Ecological niche-based algorithm resource competition cooperation evolution model construction method

Abstract

The invention discloses an ecological niche-based algorithm resource competition cooperation evolution model construction method. The method comprises the steps of obtaining N performance indexes of I algorithm resources deployed on a target platform; establishing an ecological niche fitness expression according to the performance indexes; according to the ecological niche fitness expression, obtaining the ecological niche fitness of the I algorithm resources under the required environmental factors; establishing an ecological niche similarity expression; according to the ecological niche similarity expression, calculating the ecological niche similarity between any two different algorithm resources; establishing an initial survival value expression according to the ecological niche fitness; obtaining initial survival values of the I algorithm resources according to the initial survival value expression; determining a competition influence coefficient according to the ecological niche similarity; and establishing an algorithm resource competition cooperation evolution model according to the initial survival value and the competition influence coefficient of the algorithm resource. According to the method, an algorithm resource competition and cooperation evolution model is constructed to predict the development trend of algorithm resources in a competition and cooperation environment; and effective technical support and reference opinions can be provided for algorithm resource configuration of a third-party platform.

Description

Technical Field

[0001] The present invention relates to a method for constructing an algorithm resource competition and cooperation evolution model based on ecological niche, and belongs to the technical field of data processing. Background Art

[0002] With the rapid development of big data and artificial intelligence technologies, third-party cloud platforms for data processing and algorithm deployment have been widely used in fields such as recommendation systems and intelligent manufacturing. In natural ecosystems, under conditions of limited resources, populations dependent on the same resource compete with each other, following the pattern described by the Lotka-Volterra model. The competitive and cooperative relationships between agents in resource ecosystems closely resemble the interactions between species in nature. In data processing and algorithm deployment environments, many algorithms provide the same or similar functions, resulting in fierce competition for resource allocation and prioritization, leading to a continuous survival of the fittest. At the same time, algorithms often need to collaborate, especially when tackling complex or cross-domain tasks, where multiple algorithms must work together to complete the task. However, existing algorithm resource management methods primarily rely on static performance metrics, which cannot comprehensively assess the adaptability of algorithms to different business scenarios or environmental factors. Furthermore, most integration methods focus solely on cooperative gains or pure competition, without establishing a unified model to simultaneously characterize the interaction between the two. This makes it difficult to provide a sound scientific basis for long-term platform planning.

[0003] In view of this, the present invention is proposed. Summary of the Invention

[0004] The present invention provides a method for constructing an algorithm resource competition and cooperation evolution model based on ecological niche, which is used to construct an algorithm resource competition and cooperation evolution model to predict the development trend of algorithm resources in a competition and cooperation environment; it can provide effective technical support and reference opinions for the algorithm resource configuration of third-party platforms.

[0005] The technical solution of the present invention is:

[0006] According to a first aspect of the present invention, a method for constructing an algorithm resource competition and cooperative evolution model based on ecological niche is provided, comprising:

[0007] S1. Obtain N performance indicators of I algorithm resources deployed on the target platform;

[0008] S2. Establish an ecological niche fitness expression based on the performance index; obtain the ecological niche fitness of I algorithm resources under the required environmental factors based on the ecological niche fitness expression;

[0009] S3. Establish an ecological niche similarity expression; based on the ecological niche similarity expression, calculate the ecological niche similarity between any two different algorithm resources;

[0010] S4. Establish an initial survival value expression based on the niche fitness; obtain the initial survival value of I algorithm resources based on the initial survival value expression;

[0011] S5. Determine the competition impact coefficient based on the niche similarity; establish an algorithm resource competition and cooperation evolution model based on the initial survival value and competition impact coefficient of the algorithm resources.

[0012] Furthermore, the niche fitness expression is specifically:

[0013]

[0014] in, is the niche fitness of the i-th algorithm resource under the k-th environmental factor, i = 1, 2, ... I; N is the number of performance indicators, is the jth performance index value of the i-th algorithm resource under the k-th environmental factor, is the weight of the i-th algorithm resource under the j-th performance indicator under the k-th environmental factor.

[0015] Furthermore, the niche similarity expression is specifically:

[0016]

[0017] Where, O AB is the niche similarity between algorithm resources A and algorithm resources B; P Ak is the usage ratio of algorithm resource A on the kth environmental factor, P Bk is the usage ratio of algorithm resource B on the kth environmental factor; n represents the number of environmental factors in which algorithm resources A and algorithm resources B participate together.

[0018] Furthermore, the initial survival value expression is specifically:

[0019]

[0020] Among them, S i (0) is the initial survival value of the i-th algorithm resource, is the niche fitness of the i-th algorithm resource under the l-th environmental factor; w l is the weight of the lth environmental factor, and m represents the total number of environmental factors involved in the ith algorithm resource.

[0021] Furthermore, the algorithm resource competition and cooperation evolution model has a specific mathematical expression as follows:

[0022]

[0023] Among them, Si (t+1) represents the survival value of the i-th algorithm resource at time t+1, b ij represents the competition impact coefficient between the i-th algorithm resource and the j-th algorithm resource; c ij K represents the cooperation effect coefficient between the i-th algorithm resource and the j-th algorithm resource; i , K j Indicates the upper limit of the survival value of the i-th algorithm resource and the j-th algorithm resource in the non-competitive state; r i represents the intrinsic growth rate of the i-th algorithm resource.

[0024] According to a second aspect of the present invention, there is provided a device for constructing an algorithm resource competition and cooperative evolution model based on ecological niche, comprising:

[0025] The first acquisition module is used to obtain N performance indicators of I algorithm resources deployed on the target platform;

[0026] The second acquisition module is used to establish an ecological niche fitness expression based on the performance index; and is used to obtain the ecological niche fitness of I algorithm resources under the required environmental factors based on the ecological niche fitness expression;

[0027] The calculation module is used to establish an ecological niche similarity expression; and is used to calculate the ecological niche similarity between any two different algorithm resources based on the ecological niche similarity expression;

[0028] The third acquisition module is used to establish an initial survival value expression based on the ecological niche fitness; and is used to obtain the initial survival value of I algorithm resources based on the initial survival value expression;

[0029] A module is established to determine the competition impact coefficient based on the niche similarity; it is used to establish an algorithm resource competition and cooperation evolution model based on the initial survival value and competition impact coefficient of the algorithm resources.

[0030] According to a third aspect of the present invention, an evolutionary analysis device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of the method for constructing an algorithmic resource competition and cooperation evolution model based on an ecological niche as described in any one of the above items are implemented.

[0031] The beneficial effects of the present invention are: the present invention describes the adaptability of the algorithm resources to the environment from the perspective of the algorithm resources themselves through niche fitness, while the niche similarity is analyzed from the perspective of the relationship between algorithm resources, so as to establish an initial survival value expression; further, an algorithm resource competition and cooperation evolution model is established to fill the gap in the existing research on the evolution law of the competition and cooperation relationship of multiple algorithm resources. Furthermore, through evolutionary simulation analysis, a scientific basis is provided for strategy formulation and long-term planning, which can improve the overall performance of the platform, resource utilization and the scalability and adaptability of the system, and effectively overcome the shortcomings of the existing technology in terms of single evaluation dimension and separation of competition and cooperation. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 Construct a flow chart for the resource competition and cooperative evolution model of the algorithm of the present invention;

[0033] Figure 2 This is a partial overlapping relationship diagram for the competition and cooperation evolution analysis of the present invention;

[0034] Figure 3 Adjacency relationship graph for competition and cooperation evolution analysis of the present invention;

[0035] Figure 4 This is a diagram of the internal sourcing relationship for the competition and cooperation evolution analysis of the present invention;

[0036] Figure 5 This is a separation relationship diagram for the competition and cooperation evolution analysis of the present invention;

[0037] Figure 6 This is a development trend diagram of the algorithm resources of the present invention in a competitive and cooperative environment. DETAILED DESCRIPTION

[0038] To make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. It should be noted that, in the absence of conflict, the embodiments in this application and the features in the embodiments can be combined with each other in any way.

[0039] Example 1: Figures 1-6 As shown, a method for constructing an algorithm resource competition and cooperation evolution model based on ecological niche can be executed by an evolution analysis device, that is, one or more processors in the evolution analysis device execute the following steps S1-S5.

[0040] S1. Obtain N performance indicators of I algorithm resources deployed on a target platform; the performance indicators include subjective performance indicators and objective performance indicators; illustratively, the subjective performance indicators include user satisfaction, and the objective performance indicators include at least one of accuracy and response time;

[0041] S2. Establish an ecological niche fitness expression based on the performance index; obtain the ecological niche fitness of I algorithm resources under the required environmental factors based on the ecological niche fitness expression;

[0042] The niche fitness expression is specifically:

[0043]

[0044] in, is the niche fitness of the i-th algorithm resource under the k-th environmental factor, i = 1, 2, ... I; N is the number of performance indicators, is the jth performance index value of the i-th algorithm resource under the k-th environmental factor, is the weight of the i-th algorithm resource under the j-th performance indicator under the k-th environmental factor.

[0045] S3. Establish an ecological niche similarity expression; based on the ecological niche similarity expression, calculate the ecological niche similarity between any two different algorithm resources;

[0046] The niche similarity expression is specifically:

[0047]

[0048] Where, O AB is the niche similarity between algorithm resources A and algorithm resources B; P Ak is the usage ratio of algorithm resource A on the kth environmental factor, P Bk is the usage ratio of algorithm resource B on the kth environmental factor; n represents the number of environmental factors in which algorithm resources A and algorithm resources B participate together.

[0049] In the above, environmental factors can be different tasks in the same application scenario. For example, if there are 10 prediction tasks involving 10 different objects in the prediction scenario, the number of environmental factors is 10. The usage ratio of the algorithm resource on the kth environmental factor can be the calling frequency of the algorithm resource on the factor, or the ecological fitness ratio of the algorithm resource in the factor (for example, in the case of 2 environmental factors and 3 algorithm resources, the ratio can be obtained). The ecological fitness ratio of algorithm resource 1 in environmental factor 1 is ).

[0050] S4. Establish an initial survival value expression based on the niche fitness; obtain the initial survival value of I algorithm resources based on the initial survival value expression;

[0051] The initial survival value expression is specifically:

[0052]

[0053] Among them, S i (0) is the initial survival value of the ith algorithm resource (in the embodiment of the present invention, t=0 is used as the initial time; if other times are used as the initial time, then S i (0) 0 in the match is replaced by the corresponding time), is the niche fitness of the i-th algorithm resource under the l-th environmental factor; w l is the weight of the lth environmental factor, and m represents the total number of environmental factors involved in the ith algorithm resource. (For example, in the prediction scenario, algorithm resource 1 participates in the prediction tasks of 3 different objects, then the number of environmental factors m is 3, and the weight of each environmental factor is further set to 1 / 3. The survival value of algorithm resource 1 is ).

[0054] S5. Determine the competition impact coefficient based on the niche similarity; establish an algorithm resource competition and cooperation evolution model based on the initial survival value and competition impact coefficient of the algorithm resource; and obtain the development trend of different algorithm resources in the competition and cooperation environment based on the evolution model;

[0055] The algorithm resource competition and cooperation evolution model is specifically expressed as follows:

[0056]

[0057] Among them, b ij represents the competition impact coefficient between the i-th algorithm resource and the j-th algorithm resource; c ij K represents the cooperation effect coefficient between the i-th algorithm resource and the j-th algorithm resource; i , K j Indicates the upper limit of the survival value of the i-th algorithm resource and the j-th algorithm resource in the non-competitive state; r i represents the intrinsic growth rate of the i-th algorithm resource.

[0058] The following further gives the method for determining the above parameters:

[0059] Intrinsic growth rate r i It's just about the algorithm itself. Where α can be set in the interval [0,1] according to the actual data. (For example, in the prediction scenario, the first algorithm resource participates in the prediction task of 3 different objects. Then for the first algorithm resource, its intrinsic growth rate is ).

[0060] Cooperation coefficient c ij It can be approximated by the ratio of environmental factors that two algorithm resources jointly participate in (for example, in a prediction scenario involving 10 prediction tasks of different objects, where the first algorithm resource and the second algorithm resource jointly participate in 4 tasks, the cooperation coefficient is ).

[0061] According to the general rules of resource ecology, the impact of competition on algorithm resources is related to their capabilities. Algorithmic resources with strong capabilities are less affected, while algorithm resources with weak capabilities are more affected. Therefore, a formula is introduced to calculate the competition impact coefficient between two different algorithm resources. The expression is as follows:

[0062]

[0063] Among them, β is the adjustment coefficient, which can be set in the interval [0,1] according to the actual data; ij is the niche similarity between the ith algorithm resource and the ith algorithm resource.

[0064] To more clearly illustrate possible situations that may arise between algorithm resources in a competition and cooperation environment, this embodiment selects two algorithm resources for specific analysis.

[0065] According to the mathematical expression of the evolution model, the evolution model expressions of algorithm resource 1 and algorithm resource 2 can be obtained:

[0066]

[0067] In order to further analyze the development and change of the competition and cooperation relationship between individuals, let the change rate of survival value ΔS1 = 0, ΔS2 = 0, and obtain the curve of the change rate of algorithm resource survival value ( Figure 2-Figure 5 In the figure, the solid line represents algorithm resource 1 and the dotted line represents algorithm resource 2), and the competition and cooperation parameters (b ij 、c ij ), we can get four different relationships between the two algorithm resources:

[0068] Setting b ij =0.4, c ij =0, and obtain Figure 2The curve of the survival value change rate of algorithm resources with partially overlapping relationships. Point E is the intersection of the two algorithms. The system is in an unstable equilibrium. In region ①, Algorithm 1 gradually gains the upper hand and Algorithm 2 is eliminated. In region ②, Algorithm 2 becomes dominant and Algorithm 1 is eliminated. In region ③, both algorithms can coexist temporarily, but as the system develops, they will eventually tend to be dominated by one of them.

[0069] Setting b ij =0.2, c ij = 0.1, and we get Figure 3 The curve of the change rate of algorithm resource survival value of the adjacency relationship shows that the competition near point E is relatively light. In area ①, algorithm 1 is dominant, and over time, algorithm 2 is gradually eliminated. In area ②, algorithm 2 has an advantage and algorithm 1 is weakened. Area ③ indicates that the two algorithms have almost no intersection and are independent of each other, but one will eventually dominate in its independent field.

[0070] Setting b ij =0.6, c ij =0, and obtain Figure 4 The curve of the survival value change rate of algorithm resources in the internal subcontracting relationship shows that Algorithm 1 is completely dominant at point E. The overlap of the system causes Algorithm 2 to be continuously weakened. In area ①, Algorithm 1 completely dominates and Algorithm 2 cannot survive. In area ②, Algorithm 2 has a short survival in a small area, but its demise is almost inevitable. In area ③, the possibility of coexistence is extremely low, and the system development eventually tends to be completely controlled by Algorithm 1.

[0071] Setting b ij =0, c ij =0, and obtain Figure 5 The curve of the change rate of algorithm resource survival value in the separation relationship shows that there is almost no competition at point E. The resource areas of the two algorithms are completely independent. In area ①, algorithm 1 develops stably in its independent field, and algorithm 2 is almost unaffected. In area ②, algorithm 2 also exists independently and is almost not disturbed by the competition of algorithm 1. In area ③, the two algorithms are completely separated and do not interfere with each other. The system presents a stable low-competition state.

[0072] Example 2:

[0073] Taking "prediction algorithm resources - parts demand forecasting" as the research business, the method of the present invention is explained as follows:

[0074] 1. Taking the prediction task of "exhaust pipe" accessories as a research case (i.e., the environmental factor is set to 1), the historical accessories sales, vehicle sales, accessories failure rate, and accessories demand data of the accessories "exhaust pipe" are obtained to construct a data set; Table 1 below is a selection of accessories sales, vehicle sales, accessories failure rate, and accessories demand data. The data set is divided into a training set and a test set, and the obtained prediction algorithm resources are trained and tested (the prediction algorithm resources include S1, S2, and S3, which are used to predict the accessories demand based on the accessories sales, vehicle sales, and accessories failure rate of the accessories "exhaust pipe"; in an embodiment of the present invention, S1, S2, and S3 correspond to the LSTM model, the CNN-LSTM model, and the CNN model, respectively). Through testing, the accuracy and response time performance index results of the prediction algorithm resources are obtained as shown in Table 2, and the user satisfaction index is also given in Table 2. Furthermore, the weights of the three prediction algorithm resources are set to 1 / 3.

[0075] Table 1

[0076] Accessory sales Accessory sales Accessory sales Accessory sales 127 113 0.014345214 128 129 115 0.020979621 133 113 126 0.012661477 126 129 115 0.020559799 133 113 126 0.012670085 127 126 112 0.012027116 127 127 113 0.014345214 128 113 126 0.012610825 127 110 115 0.011457407 110 126 111 0.011540269 126 126 112 0.011958646 127

[0077] Table 2

[0078] Algorithm Resources Prediction accuracy Algorithm resource response time User satisfaction S1 0.75 0.8 0.73 S2 0.95 0.88 0.96 S3 0.93 0.85 0.91

[0079] It should be noted that for other research services, if the target platform (e.g., a third-party cloud platform) has access to trained prediction algorithm resources, multiple performance indicators of these trained prediction algorithm resources can be directly invoked. Furthermore, this invention can be expanded to other types of algorithm resources, such as search algorithm resources, for use in auto parts searches, repair plan searches, claim record searches, and service provider searches.

[0080] 2. Based on the established niche fitness expression, calculate the niche fitness of the three prediction algorithm resources, that is, obtain

[0081] 3. Based on the established niche similarity expression, calculate the niche similarity of the three prediction algorithm resources, namely O 12 , O 13 , O 23 ;

[0082] 4. Based on the established initial survival value expression, calculate the initial survival values ​​of the three prediction algorithm resources;

[0083] 5. Based on the mathematical expression of the evolution model, numerical simulation experiments are conducted on the three prediction algorithm resources to observe the development trends of the three resources in a competitive and cooperative environment, such as Figure 6As shown in the figure, the survival value of the S1 algorithm resources gradually decreases and stabilizes over time: At the beginning of the simulation, the survival value of the S1 algorithm resources rises briefly, but quickly stabilizes and remains at a low level. This shows that although the S1 algorithm resources survive the competition with other algorithms, their growth rate cannot match that of the S2 and S3 algorithm resources due to their inferior fitness and call frequency. Ultimately, under the influence of competition and cooperation, the survival value remains at a low level.

[0084] Stability of S2 and S3 Algorithm Resources: The survival values ​​of S2 and S3 Algorithm Resources are close to 1, indicating that they have an advantage in competition under these environmental factors and may also have gained some positive effects through cooperation. These factors have led to their survival values ​​growing rapidly and reaching a high and stable level.

[0085] The simulation results show that the survival value of the S1 algorithm resource is lower than that of the S2 and S3 algorithm resources and tends to be stable. This indicates that the S1 algorithm resource is at a disadvantage in competition with the S2 and S3 algorithm resources, but still has a certain survival space. Furthermore, the proposed method can effectively describe and evaluate the competitive and cooperative relationships between individuals in the algorithm resource ecosystem, providing guidance for the management of data resources and platforms, and has important theoretical and practical value.

[0086] Example 3: A device for constructing an algorithmic resource competition and cooperative evolution model based on ecological niche, comprising:

[0087] The first acquisition module is used to obtain N performance indicators of I algorithm resources deployed on the target platform;

[0088] The second acquisition module is used to establish an ecological niche fitness expression based on the performance index; and is used to obtain the ecological niche fitness of I algorithm resources under the required environmental factors based on the ecological niche fitness expression;

[0089] The calculation module is used to establish an ecological niche similarity expression; and is used to calculate the ecological niche similarity between any two different algorithm resources based on the ecological niche similarity expression;

[0090] The third acquisition module is used to establish an initial survival value expression based on the ecological niche fitness; and is used to obtain the initial survival value of I algorithm resources based on the initial survival value expression;

[0091] A module is established to determine the competition impact coefficient based on the niche similarity; it is used to establish an algorithm resource competition and cooperation evolution model based on the initial survival value and competition impact coefficient of the algorithm resources.

[0092] For parts of the modules not described in detail above, please refer to the relevant descriptions of other embodiments.

[0093] The specific embodiments of the present invention are described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Various changes can be made within the knowledge of ordinary technicians in this field without departing from the scope of the present invention.

Claims

1. A method for constructing an algorithm resource competition and cooperative evolution model based on ecological niche, characterized by: include: S1. Obtain N performance indicators of I algorithm resources deployed on the target platform; S2. Establish an ecological niche fitness expression based on the performance index; obtain the ecological niche fitness of I algorithm resources under the required environmental factors based on the ecological niche fitness expression; S3. Establish an ecological niche similarity expression; based on the ecological niche similarity expression, calculate the ecological niche similarity between any two different algorithm resources; S4. Establish an initial survival value expression based on the niche fitness; obtain the initial survival value of I algorithm resources based on the initial survival value expression; S5. Determine the competition impact coefficient based on the niche similarity; establish an algorithm resource competition and cooperation evolution model based on the initial survival value and competition impact coefficient of the algorithm resources.

2. The method for constructing an algorithm resource competition and cooperative evolution model based on ecological niche according to claim 1 is characterized in that: The niche fitness expression is specifically: in, is the niche fitness of the i-th algorithm resource under the k-th environmental factor, i = 1, 2, ... I; N is the number of performance indicators, is the jth performance index value of the i-th algorithm resource under the k-th environmental factor, is the weight of the i-th algorithm resource under the j-th performance indicator under the k-th environmental factor.

3. The method for constructing a resource competition and cooperative evolution model based on ecological niche according to claim 1 is characterized in that: The niche similarity expression is specifically: Where, O AB is the niche similarity between algorithm resources A and algorithm resources B; P Ak is the usage ratio of algorithm resource A on the kth environmental factor, P Bk is the usage ratio of algorithm resource B on the kth environmental factor; n represents the number of environmental factors in which algorithm resources A and algorithm resources B participate together.

4. The method for constructing a resource competition and cooperative evolution model based on ecological niche according to claim 1 is characterized in that: The initial survival value expression is specifically: Among them, S i (0) is the initial survival value of the i-th algorithm resource, is the niche fitness of the i-th algorithm resource under the l-th environmental factor; w l is the weight of the lth environmental factor, and m represents the total number of environmental factors involved in the ith algorithm resource.

5. The method for constructing an algorithm resource competition and cooperative evolution model based on ecological niche according to claim 1 is characterized in that: The algorithm resource competition and cooperation evolution model is specifically expressed as follows: Among them, S i (t+1) represents the survival value of the i-th algorithm resource at time t+1, b ij represents the competition impact coefficient between the i-th algorithm resource and the j-th algorithm resource; c ij K represents the cooperation effect coefficient between the i-th algorithm resource and the j-th algorithm resource; i , K j Indicates the upper limit of the survival value of the i-th algorithm resource and the j-th algorithm resource in the non-competitive state; r i represents the intrinsic growth rate of the i-th algorithm resource.

6. A device for constructing a resource competition and cooperative evolution model based on ecological niche algorithm, characterized in that: include: The first acquisition module is used to obtain N performance indicators of I algorithm resources deployed on the target platform; The second acquisition module is used to establish an ecological niche fitness expression based on the performance index; and is used to obtain the ecological niche fitness of I algorithm resources under the required environmental factors based on the ecological niche fitness expression; The calculation module is used to establish an ecological niche similarity expression; and is used to calculate the ecological niche similarity between any two different algorithm resources based on the ecological niche similarity expression; The third acquisition module is used to establish an initial survival value expression based on the ecological niche fitness; and is used to obtain the initial survival value of I algorithm resources based on the initial survival value expression; A module is established to determine the competition impact coefficient based on the niche similarity; and to establish an algorithm resource competition and cooperation evolution model based on the initial survival value and competition impact coefficient of the algorithm resource.

7. An evolutionary analysis device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the method for constructing an algorithm resource competition and cooperative evolution model based on ecological niche as described in any one of claims 1 to 5 are implemented.

Images