Intelligent decision-making method based on experience matching
By adopting an intelligent decision-making method based on empirical matching in the decision-making network, the problems of low efficiency and slow convergence speed of traditional decision-making networks are solved, and a more efficient and accurate decision-making process is achieved.
Patent Information
- Application Number
- CN202411756686.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-03
- Publication Date
- 2025-04-08
AI Technical Summary
When facing new decision-making needs, traditional decision-making networks are less efficient, the network converges slowly, making it difficult to quickly adapt to changes.
Using an intelligent decision-making method based on empirical matching, by modeling decision requirements into conditional parameter sets and environmental parameter sets, matching searches are performed in the decision experience library, matching degree is calculated, and decision network initialization and training are performed based on matching degree.
Through efficient joint decision-making requirements and decision-making experience database, the network training round is shortened, decision-making efficiency and accuracy are improved, and the convergence of the network is enhanced.
Smart Images
Figure CN119783723A_ABST
Abstract
Description
Technical Field
[0001] The embodiments of the present invention relate to the field of computer network technology, and in particular to an intelligent decision-making method based on experience matching. Background Art
[0002] With the rapid development of artificial intelligence technology, decision networks have important application value in both military and civilian applications. They can help people make decisions faster and more accurately, and improve work efficiency and decision quality. However, traditional decision networks mainly rely on given data sets for learning and decision-making. Once a new decision demand enters the network, the decision output is obtained through parameter training of a deep neural network based on the "black box" mode. This mode has problems such as low decision network efficiency and slow network convergence.
[0003] Therefore, it is necessary to improve one or more problems existing in the above-mentioned related technical solutions.
[0004] It should be noted that this section is intended to provide background or context for the technical solutions of the present invention stated in the claims. The description herein is not admitted to be prior art by inclusion in this section. Summary of the invention
[0005] The purpose of the present invention is to provide an intelligent decision-making method based on experience matching, thereby solving one or more problems caused by the limitations and defects of related technologies at least to a certain extent.
[0006] The present invention provides an intelligent decision-making method based on experience matching, comprising the following steps:
[0007] Modeling decision requirements as sets of conditional parameters and environmental parameters;
[0008] Perform matching search on the condition parameter set and the environment parameter set in a decision-making experience library;
[0009] Calculating the matching degree between the condition parameter set and the environment parameter set and each experience parameter set in the decision experience library respectively;
[0010] Comparing the maximum matching degree among the matching degrees with a threshold value, and the decision network outputting decision experience according to the comparison result;
[0011] Among them, if the comparison result is completely matched, the decision experience of the decision experience library corresponding to the maximum matching degree is selected as the output decision experience. If the comparison result is not completely matched, the decision network is initialized, and the decision output is performed after the decision network is trained.
[0012] In the present invention, the overlapping similarity algorithm is used in the matching degree, and the expression is:
[0013]
[0014] In the formula, Overlap Similarity is the matching degree; P(D B |D A ),P(D A |D B ) is the conditional probability.
[0015] In the present invention, the threshold values include a maximum threshold value and a minimum threshold value, a maximum matching degree greater than or equal to the maximum threshold value indicates a complete match, and a maximum matching degree less than the maximum threshold value indicates an incomplete match, wherein the incomplete match includes a maximum matching degree less than the maximum threshold value and greater than or equal to the minimum threshold value, and the maximum matching degree is less than the minimum threshold value.
[0016] In the present invention, the threshold value includes a maximum threshold value and a minimum threshold value, the minimum threshold value is 0.2, and the maximum threshold value is 0.8.
[0017] In the present invention, when the comparison result is completely matched, the maximum matching degree is ≥ 0.8, and the decision network selects the decision experience of the decision experience library corresponding to the maximum matching degree as the output decision experience.
[0018] In the present invention, when the comparison result is not completely matched, when 0.2≤maximum matching degree<0.8, the parameters corresponding to the experience in the decision experience library are extracted to initialize the decision network, and the decision network is trained according to the parameters corresponding to the experience in the decision experience library to output the decision.
[0019] In the present invention, when the comparison result is not completely matched, when the maximum matching degree is less than 0.2, the decision network is randomly initialized, and the decision output is performed after the decision network is trained.
[0020] In the present invention, the storage of decision-making experience in the experience database includes:
[0021] Store the given condition parameter set, environment parameter set and output decision;
[0022] Store the parameter set of each experience in the decision experience library.
[0023] In the present invention, all decision outputs of the decision network are stored in the decision experience library as new experience.
[0024] The technical solution provided by the present invention may include the following beneficial effects:
[0025] In the present invention, through the above method, the decision demand and the decision experience library are efficiently combined through the constructed decision architecture, which greatly shortens the network training rounds. An intelligent decision method based on experience matching is constructed by combining the decision demand and the decision experience library, and the matching degree estimation algorithm is used to calculate the corresponding matching degree between the decision demand and the decision experience library. Based on the estimated matching degree, the decision network related parameters stored under this decision experience are extracted in the decision experience library, and the extracted network parameters are placed in the decision network as prior information, so as to accelerate the convergence of the network and improve the decision efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] The accompanying drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the specification are used to explain the principles of the present disclosure. Obviously, the accompanying drawings described below are only some embodiments of the present disclosure, and for ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without creative work.
[0027] Figure 1 A schematic diagram showing the steps of an intelligent decision-making method based on experience matching in an exemplary embodiment of the present invention.
[0028] Figure 2 A schematic flow chart of an intelligent decision-making method based on experience matching in an exemplary embodiment of the present invention is shown. DETAILED DESCRIPTION
[0029] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that the disclosure will be more comprehensive and complete and to fully convey the concepts of the example embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
[0030] In addition, the accompanying drawings are only schematic illustrations of the present disclosure and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor devices and / or microcontroller devices.
[0031] This example implementation provides an intelligent decision-making method based on experience matching, such as Figure 1 As shown, the following steps may be included:
[0032] S101: Model decision requirements as conditional parameter sets and environmental parameter sets;
[0033] S102: performing a matching search on the condition parameter set and the environment parameter set in a decision-making experience library;
[0034] S103: respectively calculating the matching degree of the condition parameter set and the environment parameter set with each experience parameter set in the decision experience library;
[0035] S104: comparing the maximum matching degree among the matching degrees with a threshold value, and the decision network outputs decision experience according to the comparison result;
[0036] Among them, if the comparison result is completely matched, the decision experience of the decision experience library corresponding to the maximum matching degree is selected as the output decision experience. If the comparison result is not completely matched, the decision network is initialized, and the decision output is performed after the decision network is trained.
[0037] This application constructs a decision-making architecture that efficiently combines decision requirements with the decision experience library, greatly shortening the network training rounds. By combining decision requirements with the decision experience library, an intelligent decision-making method based on experience matching is constructed, and a matching degree estimation algorithm is used to calculate the corresponding matching degree between the decision requirements and the decision experience library. Based on the estimated matching degree, the decision network related parameters stored under this decision experience are extracted from the decision experience library, and the extracted network parameters are placed in the decision network as prior information, thereby accelerating the convergence of the network and improving decision efficiency and accuracy.
[0038] Below, as Figure 2 As shown, each step of the above method in this exemplary embodiment is described in more detail.
[0039] Step S101: Modeling decision requirements as a condition parameter set and an environment parameter set.
[0040] Specifically, decision requirements are a set of decision input conditions and decision environment parameters.
[0041] The experience database is used to store decision-making experience: Decision-making experience refers to the optimal decision output that has been verified in the early stage under given input conditions and environment. The storage of decision-making experience is divided into two forms: (a) storing the given condition parameter set, environmental parameter set and output decision; (b) storing the parameter set of each experience in the decision experience database.
[0042] S102: performing a matching search on the condition parameter set and the environment parameter set in a decision experience library.
[0043] Specifically, for the establishment of the decision experience library, each decision requirement i has its corresponding descriptor Di = {ai1, ai2, .., ai4}, such as the descriptor is composed of 4 sub-elements to describe the decision requirement. The descriptor is usually in the form of a vector, which describes the feature information of the decision requirement in a certain artificially designed way. The descriptor is designed according to the principle that "features with similar features should have similar descriptors". Therefore, when calculating the matching degree between the decision requirement and the candidate experience in the experience library, it is only necessary to use statistical methods to calculate the two vector descriptors. When the matching degree of the descriptors of the two decision requirements is similar in statistical measurement, it can be considered that they have similar experience.
[0044] S103: Calculate the matching degree between the condition parameter set and the environment parameter set and each experience parameter set in the decision experience library respectively.
[0045] Specifically, in step S103, the matching degree between the decision requirement and the plurality of alternative decision experiences is calculated respectively, and the matching degree is calculated using an overlapping similarity algorithm, as shown in formula (1):
[0046]
[0047] In the formula, Overlap Similarity is the matching degree; P(D B |D A ),P(D A |D B ) is the conditional probability.
[0048] The matching degree estimation method is used to calculate the matching degree between the input decision demand and the decision experience in the experience library. The larger the matching degree value, the more similar the decision demand is to the decision experience in the experience library. Conversely, the smaller the similarity is. This application uses the overlapping similarity algorithm to calculate the matching degree. As an extensible algorithm, the matching degree algorithm includes but is not limited to the overlapping similarity algorithm.
[0049] S104: comparing the maximum matching degree among the matching degrees with a threshold value, and the decision network outputs decision experience according to the comparison result;
[0050] Among them, if the comparison result is completely matched, the decision experience of the decision experience library corresponding to the maximum matching degree is selected as the output decision experience. If the comparison result is not completely matched, the decision network is initialized, and the decision output is performed after the decision network is trained.
[0051] Specifically, the threshold values include a maximum threshold value and a minimum threshold value, a maximum match greater than or equal to the maximum threshold value indicates a complete match, and a maximum match less than the maximum threshold value indicates an incomplete match, wherein the incomplete match includes a maximum match less than the maximum threshold value and greater than or equal to the minimum threshold value, and the maximum match less than the minimum threshold value.
[0052] When the matching degree between decision demand and output decision experience is less than 0.2, the decision demand is an inexperienced decision demand;
[0053] When the matching degree between the decision demand and the output decision experience is ≥0.2 and <0.8, the decision demand is a similar experience decision demand;
[0054] When the matching degree between the decision requirement and the output decision experience is ≥0.8, the decision requirement is the existing experience decision requirement.
[0055] The above-mentioned threshold of the matching degree includes but is not limited to the above-mentioned specific numerical value, and those skilled in the art may adjust the above-mentioned threshold of the matching degree according to different decision-making requirements.
[0056] For decision requirements that are classified as inexperienced, it means that the decision requirements are significantly different from the experience stored in the decision experience library. In this case, the decision network parameters are randomly initialized, and the decision network is trained for more rounds before making a decision output.
[0057] When the decision requirement is a similar experience decision requirement, the network parameters stored in the output decision experience are extracted as prior information of the decision network, the decision network is initialized by the network parameters stored in the output decision experience, and the decision network outputs the decision after network training.
[0058] For decision requirements with similar classification, it means that there are decision experiences in the decision library that are similar to the decision requirements. Then, the network parameters stored under this decision experience are extracted from the decision experience library. These extracted parameters are used as prior knowledge to initialize the nodes involved in the network. The convergence of decision network training can be improved by utilizing prior knowledge. In this case, the network can quickly output decisions with fewer rounds of network training.
[0059] Because the decision experience in the decision library is similar to the decision demand, extracting the network parameters stored under similar decision experience in the decision experience library as prior information reduces the number of network training times of the decision network, thereby reducing the time it takes for the decision network to output decisions.
[0060] When the decision requirement is an existing experience decision requirement, the network parameters stored in the output decision experience are extracted and placed into the decision network as prior information, and the decision network directly outputs the decision.
[0061] For decision requirements whose classification is existing experience, it means that there are existing decision experiences with exactly the same decision requirements in the decision library. In this case, the network parameters stored under the decision experiences with exactly the same similarity are extracted from the decision experience library, and the extracted network parameters are placed in the decision network as prior information. Since all training parameters are known and valid, there is no need for network training in the decision network, and the network can output decisions in real time.
[0062] For decision requirements with no experience and decision requirements with similar experience, after the decision network is trained and outputs decisions, the decision requirements, decision outputs and network parameters stored by the decision network are stored in the decision experience library as new decision experiences. The decision experiences in the decision experience library are updated in real time, and the number of decision experiences in the decision experience library is increased, thereby improving the matching degree between the decision requirements and the decision experiences in the decision experience library, thereby improving the decision efficiency and accuracy.
[0063] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example" or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification.
[0064] Those skilled in the art will readily appreciate other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses or adaptations of the present invention, which follow the general principles of the present invention and include common knowledge or customary techniques in the art that are not disclosed by the present invention. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present invention are indicated by the appended claims.
Claims
1. An intelligent decision-making method based on experience matching, characterized in that: Follow these steps to implement: Modeling decision requirements as sets of conditional parameters and environmental parameters; Perform matching search on the condition parameter set and the environment parameter set in a decision-making experience library; Calculating the matching degree between the condition parameter set and the environment parameter set and each experience parameter set in the decision experience library respectively; Comparing the maximum matching degree among the matching degrees with a threshold value, and the decision network outputting decision experience according to the comparison result; Among them, if the comparison result is completely matched, the decision experience of the decision experience library corresponding to the maximum matching degree is selected as the output decision experience. If the comparison result is not completely matched, the decision network is initialized, and the decision output is performed after the decision network is trained.
2. The intelligent decision-making method based on experience matching according to claim 1 is characterized in that: The matching degree adopts the overlapping similarity algorithm, and the expression is: In the formula, Overlap Similarity is the matching degree; P(D B |D A ),P(D A |D B ) is the conditional probability.
3. The intelligent decision-making method based on experience matching according to claim 1 is characterized in that: The threshold values include a maximum threshold value and a minimum threshold value. A complete match is defined as the maximum match degree being greater than or equal to the maximum threshold value, and an incomplete match is defined as the maximum match degree being less than the maximum threshold value. The incomplete match includes a maximum match degree being less than the maximum threshold value and greater than or equal to the minimum threshold value, and the maximum match degree being less than the minimum threshold value.
4. The intelligent decision-making method based on experience matching according to claim 3 is characterized in that: The minimum threshold value is 0.2, and the maximum threshold value is 0.
8.
5. The intelligent decision-making method based on experience matching according to claim 4 is characterized in that: When the comparison result is completely matched, the maximum matching degree is ≥ 0.8, and the decision network selects the decision experience of the decision experience library corresponding to the maximum matching degree as the output decision experience.
6. The intelligent decision-making method based on experience matching according to claim 5 is characterized in that: When the comparison result is not completely matched, when 0.2≤maximum matching degree<0.8, the parameters of corresponding experience in the decision experience library are extracted to initialize the decision network, and the decision network is trained according to the parameters of corresponding experience extracted from the decision experience library to output the decision.
7. The intelligent decision-making method based on experience matching according to claim 6 is characterized in that: When the comparison result is not completely matched, when the maximum matching degree is less than 0.2, the decision network is randomly initialized, and the decision output is performed after the decision network is trained.
8. The intelligent decision-making method based on experience matching according to claim 1 is characterized in that: The storage of decision-making experience in the experience database includes: Store the given condition parameter set, environment parameter set and output decision; Store the parameter set of each experience in the decision experience library.
9. The intelligent decision-making method based on experience matching according to claim 1, characterized in that: All decision outputs are stored in the decision experience database as new experience.