on the correlation variable information of the candidate variables and the variable relationships between the candidate variables and the correlation variables to obtain fused information of the candidate variables; and determining a target variable among the candidate variables based on the fused information, so as to perform corresponding task processing on the child nodes of the task node based on the target variable. Attached Figure Description

[0007] To more clearly illustrate the technical solutions in one or more embodiments of this specification or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1 A schematic diagram of an implementation environment for a task processing method provided in one or more embodiments of this specification;

[0009] Figure 2 A flowchart illustrating a task processing method provided in one or more embodiments of this specification;

[0010] Figure 3 A flowchart illustrating a task processing method applied to a container scheduling scenario, provided by one or more embodiments of this specification;

[0011] Figure 4 A schematic diagram of one or more embodiments of a task processing device provided in this specification;

[0012] Figure 5 This is a schematic diagram of the structure of a task processing device provided for one or more embodiments of this specification. Detailed Implementation

[0013] To enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the technical solutions in one or more embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this specification, and not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of this document.

[0014] Reference Figure 1 This specification provides a schematic diagram of the implementation environment for one or more embodiments of the task processing method.

[0015] The task processing method provided in one or more embodiments of this specification can be applied to the implementation environment of determining the target variable based on the variable information of candidate variables on the task node of the task to be processed. The implementation environment of this method includes at least a server 101 for determining the target variable.

[0016] Server 101 can be a single server, a server cluster consisting of several servers, or one or more cloud servers in a cloud computing platform. Server 101 is used to determine the target variable based on the variable information of the candidate variables on the task node of the task to be processed.

[0017] In this implementation environment, after obtaining the variable information of candidate variables on the task node of the task to be processed, server 101 calculates the variable relationship between candidate variables using the variable information of candidate variables, and performs information fusion processing using the variable information of the associated variables of candidate variables and the variable relationship between candidate variables and associated variables to obtain fused information of candidate variables. Based on this, the target variable is determined from the candidate variables using the fused information of candidate variables, so that the corresponding task processing is performed on the child nodes of the task node based on the target variable. In this way, by introducing the variable relationship between candidate variables, the effectiveness and comprehensiveness of the fused information of candidate variables are improved, the efficiency of the determination of the target variable is improved, and thus the task processing efficiency of the task to be processed is improved.

[0018] One or more embodiments of a task processing method provided in this specification are as follows:

[0019] Reference Figure 2 The task processing method provided in this embodiment specifically includes steps S202 to S208.

[0020] Step S202: Obtain the variable information of candidate variables on the task node of the task to be processed.

[0021] In this embodiment, the task to be processed refers to a task waiting to be processed, such as a container scheduling task, a traffic management task, or an address selection task. For example, the task to be processed could be an integer linear programming task for container scheduling, an integer linear programming task for traffic management, or an integer linear programming task for address selection. Specifically, the task to be processed can be a problem to be solved, and the integer linear programming task can be a mixed integer linear programming problem. Furthermore, the task to be processed can also be an integer linear programming task from other domains.

[0022] The task node refers to the node involved in the processing of the task to be processed. In practical applications, during the processing of the task to be processed, such as the processing of a mixed integer linear programming task for container scheduling, the branch and bound method can be used to process the mixed integer linear programming task. During the task processing, a branch and bound tree can be generated, and the nodes in the branch and bound tree can be called task nodes.

[0023] Optionally, the task node of the task to be processed corresponds to the task to be processed or a subtask of the task to be processed; when the task node of the task to be processed corresponds to the task to be processed, the task node of the task to be processed can be the main task node; when the task node of the task to be processed corresponds to a subtask of the task to be processed, the task node of the task to be processed can be a subtask node, and there can be multiple subtask nodes; for example, the integer linear programming task of container scheduling is:

[0024] max z = 4x1 + 3x2

[0025]

[0026] Where max z = 4x1 + 3x2 represents the objective function of the integer linear programming task of container scheduling; st represents the constraint condition of the integer linear programming task of container scheduling; x1 and x2 are integers, representing integer constraints; x1 and x2 both represent variables in the integer linear programming task.

[0027] The candidate variables include candidate variables on the task nodes of the task to be processed; optionally, the candidate variables are determined based on the fact that the relaxation result of the relaxation task corresponding to the task node is a non-integer variable and the variable attribute is an integer variable; the variable information includes the relaxation result of the candidate variable and / or the upper limit value of the objective function of the task node where the candidate variable is located. The variable information may be the variable characteristics of the candidate variable, specifically a variable characteristic matrix. There are multiple candidate variables; optionally, the candidate variables are determined from the integer variables corresponding to the task to be processed, where integer variables refer to variables with integer attributes, such as x1 and x2 in the above example; the relaxation task corresponding to the task node refers to the variable value when the integer constraints in the task corresponding to the task node are ignored.

[0028] In practical applications, when a task processing platform processes a task, such as the integer linear programming task for container scheduling mentioned above, it can use the branch and bound method to process the integer linear programming task for container scheduling. Taking the task node as the main task node as an example, the main task node determines the relaxation task of the task to be processed and calculates the relaxation result of the relaxation task. In the relaxation result of the relaxation task, variables whose relaxation result is not an integer and whose variable attribute is an integer are determined. Candidate variables of the main task node can be determined from these variables (one or more of these variables can be selected, or the variable can be used as a candidate variable). The variable information of the candidate variables on the main task node is sent to the server. This embodiment can be applied to the server.

[0029] Subsequently, the server determines the target variable from the candidate variables using this embodiment. The target variable can be sent to the task processing platform. For example, if the relaxation result of the task to be processed on the task processing platform is (x1 = 1.2, x2 = 2.1), the variables in the relaxation result that are non-integers with integer attributes include x1 and x2. Among the variables x1 and x2 whose relaxation results are non-integers with integer attributes, candidate variables for the main task node of the task to be processed are determined as x1 and x2. In this embodiment, the server can obtain the candidate variables on the main task node and determine the target variable x1 from x1 and x2. After obtaining the target variable x1, the task processing platform determines the sub-task nodes of the main task node based on the relaxation result of the target variable x1. The sub-tasks corresponding to the two sub-task nodes are p1 and p2, respectively. Specifically, the first constraint x1 ≤ 1 and the second constraint x1 ≥ 2 can be determined based on the relaxation result of the target variable x1. The task to be processed is updated based on the first and second constraints to obtain the first sub-task p1 and the second sub-task p2.

[0030]

[0031]

[0032] The task processing platform calculates the relaxation results of the first subtask and / or the second subtask at nodes p1 and / or p2, and determines candidate variables based on the relaxation results, and sends them to the server in this embodiment. It should be noted that in this embodiment, during the task processing of a task to be processed, the variable information of the candidate variables on the task nodes of the task to be processed can be obtained once or multiple times until the task processing result of the task to be processed is obtained; the task nodes corresponding to the variable information of the candidate variables are different each time.

[0033] It should be added that the above description of the process of obtaining variable information of candidate variables, taking the task node as the main task node, is similar to the process of obtaining variable information of candidate variables on the task nodes of other tasks to be processed. It will not be repeated here. This embodiment is the process of determining the target variable from the candidate variables after obtaining the variable information of candidate variables on the task nodes of the tasks to be processed sent by the task processing platform, so as to achieve efficient task processing of the tasks to be processed through the target variable.

[0034] Step S204: Calculate the variable relationships between the candidate variables based on the variable information of the candidate variables.

[0035] In the previous step of obtaining the variable information of candidate variables on the task node of the task to be processed, in order to improve the comprehensiveness and effectiveness of the variable information of candidate variables, the related variables of candidate variables can be introduced. Specifically, the variable relationship between candidate variables can be calculated based on the variable information of candidate variables.

[0036] The associated variable mentioned in this embodiment refers to a variable that has an association relationship with the candidate variable. Optionally, the associated variable of any candidate variable includes the candidate variable itself, that is, the associated variable of any candidate variable includes the candidate variable and the remaining variables of the candidate variable excluding the candidate variable. In addition, the associated variable of the candidate variable can also be a variable adjacent to the candidate variable. Specifically, it can be a variable connected to the candidate variable in a graph network, an associated variable whose variable relationship with the candidate variable is a preset variable relationship, or an associated variable whose variable relationship index with the candidate variable is greater than a preset index threshold.

[0037] The variable relationships include variable relationship indicators, such as variable relationship values. The larger the variable relationship value, the closer the relationship between the candidate variables. For example, the variable relationship indicator is m. The variable relationships include indicators used to characterize the closeness of the relationship between each pair of candidate variables.

[0038] In specific implementation, in order to improve the convenience of determining the variable relationship between candidate variables, meet the diverse needs of determining the variable relationship between candidate variables, and at the same time improve the flexibility of determining the variable relationship, the variable relationship between every two candidate variables can be calculated based on the variable information of the candidate variables. Specifically, the variable relationship between candidate variables can be determined through any one of the following three optional implementation methods.

[0039] In the first optional implementation provided in this embodiment, during the process of calculating the variable relationship between candidate variables based on the variable information of the candidate variables, the following operations are performed:

[0040] Determine the constraint factors for the first and second candidate variables among the candidate variables;

[0041] Based on the variable information of the first candidate variable, the variable information of the second candidate variable, and the constraint factor, the variable relationship index between the first candidate variable and the second candidate variable is calculated.

[0042] The constraint factor includes constraints containing a first candidate variable and a second candidate variable. For example, if the candidate variables on the task node of the integer linear programming task for container scheduling are x1 and x2, then the constraint factor could be the constraints 4x1+2x2≤9 and / or 3x1+4x2≤12. The variable information of the first candidate variable includes the variable characteristics of the first candidate variable; the variable information of the second candidate variable includes the variable characteristics of the second candidate variable, specifically, it can be a variable characteristic matrix.

[0043] Specifically, in the process of calculating the variable relationship index between the first and second candidate variables based on the variable information of the first candidate variable, the variable information of the second candidate variable, and the constraint factors, the variable state of the first and second candidate variables can be determined based on the constraint factors of the first and second candidate variables. That is, if the constraint factors of the first and second candidate variables have constraints, the variable state of the first and second candidate variables is determined to be a variable association state; if the constraint factors of the first and second candidate variables do not have constraints, the connection state of the first and second candidate variables is determined to be a variable non-association state. On this basis, the variable relationship index between the first and second candidate variables is calculated according to the variable information of the first and second candidate variables and the variable state of the first and second candidate variables. The variable association state includes the state where there is an association between the first and second candidate variables; the variable non-association state includes the state where there is no association between the first and second candidate variables.

[0044] Furthermore, in the process of calculating the variable relationship between the candidate variables based on the variable information of the candidate variables, the constraint factors of the first candidate variable and the second candidate variable, as well as the variable characteristics of the first candidate variable and the second candidate variable, can be input into the first relationship calculation algorithm to calculate the variable relationship index and obtain the variable relationship index between the first candidate variable and the second candidate variable.

[0045] For example, the algorithm for calculating the first relation is:

[0046] e ij =S(x i ,x j ,ψ ij )

[0047] Among them, e ij A variable relationship indicator representing the relationship between the first and second candidate variables; x i x represents the variable characteristic of the i-th candidate variable among the candidate variables, which can be the variable characteristic of the first candidate variable; j ψ represents the variable characteristic of the j-th candidate variable among the candidate variables, and can be the variable characteristic of the second candidate variable; ij Represents x i The corresponding first candidate variable and x j The constraint factor for the corresponding second candidate variable.

[0048] Based on the variable information of candidate variables as variable features at the task nodes of the aforementioned tasks to be processed, in the second optional implementation provided in this embodiment, the following operations are performed during the process of calculating the variable relationships between candidate variables based on the variable information of the candidate variables:

[0049] A first feature is calculated based on the variable characteristics and the first transformation characteristics of the first candidate variable among the candidate variables, and a second feature is calculated based on the variable characteristics and the second transformation characteristics of the second candidate variable among the candidate variables;

[0050] Based on the first feature and the second feature, calculate the variable relationship index between the first candidate variable and the second candidate variable.

[0051] Optionally, the first transformation feature and the second transformation feature may be the same or different. The first feature may be a first feature vector, and the second feature may be a second feature vector.

[0052] Specifically, in the process of calculating the first feature based on the variable features and the first transformation features of the first candidate variable, a first product can be calculated as the first intermediate feature based on the variable features and the first transformation features of the first candidate variable, and the first intermediate feature can be transformed (e.g., transposed, converting an n*m matrix into an m*n matrix) to obtain the first feature; in the process of calculating the second feature based on the variable features and the second transformation features of the second candidate variable, a second product can be calculated as the second feature based on the variable features and the second transformation features of the second candidate variable; in the process of calculating the variable relationship index of the first candidate variable and the second candidate variable based on the first feature and the second feature, the product of the first feature and the second feature can be calculated as the variable relationship index of the first candidate variable and the second candidate variable.

[0053] Furthermore, in the process of calculating the variable relationship between the candidate variables based on the variable information, the variable characteristics of the first candidate variable and the variable characteristics of the second candidate variable can also be input into the second relationship calculation algorithm to calculate the variable relationship index and obtain the variable relationship index of the first candidate variable and the second candidate variable; that is, the second optional implementation of the above-mentioned variable relationship calculation can be executed by the second relationship calculation algorithm, which can be an algorithm using the transformer (a neural network) mechanism.

[0054] For example, the algorithm for calculating the second relation is:

[0055] e ij =(W k x i ) T W q x j

[0056] Among them, e ij Represents the relationship between variables; W k Represents the first transformation feature; (W) k x i ) represents the first intermediate feature; (W k x i ) T W represents the first feature obtained by transposing the first intermediate feature; q Represents the second transformation feature; x i x represents the variable characteristic of the i-th candidate variable among the candidate variables. i It can be the first candidate variable; x j x represents the variable characteristic of the j-th candidate variable among the candidate variables. j It can be the second candidate variable.

[0057] In addition, the second optional implementation method for calculating variable relationships described above can also be executed by a third relationship calculation algorithm. The third relationship calculation algorithm can be a metric learning-based method, specifically using the inner-product mechanism of GRCN (Graph-Refined Convolutional Network).

[0058] For example, the algorithm for calculating the third relation is:

[0059] e ij =(W1x i ) T (W1x j )

[0060] Among them, e ij Represents the variable relationship index; W1 represents the first transformation feature and the second transformation feature; (W1x i ) represents the first intermediate feature; (W1x i ) T This represents the first feature obtained by transposing the first intermediate feature; (W1x) j ) represents the second characteristic; x i x represents the variable characteristic of the i-th candidate variable among the candidate variables. i It can be the first candidate variable; x j x represents the variable characteristic of the j-th candidate variable among the candidate variables. j It can be the second candidate variable.

[0061] Based on the variable information of candidate variables as variable features at the task nodes of the aforementioned tasks to be processed, in the third optional implementation provided in this embodiment, during the process of calculating the variable relationship between candidate variables based on the variable information of candidate variables, the variable relationship index between the first candidate variable and the second candidate variable can be calculated according to the variable features and transformation features of the first candidate variable and the second candidate variable. Specifically, the following operations can be performed:

[0062] Based on the variable characteristics of the first and second candidate variables among the candidate variables, the target characteristics are determined;

[0063] A first feature is calculated based on the variable features and transformation features of the first candidate variable, and a second feature is calculated based on the variable features and transformation features of the second candidate variable;

[0064] Based on the target feature, the first feature, and the second feature, calculate the variable relationship index between the first candidate variable and the second candidate variable.

[0065] The transformation features include a transformation matrix used for matrix transformation. Optionally, the variable features of the first candidate variable and the variable features of the second candidate variable have the same feature dimension. For example, if the variable features of the first candidate variable are a-dimensional feature vectors, then the feature dimension of the variable features of the first candidate variable is a.

[0066] Specifically, in the process of determining the target feature based on the variable features of the first candidate variable and the second candidate variable, the feature dimension of the variable feature of the first candidate variable or the variable feature of the second candidate variable can be determined, an intermediate feature can be generated based on the feature dimension, and the intermediate feature can be transformed to obtain the target feature (the transformation process includes transpose processing or matrix transpose processing); in the process of calculating the first feature based on the variable features of the first candidate variable and the transformed feature, the first product of the variable features of the first candidate variable and the transformed feature can be calculated as the first feature; in the process of calculating the second feature based on the variable features of the second candidate variable and the transformed feature, the second product of the variable features of the second candidate variable and the transformed feature can be calculated as the second feature.

[0067] To further improve the convenience of calculating the variable relationship index, an activation function can be introduced. In one optional implementation of this embodiment, the following operation is performed during the process of calculating the variable relationship index of the first candidate variable and the second candidate variable based on the target feature, the first feature, and the second feature:

[0068] Calculate the target value based on the target feature, the first feature, and the second feature;

[0069] The target value is input into the activation function to calculate the index, thereby obtaining the variable relationship index.

[0070] The activation function can be any type of activation function, such as Leaky ReLU.

[0071] Specifically, in the process of calculating the target value based on the target feature, the first feature, and the second feature, the first feature and the second feature can be concatenated to obtain the third feature, and the target feature and the third feature can be multiplied to obtain the target value.

[0072] In addition, the third optional implementation method for calculating variable relationships described above can also be executed by a fourth relationship calculation algorithm. The fourth relationship calculation algorithm can be a network-based algorithm, specifically an algorithm using GAT (Graph Attention Network).

[0073] For example, the algorithm for calculating the fourth relation is:

[0074] e ij =σ(α) T ·[W2x i ||W2x j ])

[0075] Among them, e ij σ represents the variable relationship index; σ represents the activation function, such as Leaky ReLU; α T The target feature is the transpose of the intermediate feature α; x i x represents the variable characteristic of the i-th candidate variable among the candidate variables. i It can be the first candidate variable; x j x represents the variable characteristic of the j-th candidate variable among the candidate variables. j It can be the second candidate variable; W2 represents the transformation feature; (W2x i ) represents the transformation feature and x i The first product, (W2x) j ) represents the transformation feature and x j The second product; [W2x i ||W2x j ] represents (W2x) i ) and (W2x j The third feature is obtained by splicing together the features.

[0076] After obtaining the variable relationships between the candidate variables through the above calculations, the related variables of the candidate variables can also be determined based on these relationships; specifically, the related variables of the candidate variables can be determined in the following ways:

[0077] Among the candidate variables, a second candidate variable whose variable relationship index with the first candidate variable is greater than a preset index threshold is detected;

[0078] The detected second candidate variable is used as the associated variable of the first candidate variable.

[0079] The preset index threshold can be 0.

[0080] Specifically, when there are multiple candidate variables, among the multiple candidate variables, candidate variables whose variable relationship index with each of the multiple candidate variables is greater than a preset index threshold are detected, and the detected candidate variables are used as the associated variables of each candidate variable.

[0081] It should be noted that the first candidate variable mentioned above can be the first candidate variable among every two candidate variables, and the second candidate variable can be the second candidate variable among every two candidate variables.

[0082] Step S206: Based on the correlation information of the correlation variables of the candidate variables and the variable relationship between the candidate variables and the correlation variables, perform information fusion processing to obtain the fusion information of the candidate variables.

[0083] The above-mentioned calculation of the variable relationships between candidate variables based on the variable information of the candidate variables involves, in this step, combining the correlation variable information of the related variables and the variable relationships between the candidate variables and the related variables to perform information fusion processing, thereby obtaining fused information of the candidate variables. The fused information mentioned in this embodiment refers to the fused information generated after fusing the related variable information into the variable information of the candidate variables.

[0084] In practical applications, the variable information of a single candidate variable may not accurately represent the situation of the candidate variable, resulting in low accuracy of the determined target variable. Therefore, to improve the comprehensiveness and accuracy of the variable information of candidate variables, thereby enhancing the effectiveness of the determined target variable and improving the task processing efficiency, this embodiment provides an optional implementation method. In the process of obtaining the fused information of the candidate variable by performing information fusion processing based on the correlation variable information of the candidate variable's related variables and the variable relationship between the candidate variable and the related variables, the following operations are performed:

[0085] Based on the associated variable information and the variable relationship index between the candidate variable and the associated variable, the information to be fused is calculated;

[0086] Based on the information to be fused, calculate the fusion information of the candidate variables.

[0087] The information to be fused can be a feature to be fused, specifically a matrix to be fused.

[0088] Specifically, in the process of calculating the information to be fused based on the associated variable information and the variable relationship index between the candidate variable and the associated variable, the product of the associated variable information of each associated variable and the corresponding variable relationship value between the candidate variable and each associated variable is calculated as each piece of information to be fused; in the process of calculating the fusion information of the candidate variable based on the information to be fused, the fusion information of the candidate variable can be obtained by summing the information to be fused.

[0089] For example, candidate variables include a, b, c, and d. The related variables of candidate variables a, b, c, and d can all include a, b, c, and d. That is, the related variables of each candidate variable in candidate variables a, b, c, and d can include a, b, c, and d. Taking candidate variable a as an example, assuming that the variable features of candidate variables a, b, c, and d are i1, i2, i3, and i4 respectively, and the variable relationship indicators between candidate variable a and related variables a, b, c, and d are m1, m2, m3, and m4 respectively, then the information to be fused is calculated as (i1×m1), (i2×m2), (i3×m3), and (i4×m4). The fusion information of candidate variable a is (i1×m1+i2×m2+i3×m3+i4×m4).

[0090] Furthermore, during the process of performing information fusion processing based on the correlation information of the candidate variables and the variable relationship between the candidate variables and the correlation variables to obtain the fused information of the candidate variables, the following operations can also be performed:

[0091] The variable relationship between the candidate variables and the associated variables is normalized to obtain the normalized variable relationship between the candidate variables and the associated variables;

[0092] Based on the correlation information of the candidate variables and the normalized variable relationship between the candidate variables and the correlation variables, calculate the information to be fused.

[0093] Based on the information to be fused, calculate the fusion information of the candidate variables.

[0094] Specifically, the variable relationship index between the candidate variable and each associated variable can be normalized to obtain the normalized index between the candidate variable and each associated variable; the product of the associated variable feature of each associated variable and the corresponding normalized index can be calculated as each piece of information to be fused; and the fused information of the candidate variable can be obtained by summing the fused information of each piece of information.

[0095] It should be noted that steps S204 to S206 can be repeated multiple times. That is, based on the variable information of the candidate variables, the variable relationship between the candidate variables is calculated. Based on the correlation variable information of the candidate variables and the variable relationship between the candidate variables and the correlation variables, information fusion processing is performed to obtain the fusion information of the candidate variables. The fusion information of the candidate variables is used as the variable information of the candidate variables. The process is then repeated to calculate the variable relationship between the candidate variables based on the variable information of the candidate variables. Based on the correlation variable information of the candidate variables and the variable relationship between the candidate variables and the correlation variables, information fusion processing is performed to obtain the fusion information of the candidate variables. If the fusion information is found to meet the variable determination conditions, step S208 is executed. The variable determination conditions include that the number of fusion processes is greater than a preset threshold.

[0096] Step S208: Based on the fusion information, determine the target variable from the candidate variables, and perform corresponding task processing on the child nodes of the task node based on the target variable.

[0097] The above-mentioned information fusion processing, based on the correlation information of the candidate variables and the variable relationship between the candidate variables and the correlation variables, yields the fusion information of the candidate variables. In this step, the target variable is determined from the candidate variables using the fusion information, so that corresponding task processing is performed based on the target variable at the child nodes of the task node. The child nodes of the task node refer to the nodes at the next level below the task node.

[0098] In specific implementation, to improve the convenience of determining the target variable, in an optional implementation method provided in this embodiment, the following operations are performed during the process of determining the target variable from the candidate variables based on the fusion information:

[0099] The dimensionality-reduced features are obtained by performing feature dimensionality reduction on the fusion features of the candidate variables;

[0100] The reduced-dimensional features are normalized to obtain normalized features, and the target variable is determined from the candidate variables based on the normalized features.

[0101] The normalized features include normalized feature vectors with a preset feature dimension; the preset feature dimension can be 1-dimensional.

[0102] In one optional implementation of this embodiment, during the process of determining the target variable from candidate variables based on the normalized features, the following operations are performed:

[0103] The normalized feature values ​​in the normalized features are sorted.

[0104] The candidate variable corresponding to the normalized feature value with a preset ranking position is determined as the target variable.

[0105] In the specific execution process, in order to improve the flexibility and efficiency of task processing, the task processing method can optionally be implemented by using a variable-determined model. In one optional implementation of this embodiment, the variable-determined model is trained in the following manner:

[0106] A first training sample is determined from the training sample sequence, and the first training sample is merged with the second training sample to obtain a merged sample.

[0107] The merged samples are input into the candidate variable determination model for model training to obtain the variable determination model.

[0108] Specifically, the first training sample can be extracted from the training sample sequence according to the extraction ratio, and the first training sample and the second training sample can be merged to obtain a merged sample. The candidate variable determination model can be trained for a preset number of rounds using the merged sample to obtain the variable determination model.

[0109] The second training sample may be empty or not empty. If the second training sample is empty, the first training sample is the training sample obtained from the first sample extraction. If the second training sample is not empty, the second training sample is the training sample obtained from subsequent sample extractions after the first extraction.

[0110] For example, first, 30% of the training samples are extracted from the training sample sequence and used as the merged sample. The candidate variable determination model is trained for m1 rounds using the merged sample to obtain the first intermediate model. Then, 20% of the training samples are extracted from the remaining training sample sequence and merged with the 30% training samples to obtain the merged sample. The first intermediate model is trained for m2 rounds to obtain the second intermediate model. Another 20% of the training samples are extracted and merged with the previous 30% and 20% training samples to obtain the merged sample. The second intermediate model is trained for m3 rounds using the merged sample to obtain the third intermediate model. Finally, 30% of the training samples are extracted and merged with the previous 30%, 20%, and 20% training samples to obtain the merged sample (which is exactly the entire training sample sequence). The third intermediate model is trained for m4 rounds using the merged sample to obtain the variable determination model.

[0111] In one optional implementation of this embodiment, the training sample sequence is obtained in the following manner:

[0112] The candidate variables are determined by training the model to be trained based on the training data;

[0113] The candidate variables are used to determine the model to calculate the training loss corresponding to each training sample in the training data, and the training samples are sorted according to the training loss to obtain the training sample sequence.

[0114] The training data includes variable information of candidate variables on task nodes of one or more training tasks; the training task refers to the task to be processed for model training.

[0115] Specifically, the training samples can be sorted in order of increasing or decreasing training loss to obtain a training sample sequence.

[0116] It should be noted that the process of sorting the training samples according to the training loss to obtain the training sample sequence can be executed by a Difficulty Measurer; the process of determining the first training sample in the training sample sequence and merging the first training sample with the second training sample to obtain the merged sample can be executed by a Training Scheduler.

[0117] In one optional implementation of this embodiment, during the corresponding task processing at the child nodes of the task node based on the target variable, the following operations are performed:

[0118] Based on the task processing result of the target variable at the task node, determine the child nodes of the task node;

[0119] Candidate variables on the child node are determined based on the task processing results of the child node.

[0120] For example, among candidate variables x1 and x2, the target variable is determined to be x1. Based on the task processing result of the target variable x1, i.e. relaxation result 1.2, the child nodes of the task node are determined to be x1≤1 and x1≥2. The task processing results of the first subtask p1 and the second subtask p2 corresponding to the child nodes x1≤1 and x1≥2 are determined, i.e. relaxation results. The candidate variables on the child nodes are determined according to the task processing results of the child nodes.

[0121] It should be added that steps S202 to S208 above can be replaced by: obtaining variable information of multiple candidate variables on the task node of the task to be processed; calculating the variable relationship between every two candidate variables based on the variable information of the multiple candidate variables; performing information fusion processing according to the variable relationship between each candidate variable and its associated variable and the associated variable information of the associated variable of each candidate variable to obtain the fusion information of each candidate variable; determining the target variable among the multiple candidate variables based on the fusion information, so as to perform corresponding task processing on the child node of the task node based on the target variable;

[0122] Alternatively, it can be replaced by: obtaining the variable features of multiple candidate variables on the task node of the task to be processed; calculating the variable relationship between every two candidate variables based on the variable features of the multiple candidate variables; performing feature fusion processing based on the variable relationship between each candidate variable and its associated variable and the associated variable features of each candidate variable to obtain the fused features of each candidate variable; determining the target variable among the multiple candidate variables based on the fused features of each candidate variable, so as to perform corresponding task processing on the child nodes of the task node based on the target variable.

[0123] Optionally, the associated variables of each candidate variable among the plurality of candidate variables include the plurality of candidate variables, that is, the associated variables of each candidate variable can be multiple candidate variables, that is, the associated variables of each candidate variable include each candidate variable and the remaining candidate variables among the plurality of candidate variables other than the candidate variable itself.

[0124] In the process of calculating the variable relationship between every two candidate variables based on the variable information of multiple candidate variables, the constraint factors of every two candidate variables can be determined, and the variable relationship index of every two candidate variables can be calculated based on the variable information and constraint factors of every two candidate variables. In the process of calculating the variable relationship index of every two candidate variables based on the variable information and constraint factors of every two candidate variables, the intermediate variable relationship index of every two candidate variables can be calculated based on the variable information and constraint factors of every two candidate variables, and the intermediate variable relationship index of every two candidate variables can be normalized to obtain the variable relationship index of every two candidate variables. In the process of calculating the variable relationship between every two candidate variables based on the variable information of multiple candidate variables, a first feature can be calculated based on the variable characteristics and first transformation characteristics of the first candidate variable in every two candidate variables, and a second feature can be calculated based on the variable characteristics and second transformation characteristics of the second candidate variable in every two candidate variables. Based on the first feature and the second feature, a variable relationship index for every two candidate variables can be calculated. Alternatively, a target feature can be determined based on the variable characteristics of every two candidate variables, and a first feature can be calculated based on the variable characteristics and transformation characteristics of the first candidate variable in every two candidate variables, and a second feature can be calculated based on the variable characteristics and transformation characteristics of the second candidate variable in every two candidate variables. Based on the target feature, the first feature, and the second feature, a variable relationship index for every two candidate variables can be calculated.

[0125] In the process of obtaining fused information for each candidate variable by performing information fusion processing based on the variable relationships between each candidate variable and its associated variables and the associated variable information of the associated variables, the fused information of each candidate variable can be calculated based on the variable relationships between each candidate variable and its associated variables and the associated variable information of the associated variables. It should be noted that the specific implementation method involved here is similar to that described above, and will not be repeated here.

[0126] In summary, the one or more task processing methods provided in this embodiment first obtain the variable information of candidate variables at the task nodes of the task to be processed; secondly, determine the constraint factors of the first and second candidate variables among the candidate variables; and calculate the variable relationship index of the first and second candidate variables based on the variable information of the first and second candidate variables and the constraint factors; or, calculate a first feature based on the variable characteristics and a first transformation feature of the first candidate variable among the candidate variables, and calculate a second feature based on the variable characteristics and a second transformation feature of the second candidate variable among the candidate variables, and calculate the variable relationship index of the first and second candidate variables according to the first and second features; or, determine the target feature based on the variable characteristics of the first and second candidate variables among the candidate variables. The first feature is calculated based on the variable characteristics and transformation characteristics of the first candidate variable, and the second feature is calculated based on the variable characteristics and transformation characteristics of the second candidate variable. Based on the target feature, the first feature, and the second feature, the variable relationship index between the first and second candidate variables is calculated. On this basis, the information to be fused is calculated based on the correlation information of the candidate variables and the variable relationship index between the candidate variables and the correlation variables. The fusion information of the candidate variables is then calculated based on the information to be fused. Finally, the target variable is determined from the candidate variables based on the fusion information, and corresponding task processing is performed at the child nodes of the task node based on the target variable. Thus, by introducing the variable relationship between candidate variables, the effectiveness and comprehensiveness of the fusion information of the candidate variables are improved, the efficiency of determining the target variable is improved, and the task processing efficiency of the task to be processed is improved.

[0127] The following description uses the application of a task processing method provided in this embodiment in a container scheduling scenario as an example to further illustrate the task processing method provided in this embodiment. (See also...) Figure 3 The task processing method applied to container scheduling scenarios includes the following steps.

[0128] Step S302: Obtain the variable characteristics of multiple candidate variables on the task node of the container scheduling task.

[0129] Step S304: Determine the constraint factors for every two candidate variables among multiple candidate variables, and calculate the variable relationship index for every two candidate variables based on the variable characteristics and constraint factors of each pair of candidate variables.

[0130] Step S306: Calculate the features to be fused based on the variable relationship index between each candidate variable and its associated variable, and the associated variable characteristics of each candidate variable's associated variable.

[0131] Step S308: Calculate the fusion features of each candidate variable based on the features to be fused.

[0132] Step S310: Perform feature dimensionality reduction on the fused features to obtain dimensionality-reduced features, and perform feature normalization on the dimensionality-reduced features to obtain normalized features.

[0133] Step S312: Sort the normalized feature values ​​in the normalized features, and determine the candidate variable corresponding to the normalized feature value with the preset sorting position among multiple candidate variables as the target variable, so as to perform corresponding scheduling task processing on the child nodes of the task node based on the target variable.

[0134] It should be added that step S304 above can be replaced by calculating a first feature based on the variable features and first transformation features of the first candidate variable among every two candidate variables in a plurality of candidate variables, and calculating a second feature based on the variable features and second transformation features of the second candidate variable among every two candidate variables; and calculating a variable relationship index for every two candidate variables based on the first feature and the second feature.

[0135] It can also be replaced by determining the target feature based on the variable characteristics of every two candidate variables among multiple candidate variables; calculating the first feature based on the variable characteristics and transformation characteristics of the first candidate variable among every two candidate variables, and calculating the second feature based on the variable characteristics and transformation characteristics of the second candidate variable among every two candidate variables; and calculating the variable relationship index of every two candidate variables based on the target feature, the first feature, and the second feature.

[0136] This specification provides an embodiment of a task processing device as follows:

[0137] In the above embodiments, a task processing method is provided, and correspondingly, a task processing device is also provided, which will be described below with reference to the accompanying drawings.

[0138] Reference Figure 4 The diagram illustrates an embodiment of a task processing device provided in this embodiment.

[0139] Since the apparatus embodiments correspond to the method embodiments, the descriptions are relatively simple. For relevant parts, please refer to the corresponding descriptions of the method embodiments provided above. The apparatus embodiments described below are merely illustrative.

[0140] This embodiment provides a task processing device, including:

[0141] The information acquisition module 402 is configured to acquire variable information of candidate variables on the task node of the task to be processed;

[0142] The relationship calculation module 404 is configured to calculate the variable relationship between the candidate variables based on the variable information of the candidate variables;

[0143] The fusion processing module 406 is configured to perform information fusion processing based on the association variable information of the association variables of the candidate variables and the variable relationship between the candidate variables and the association variables to obtain the fusion information of the candidate variables.

[0144] The variable determination module 408 is configured to determine a target variable from the candidate variables based on the fusion information, so as to perform corresponding task processing on the child nodes of the task node based on the target variable.

[0145] This specification provides an example of a task processing device as follows:

[0146] Corresponding to the task processing method described above, based on the same technical concept, one or more embodiments of this specification also provide a task processing device for executing the task processing method provided above. Figure 5 This is a schematic diagram of the structure of a task processing device provided for one or more embodiments of this specification.

[0147] This embodiment provides a task processing device, including:

[0148] like Figure 5 As shown, task processing devices can vary considerably due to differences in configuration or performance. They may include one or more processors 501 and memory 502, with memory 502 storing one or more application programs or data. Memory 502 can be temporary or persistent storage. The application programs stored in memory 502 may include one or more modules (not shown), each module including a series of computer-executable instructions from the task processing device. Furthermore, processor 501 may be configured to communicate with memory 502, executing the series of computer-executable instructions stored in memory 502 on the task processing device. The task processing device may also include one or more power supplies 503, one or more wired or wireless network interfaces 504, one or more input / output interfaces 505, one or more keyboards 506, etc.

[0149] In one specific embodiment, the task processing device includes a memory and one or more programs, wherein the one or more programs are stored in the memory, and the one or more programs may include one or more modules, and each module may include a series of computer-executable instructions for the task processing device, and is configured to be executed by one or more processors. The one or more programs include computer-executable instructions for performing the following:

[0150] Obtain variable information of candidate variables on the task node of the task to be processed;

[0151] Based on the variable information of the candidate variables, calculate the variable relationships between the candidate variables;

[0152] Based on the correlation information of the candidate variable and the variable relationship between the candidate variable and the correlation variable, information fusion processing is performed to obtain the fusion information of the candidate variable;

[0153] Based on the fusion information, a target variable is determined from the candidate variables, and corresponding task processing is performed on the child nodes of the task node based on the target variable.

[0154] This specification provides an example of a storage medium as follows:

[0155] Corresponding to the task processing method described above, based on the same technical concept, one or more embodiments of this specification also provide a storage medium.

[0156] The storage medium provided in this embodiment is used to store computer-executable instructions, which, when executed by a processor, implement the following process:

[0157] Obtain variable information of candidate variables on the task node of the task to be processed;

[0158] Based on the variable information of the candidate variables, calculate the variable relationships between the candidate variables;

[0159] Based on the correlation information of the candidate variable and the variable relationship between the candidate variable and the correlation variable, information fusion processing is performed to obtain the fusion information of the candidate variable;

[0160] Based on the fusion information, a target variable is determined from the candidate variables, and corresponding task processing is performed on the child nodes of the task node based on the target variable.

[0161] It should be noted that the embodiments of a storage medium described in this specification and the embodiments of a task processing method described in this specification are based on the same inventive concept. Therefore, the specific implementation of this embodiment can be referred to the implementation of the corresponding method described above, and the repeated parts will not be described again.

[0162] The various embodiments in this specification are described in a progressive manner. For the same or similar parts between the various embodiments, please refer to each other. Each embodiment focuses on describing the differences from other embodiments. For example, the device embodiment, equipment embodiment, and storage medium embodiment are all similar to the method embodiment, so the description is relatively simple. For reading the relevant content of the device embodiment, equipment embodiment, and storage medium embodiment, please refer to the description of the method embodiment.

[0163] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0164] In the 1930s, improvements to a technology could be clearly distinguished as either hardware improvements (e.g., improvements to the circuit structure of diodes, transistors, switches, etc.) or software improvements (improvements to the methodology). However, with technological advancements, many improvements to the methodology today can be considered direct improvements to the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved methodology into the hardware circuit. Therefore, it cannot be said that an improvement to the methodology cannot be implemented using a hardware physical module. For example, a Programmable Logic Device (PLD) (e.g., a Field Programmable Gate Array (FPGA)) is such an integrated circuit whose logic function is determined by the user programming the device. Designers can program a digital system themselves to "integrate" it onto a PLD, without needing chip manufacturers to design and manufacture dedicated integrated circuit chips. Furthermore, nowadays, instead of manually manufacturing integrated circuit chips, this programming is mostly implemented using "logic compiler" software. Similar to the software compiler used in program development, the original code before compilation must be written in a specific programming language, called a Hardware Description Language (HDL). There are many HDLs, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, and RHDL (Ruby Hardware Description Language). Currently, the most commonly used are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should understand that by simply performing some logic programming on the method flow using one of these hardware description languages ​​and programming it into an integrated circuit, the hardware circuit implementing the logical method flow can be easily obtained.

[0165] The controller can be implemented in any suitable manner. For example, it can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicon Labs C8051F320. A memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art will also recognize that, in addition to implementing the controller in purely computer-readable program code form, the same functionality can be achieved by logically programming the method steps to make the controller take the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the means included therein for implementing various functions can also be considered as structures within the hardware component. Alternatively, the means for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.

[0166] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0167] For ease of description, the above apparatus is described by dividing it into various functional units. Of course, when implementing the embodiments of this specification, the functions of each unit can be implemented in one or more software and / or hardware.

[0168] Those skilled in the art will understand that one or more embodiments of this specification can be provided as a method, system, or computer program product. Therefore, one or more embodiments of this specification may take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this specification may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0169] This specification is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this specification. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable task processing device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable task processing device, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0170] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable task processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0171] These computer program instructions may also be loaded onto a computer or other programmable task processing device to cause a series of operational steps to be performed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable device for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0172] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0173] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0174] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0175] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0176] One or more embodiments of this specification can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a particular task or implement a particular abstract data type. One or more embodiments of this specification can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0177] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0178] The above description is merely an embodiment of this document and is not intended to limit the scope of this document. Various modifications and variations can be made to this document by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this document should be included within the scope of the claims of this document.

Claims

1. A task processing method applied to a server, the method comprising: Obtain variable information for multiple candidate variables on the task node of the container scheduling task sent by the task processing platform; The multiple candidate variables are obtained by the task processing platform after the task node performs task processing on the container scheduling task; the task node corresponds to the container scheduling task or a subtask of the container scheduling task; the multiple candidate variables are determined based on variables whose relaxation result of the relaxation task corresponding to the task node is a non-integer and whose variable attribute is an integer. Based on the variable information of the multiple candidate variables, a variable relationship index representing the closeness of the relationship between each pair of candidate variables is calculated. Among the multiple candidate variables, candidate variables whose variable relationship index with each candidate variable is greater than a preset index threshold are detected, and the detected candidate variables are used as the associated variables of each candidate variable. The correlation variable information of the correlation variable is fused into the variable information of each candidate variable to obtain the fused information of each candidate variable; Based on the fusion information, a target variable is determined from the multiple candidate variables. The target variable is then sent to the task processing platform. The task processing platform determines the child nodes of the task node based on the task processing result of the task node using the target variable. Based on the task processing result of the subtask corresponding to the child node, candidate variables on the child node are determined. The variable information of the determined candidate variables is then sent to the server until the task processing result of the container scheduling task is obtained.

2. The task processing method according to claim 1, wherein determining the target variable based on the fusion information from the plurality of candidate variables includes: The dimensionality-reduced features are obtained by performing feature dimensionality reduction on the fusion features of each candidate variable; The reduced-dimensional features are normalized to obtain normalized features, and the target variable is determined from the plurality of candidate variables based on the normalized features.

3. The task processing method according to claim 2, wherein determining the target variable based on the normalized feature from the plurality of candidate variables includes: The normalized feature values ​​in the normalized features are sorted. The candidate variable corresponding to the normalized feature value with a preset ranking position is determined as the target variable.

4. The task processing method according to claim 1, wherein the task processing method determines the model execution through variables; in, The variable determination model is trained in the following manner: A first training sample is determined from the training sample sequence, and the first training sample is merged with the second training sample to obtain a merged sample. The merged samples are input into the candidate variable determination model for model training to obtain the variable determination model.

5. The task processing method according to claim 4, wherein the training sample sequence is obtained in the following manner: The candidate variables are determined by training the model to be trained based on the training data; The candidate variables are used to determine the model to calculate the training loss corresponding to each training sample in the training data, and the training samples are sorted according to the training loss to obtain the training sample sequence.

6. The task processing method according to claim 1, wherein calculating the variable relationship index representing the closeness of the relationship between every two candidate variables based on the variable information of the plurality of candidate variables includes: Determine the constraint factors for every two candidate variables among the plurality of candidate variables; Based on the variable information of each pair of candidate variables and the constraint factors, a variable relationship index representing the closeness of the relationship between each pair of candidate variables is calculated.

7. The task processing method according to claim 1, wherein calculating the variable relationship index representing the closeness of the relationship between every two candidate variables based on the variable information of the plurality of candidate variables includes: A first feature is calculated based on the variable characteristics and first transformation characteristics of the first candidate variable in every pair of candidate variables, and a second feature is calculated based on the variable characteristics and second transformation characteristics of the second candidate variable in every pair of candidate variables; Based on the first feature and the second feature, calculate the variable relationship index representing the degree of closeness of the relationship between each pair of candidate variables.

8. The task processing method according to claim 1, wherein calculating the variable relationship index representing the closeness of the relationship between every two candidate variables based on the variable information of the plurality of candidate variables includes: Based on the variable characteristics of the first and second candidate variables in every pair of candidate variables, the target characteristics are determined; A first feature is calculated based on the variable features and transformation features of the first candidate variable, and a second feature is calculated based on the variable features and transformation features of the second candidate variable; Based on the target feature, the first feature, and the second feature, calculate the variable relationship index representing the closeness of the relationship between each pair of candidate variables.

9. The task processing method according to claim 8, wherein calculating the variable relationship index representing the closeness of the relationship between each pair of candidate variables based on the target feature, the first feature, and the second feature includes: Calculate the target value based on the target feature, the first feature, and the second feature; The target value is input into the activation function to calculate the index, thereby obtaining the variable relationship index.

10. A task processing apparatus, operating on a server, the apparatus comprising: The information acquisition module is configured to acquire variable information of multiple candidate variables on the task nodes of the container scheduling task sent by the task processing platform; The multiple candidate variables are obtained by the task processing platform after the task node performs task processing on the container scheduling task; the task node corresponds to the container scheduling task or a subtask of the container scheduling task; the multiple candidate variables are determined based on variables whose relaxation result of the relaxation task corresponding to the task node is a non-integer and whose variable attribute is an integer. The relationship calculation module is configured to calculate a variable relationship index representing the closeness of the relationship between each pair of candidate variables based on the variable information of the multiple candidate variables, detect candidate variables whose variable relationship index with each candidate variable in the multiple candidate variables is greater than a preset index threshold, and use the detected candidate variables as the associated variables of each candidate variable. The fusion processing module is configured to fuse the association variable information of the associated variables into the variable information of each candidate variable to obtain the fusion information of each candidate variable; The variable determination module is configured to determine a target variable from multiple candidate variables based on the fusion information, send the target variable to the task processing platform, and use the task processing platform to determine the child nodes of the task node based on the task processing result of the task node based on the target variable, determine the candidate variables on the child node based on the task processing result of the subtask corresponding to the child node, and send the variable information of the determined candidate variables to the server until the task processing result of the container scheduling task is obtained.

11. A task processing device, comprising: processor; And, a memory configured to store computer-executable instructions, which, when executed, cause the processor to: The variable information of multiple candidate variables on the task node of the container scheduling task sent by the task processing platform is obtained; the multiple candidate variables are obtained by the task processing platform after the task node performs task processing on the container scheduling task; the task node corresponds to the container scheduling task or a subtask of the container scheduling task; the multiple candidate variables are determined based on the fact that the relaxation result of the relaxation task corresponding to the task node is a non-integer variable and the variable attribute is an integer. Based on the variable information of the multiple candidate variables, a variable relationship index representing the closeness of the relationship between each pair of candidate variables is calculated. Among the multiple candidate variables, candidate variables whose variable relationship index with each candidate variable is greater than a preset index threshold are detected, and the detected candidate variables are used as the associated variables of each candidate variable. The correlation variable information of the correlation variable is fused into the variable information of each candidate variable to obtain the fused information of each candidate variable; Based on the fusion information, a target variable is determined from the multiple candidate variables. The target variable is then sent to the task processing platform. The task processing platform determines the child nodes of the task node based on the task processing result of the task node using the target variable. Based on the task processing result of the subtask corresponding to the child node, candidate variables on the child node are determined. The variable information of the determined candidate variables is then sent to the server until the task processing result of the container scheduling task is obtained.

12. A storage medium for storing computer-executable instructions, which, when executed by a processor, perform the following process: The variable information of multiple candidate variables on the task node of the container scheduling task sent by the task processing platform is obtained; the multiple candidate variables are obtained by the task processing platform after the task node performs task processing on the container scheduling task; the task node corresponds to the container scheduling task or a subtask of the container scheduling task; the multiple candidate variables are determined based on the fact that the relaxation result of the relaxation task corresponding to the task node is a non-integer variable and the variable attribute is an integer. Based on the variable information of the multiple candidate variables, a variable relationship index representing the closeness of the relationship between each pair of candidate variables is calculated. Among the multiple candidate variables, candidate variables whose variable relationship index with each candidate variable is greater than a preset index threshold are detected, and the detected candidate variables are used as the associated variables of each candidate variable. The correlation variable information of the correlation variable is fused into the variable information of each candidate variable to obtain the fused information of each candidate variable; Based on the fusion information, a target variable is determined from the multiple candidate variables. The target variable is then sent to the task processing platform. The task processing platform determines the child nodes of the task node based on the task processing result of the task node using the target variable. Based on the task processing result of the subtask corresponding to the child node, candidate variables on the child node are determined. The variable information of the determined candidate variables is then sent to the server until the task processing result of the container scheduling task is obtained.

Images