A thinking learning method and system based on AI convergent evolution

Abstract

The present invention discloses a thinking learning method and system based on AI convergent evolution, including inputting the historical designs of the current designer and the historical designs of other designers of the same type as the current designer, extracting design features in the historical designs, and generating multiple first behavior samples; limiting the behavior logic of the first behavior samples based on the process tree; expanding the first behavior samples based on the behavior logic to generate multiple second behavior samples; performing convergent evolution on the first behavior samples and the second behavior samples, and outputting a final design sample. The present invention generates multiple behavior samples by performing behavioral learning, logical restrictions, and behavioral expansion on the historical works of the designer and other designers of the same type, thereby generating a larger number of richer samples, and then performing AI convergent evolution on the generated behavior samples, thereby enhancing the iterative nature of AI design technology, and ultimately outputting diversified sample results to provide to designers for reference, thereby stimulating the designers' creativity.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a thinking learning method and system based on AI convergent evolution. Background Art

[0002] Currently, AI is used in painting or voice to liberate productivity, but the use of AI is increasingly moving towards industrial assembly lines, making humans lazier. AI design replaces part of human work, causing people to lose their creativity, breakthrough power, and perception ability, and the core AI design technology lacks iterativeness. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a thinking learning method and system based on AI convergent evolution to enhance the iterative nature of AI design technology and thus stimulate human creativity.

[0004] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0005] A thinking learning method based on AI convergent evolution, comprising the following steps:

[0006] S1. Input the historical designs of the current designer and the historical designs of other designers of the same type as the current designer, extract design features from the historical designs, and generate a plurality of first behavior samples;

[0007] S2. limiting the behavior logic of the first behavior sample based on the process tree;

[0008] S3. Expand the first behavior sample based on the behavior logic to generate multiple second behavior samples;

[0009] S4. Perform convergent evolution on the first behavior sample and the second behavior sample, and output a final design sample.

[0010] In order to solve the above technical problems, another technical solution adopted by the present invention is:

[0011] A thinking learning system based on AI convergent evolution, including:

[0012] A behavior learning module is configured to input historical designs of a current designer and historical designs of other designers of the same type as the current designer, extract design features from the historical designs, and generate a plurality of first behavior samples;

[0013] a grammar learning module, configured to restrict the behavior logic of the first behavior sample based on a process tree;

[0014] a behavior extension module, configured to extend the first behavior sample based on the behavior logic to generate a plurality of second behavior samples;

[0015] The convergent evolution module is used to perform convergent evolution on the first behavior sample and the second behavior sample, and output a final design sample.

[0016] The beneficial effects of the present invention are as follows: the present invention provides a thinking learning method and system based on AI convergent evolution, which generates multiple behavior samples by performing behavioral learning, logical restrictions and behavioral expansion on the historical works of designers and other designers of the same type, thereby generating more and richer samples, and then performing AI convergent evolution on the generated behavior samples to enhance the iterative nature of AI design technology, and finally output diversified sample results to provide to designers for reference, thereby stimulating the designers' creativity. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a flow chart of a thinking learning method based on AI convergent evolution according to an embodiment of the present invention;

[0018] Figure 2 This is a structural diagram of a thinking learning system based on AI convergent evolution.

[0019] Description of labels:

[0020] 1. A thinking learning system based on AI convergent evolution; 2. Behavior learning module; 3. Grammar learning module; 4. Behavior expansion module; 5. Logic analysis module; 6. Convergent evolution module. DETAILED DESCRIPTION

[0021] To illustrate the technical content, achieved objectives and effects of the present invention in detail, the following description is given in conjunction with the embodiments and accompanying drawings.

[0022] Please refer to Figure 1 , a thinking learning method based on AI convergent evolution, including the steps of:

[0023] S1. Input the historical designs of the current designer and the historical designs of other designers of the same type as the current designer, extract design features from the historical designs, and generate a plurality of first behavior samples;

[0024] S2. limiting the behavior logic of the first behavior sample based on the process tree;

[0025] S3. Expand the first behavior sample based on the behavior logic to generate multiple second behavior samples;

[0026] S4. Perform convergent evolution on the first behavior sample and the second behavior sample, and output a final design sample.

[0027] From the above description, it can be seen that the beneficial effect of the present invention is that multiple behavior samples are generated by performing behavioral learning, logical restrictions, and behavioral expansion on the historical works of the designer and other designers of the same type, thereby generating more and richer samples, and then performing AI convergent evolution on the generated behavior samples to enhance the iterative nature of AI design technology, and finally output diversified sample results to provide to designers for reference, thereby stimulating the designer's creativity.

[0028] Furthermore, between step S3 and step S4, the following steps are further included:

[0029] S34: Expand the first behavior sample based on the behavior logic to generate multiple second behavior samples, and output the first behavior sample and the second behavior sample by permutation and combination.

[0030]

[0031]

[0032] Wherein, A is the permutation formula, C is the combination formula, n is the number of samples in the second behavior, and m is the number of samples in the first behavior.

[0033] It can be seen from the above description that by arranging and combining the first behavior sample and the second behavior sample, the subsequent designed scenarios are enriched.

[0034] Furthermore, the step S4 further includes the following steps:

[0035] S40. Establish a convergent evolution function, as shown in formula (3):

[0036] N = L × (1 + d%) k (3);

[0037] Among them, N is the iterative result of the current branch evolution, k is the number of iterations, d is the degree of sample deviation, and L is the level, which includes the original evolution layer, the same evolution layer, and the differential evolution layer.

[0038] From the above description, it can be seen that by establishing a convergent evolution function, the original behavioral data can be convergently evolved based on the hierarchy to create differential data and realize multi-level data.

[0039] Furthermore, the step S4 is specifically as follows:

[0040] S41, sequentially substituting the original data, similar data, and differential data into the original evolution layer, the similar evolution layer, and the differential evolution layer in the L branch of the convergent evolution function for hierarchical evolution, to obtain an original evolution result, a similar evolution result, and a differential evolution result, respectively;

[0041] S42: Perform convergent evolution on the original evolution result, the similar evolution result, and the differential evolution result, and output the final design sample.

[0042] Furthermore, the original data is behaviors in the historical designs of the current designer included in the first behavior sample and the second behavior sample;

[0043] The similar data are behaviors in the historical designs of the other designers included in the first behavior sample and the second behavior sample;

[0044] The difference data is behaviors that are different between the historical designs of the current designer and the other designers, included in the first behavior sample and the second behavior sample.

[0045] From the above description, we can see that different original behavioral data - original data, similar data and differential data - are sequentially substituted into the L branch of the convergent evolution function to perform convergent evolution at their own level. As the level changes, AI will perform differential evolution on the results of each convergent evolution and the original behavioral data to create differential data, ultimately achieving multi-level data to stimulate designers' more diversified creativity.

[0046] Please refer to Figure 2 , a thinking learning system based on AI convergent evolution, including:

[0047] A behavior learning module is configured to input historical designs of a current designer and historical designs of other designers of the same type as the current designer, extract design features from the historical designs, and generate a plurality of first behavior samples;

[0048] a grammar learning module, configured to restrict the behavior logic of the first behavior sample based on a process tree;

[0049] a behavior extension module, configured to extend the first behavior sample based on the behavior logic to generate a plurality of second behavior samples;

[0050] The convergent evolution module is used to perform convergent evolution on the first behavior sample and the second behavior sample, and output a final design sample.

[0051] From the above description, it can be seen that the beneficial effects of the present invention are: based on the same technical concept, in conjunction with the above-mentioned thinking learning method based on AI convergent evolution, a thinking learning system based on AI convergent evolution is provided, which generates multiple behavior samples by performing behavioral learning, logical restrictions and behavioral expansion on the historical works of designers and other designers of the same type, thereby generating more and richer samples, and then performing AI convergent evolution on the generated behavior samples to enhance the iterative nature of AI design technology, and finally output diversified sample results to provide to designers for reference, thereby stimulating the designers' creativity.

[0052] Furthermore, a logic parsing module is included, which is used to expand the first behavior sample based on the behavior logic, generate multiple second behavior samples, and arrange and combine the first behavior sample and the second behavior sample to output:

[0053]

[0054]

[0055] Wherein, A is the permutation formula, C is the combination formula, n is the number of samples in the second behavior, and m is the number of samples in the first behavior.

[0056] It can be seen from the above description that by arranging and combining the first behavior sample and the second behavior sample, the subsequent designed scenarios are enriched.

[0057] Furthermore, the convergent evolution module is also used to:

[0058] Convergent evolution function is established as follows:

[0059] N = L × (1 + d%) k (3);

[0060] Among them, N is the iterative result of the current branch evolution, k is the number of iterations, d is the degree of sample deviation, and L is the level, which includes the original evolution layer, the same evolution layer, and the differential evolution layer.

[0061] From the above description, it can be seen that by establishing a convergent evolution function, the original behavioral data can be convergently evolved based on the hierarchy to create differential data and realize multi-level data.

[0062] Furthermore, the convergent evolution module is specifically used to:

[0063] Substituting the original data, similar data and differential data into the original evolution layer, the similar evolution layer and the differential evolution layer in the L branch of the convergent evolution function in turn for hierarchical evolution, and obtaining the original evolution result, the similar evolution result and the differential evolution result respectively;

[0064] The original evolution result, the similar evolution result and the differential evolution result are subjected to convergent evolution, and the final design sample is output.

[0065] Furthermore, the original data is behaviors in the historical designs of the current designer included in the first behavior sample and the second behavior sample;

[0066] The similar data are behaviors in the historical designs of the other designers included in the first behavior sample and the second behavior sample;

[0067] The difference data is behaviors that are different between the historical designs of the current designer and the other designers, included in the first behavior sample and the second behavior sample.

[0068] From the above description, we can see that different original behavioral data - original data, similar data and differential data - are sequentially substituted into the L branch of the convergent evolution function to perform convergent evolution at their own level. As the level changes, AI will perform differential evolution on the results of each convergent evolution and the original behavioral data to create differential data, ultimately achieving multi-level data to stimulate designers' more diversified creativity.

[0069] The present invention provides a thinking learning method and system based on AI convergent evolution, which is applicable to scenarios such as game design and painting design that require designers to have the skills to design characters and scenes. The following is a detailed description with reference to the following embodiments.

[0070] Please refer to Figure 1 , embodiment 1 of the present invention is:

[0071] A thinking learning method based on AI convergent evolution, such as Figure 1 As shown, the steps include:

[0072] S1. Input the historical designs of the current designer and the historical designs of other designers of the same type as the current designer, extract the design features that are manifested by the solidified ideas in the historical designs, and generate multiple first behavior samples. That is, conduct behavioral learning on the historical works of the current designer and other designers, seeking common ground while reserving differences.

[0073] For example, the current designer designs a cat, while other designers of the same type have also designed multiple cats. The current designer's cat can climb trees, while other designers' cats can walk. In this case, behavioral learning can be performed to extract the behaviors of a cat that can climb trees and walk. At the same time, the current designer's cat is black, while other designers' cats are orange, white, gray, etc. In this case, behavioral learning can be performed to extract colorful cats.

[0074] S2. Restrict the behavior logic of the first behavior sample based on the process tree to make the extracted behavior conform to the logical design.

[0075] For example, let's say you need to design a cat running. How do you express the cat running? You'll definitely need a reference, or you can draw some elements on the cat's feet or back to represent it (such as wind wheels, wings, etc.), or a vehicle (such as a bicycle, motorcycle, etc.). Based on the process tree, you can restrict the logic of the above behaviors. For example, in real life, a cat can't ride a bicycle or a motorcycle, but it can run or jump high.

[0076] S3. Expand the first behavior sample based on the behavior logic to generate multiple second behavior samples. The expansion of the behavior also needs to be based on the behavior logic so that the expanded behavior also carries the logic design.

[0077] Let's take cats as an example. For example, if a designer needs a cat that looks very fast, the first behavior sample includes elements such as wind, fire, thunder, and lightning. Through behavior expansion, transportation tools such as bicycles and motorcycles are added to the cat's behavior. The expansion can get a cat sitting on a bicycle or motorcycle and being carried forward by humans. In the process of moving forward, the cat drives the reference objects around it (trees, houses, etc.) backward, which is reflected in the form of wind, fire, thunder, and lightning.

[0078] S4. Perform convergent evolution on the first behavior sample and the second behavior sample, and output the final design sample.

[0079] For example, in game design, designers often design skill scenarios for characters. To do this, they need to draw on similar products to find commonalities while reserving differences. As a result, the resulting products often give people a same-feeling experience, which quickly makes players lose interest. Based on this, this embodiment generates multiple behavior samples by learning behaviors, logically restricting, and expanding behaviors from the designer's and other similar designers' historical works, thereby generating a larger and richer number of samples. These generated behavior samples are then subjected to AI convergent evolution, enhancing the iterative nature of AI design technology. Ultimately, the resulting diversified sample results are output as references for designers, thereby stimulating their creativity.

[0080] The second embodiment of the present invention is:

[0081] A thinking learning method based on AI convergent evolution, based on the above embodiment 1, in this embodiment, further includes between step S3 and step S4:

[0082] S34: Expand the first behavior sample based on the behavior logic to generate multiple second behavior samples, and then permutate and combine the first behavior sample and the second behavior sample to output:

[0083]

[0084]

[0085] Where A is the permutation formula, C is the combination formula, n is the number of samples in the second line, and m is the number of samples in the first line.

[0086] That is, in this embodiment, by arranging and combining the first behavior sample and the second behavior sample, the subsequent designed scenarios are enriched.

[0087] Wherein, before step S4, the method further includes the following steps:

[0088] S40. Establish a convergent evolution function, as shown in formula (3):

[0089] N = L × (1 + d%) k (3).

[0090] Among them, N is the iterative result of the current branch evolution, k is the number of iterations, d is the degree of sample deviation, and L is the level, which includes the original evolution layer, the same evolution layer, and the differential evolution layer.

[0091] That is, by establishing a convergent evolution function, the original behavioral data can be convergently evolved based on the hierarchy to create differential data and realize multi-level data. In this embodiment, the degree of sample deviation can be pre-set and compared with the original data. For example, the original data is that as d increases, the breeze becomes a gale.

[0092] Finally, step S4 is specifically as follows:

[0093] S41. Substitute the original data, similar data and differential data into the original evolution layer, similar evolution layer and differential evolution layer in the L branch of the convergent evolution function in turn for hierarchical evolution, and obtain the original evolution result, similar evolution result and differential evolution result respectively.

[0094] S42. Perform convergent evolution on the original evolution results, similar evolution results, and differential evolution results, and output the final design sample.

[0095] Among them, the original data refers to the behaviors in the historical designs of the current designer included in the first and second behavior samples; the similar data refers to the behaviors in the historical designs of other designers included in the first and second behavior samples; and the difference data refers to the behaviors in the historical designs of the current designer and other designers included in the first and second behavior samples that are different.

[0096] That is, in this embodiment, different original behavioral data - original data, similar data and difference data - are sequentially substituted into the L branch of the convergent evolution function to perform convergent evolution of their own levels. As the levels change, AI will perform differential evolution on the results of each convergent evolution and the original behavioral data to create differential data, ultimately realizing multi-level data to stimulate more diversified creativity of designers.

[0097] The following two cases are listed to specifically illustrate the thinking learning method based on AI convergent evolution of the present invention.

[0098] Case 1: Designer A is an excellent designer, but as he ages, his brainpower feels less and less efficient. For inspiration, he increasingly relies on a vast library of online resources and the collaboration of his subordinates, resulting in low work efficiency. The company happened to need a designer for a game character, aiming for a lively, adorable character with eye-catching skills. Designer A spent a week collecting and integrating various materials, but didn't find a satisfactory result. He believed the resources lacked innovation. Using this patent, Designer A combined his own and other designers' materials through AI-powered "convergent evolution," which generated a collection of reference materials. Guided by this evolution, Designer A successfully found a satisfactory design.

[0099] Case 2: Designer B wants to draw a cat that is different from other cats. It needs to look mischievous but also cute, cute but also smart, and smart with some little thoughts. This requirement is subject to many restrictions and requires the designer to draw on a large amount of online materials. Through this patent, the designer divides cute cats, mischievous cats, smart cats, and cats with little thoughts into L-level training to obtain the corresponding evolutionary levels. By performing "convergent evolution" on the levels, a cat with all the above descriptions can be obtained.

[0100] Please refer to Figure 2 , the third embodiment of the present invention is:

[0101] like Figure 2 As shown, a thinking learning system 1 based on AI convergent evolution includes:

[0102] Behavioral learning module 2 is used to input the historical designs of the current designer and the historical designs of other designers of the same type as the current designer, extract design features from the historical designs, and generate multiple first behavioral samples, that is, to conduct behavioral learning on the historical works of the current designer and other designers, seeking common ground while reserving differences.

[0103] The grammar learning module 3 is used to restrict the behavior logic of the first behavior sample based on the process tree so that the extracted behavior conforms to the logical design.

[0104] The behavior extension module 4 is used to extend the first behavior sample based on the behavior logic and generate multiple second behavior samples. The extension of the behavior also needs to be based on the behavior logic, so that the extended behavior also carries the logic design.

[0105] The convergent evolution module 6 is used to perform convergent evolution on the first behavior sample and the second behavior sample, and output a final design sample.

[0106] That is, in this embodiment, based on the same technical concept, in conjunction with a thinking learning method based on AI convergent evolution in embodiment one, a thinking learning system 1 based on AI convergent evolution is provided, which includes a behavior learning module 2, a grammar learning module 3, a behavior expansion module 4 and a convergent evolution module 6. The behavior learning module 2 performs behavior learning on the historical works of the designer and other designers of the same type, the grammar learning module 3 performs logical restrictions on the behavior, and the behavior expansion module 4 expands more behavior samples, thereby generating more and richer samples. Finally, the convergent evolution module 6 performs AI convergent evolution on the generated behavior samples, thereby enhancing the iterative nature of the AI ​​design technology and ultimately outputting diversified sample results to provide to designers for reference, thereby stimulating the designers' creativity.

[0107] Please refer to Figure 2 , the fourth embodiment of the present invention is:

[0108] A thinking learning system 1 based on AI convergent evolution, based on the above embodiment 3, in this embodiment, Figure 2 As shown, it also includes a logic analysis module 5, which is used to expand the first behavior sample based on the behavior logic, generate multiple second behavior samples, and arrange and combine the first behavior sample and the second behavior sample to output:

[0109]

[0110]

[0111] Where A is the permutation formula, C is the combination formula, n is the number of samples in the second line, and m is the number of samples in the first line.

[0112] That is, in this embodiment, the first behavior sample and the second behavior sample are arranged and combined by the logic analysis module 5, so that the subsequent designed scenarios are richer.

[0113] The convergent evolution module 6 is also used for:

[0114] Convergent evolution function is established as follows:

[0115] N = L × (1 + d%) k (3).

[0116] Among them, N is the iterative result of the current branch evolution, k is the number of iterations, d is the degree of sample deviation, and L is the level, which includes the original evolution layer, the same evolution layer, and the difference evolution layer. In this embodiment, the degree of sample deviation can be pre-set and compared with the original data. For example, the original data is that as d increases, a breeze turns into a gale.

[0117] In addition, in this embodiment, the convergent evolution module 6 is specifically used to:

[0118] Substitute the original data, similar data and differential data into the original evolution layer, similar evolution layer and differential evolution layer in the L branch of the convergent evolution function in turn for hierarchical evolution, and obtain the original evolution result, similar evolution result and differential evolution result respectively;

[0119] The original evolution results, similar evolution results and differential evolution results are subjected to convergent evolution to output the final design sample.

[0120] The original data refers to the behaviors in the historical designs of the current designer included in the first and second behavior samples; the similar data refers to the behaviors in the historical designs of other designers included in the first and second behavior samples; and the difference data refers to the behaviors in the historical designs of the current designer and other designers included in the first and second behavior samples that are different.

[0121] That is, in this embodiment, different original behavioral data - original data, similar data and difference data - are sequentially substituted into the L branch of the convergent evolution function to perform convergent evolution of their own levels. As the levels change, AI will perform differential evolution on the results of each convergent evolution and the original behavioral data to create differential data, ultimately realizing multi-level data to stimulate more diversified creativity of designers.

[0122] In summary, the present invention provides a thinking learning method and system based on AI convergent evolution. It generates multiple behavior samples by performing behavioral learning, logical restrictions, and behavioral expansion on the historical works of designers and other designers of the same type, thereby producing more and richer samples. The generated behavior samples are then subjected to AI convergent evolution to enhance the iterative nature of AI design technology. Ultimately, diversified sample results are output to provide to designers for reference, thereby stimulating their creativity and saving creative time.

[0123] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent transformations made using the contents of the present invention's description and drawings, or directly or indirectly applied in related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A thinking learning method based on AI convergent evolution, characterized by: Including steps: S1. Input the historical designs of the current designer and the historical designs of other designers of the same type as the current designer, extract design features from the historical designs, and generate a plurality of first behavior samples; S2. limiting the behavior logic of the first behavior sample based on the process tree; S3. Expand the first behavior sample based on the behavior logic to generate multiple second behavior samples; S4. Perform convergent evolution on the first behavior sample and the second behavior sample, and output a final design sample; The step S4 also includes the following steps: S40. Establish a convergent evolution function, as shown in formula (3): （3）； Wherein, N is the result of the current branch evolution iteration, k is the number of iterations, d is the degree of sample deviation, and L is the level, which includes the original evolution layer, the same evolution layer, and the differential evolution layer; The step S4 is specifically as follows: S41, sequentially substituting the original data, similar data, and differential data into the original evolution layer, the similar evolution layer, and the differential evolution layer in the L branch of the convergent evolution function for hierarchical evolution, to obtain an original evolution result, a similar evolution result, and a differential evolution result, respectively; S42, performing convergent evolution on the original evolution result, the similar evolution result, and the differential evolution result, and outputting the final design sample; The original data is the behaviors in the historical designs of the current designer included in the first behavior sample and the second behavior sample; The similar data are behaviors in the historical designs of the other designers included in the first behavior sample and the second behavior sample; The difference data is behaviors that are different between the historical designs of the current designer and the other designers, included in the first behavior sample and the second behavior sample.

2. A thinking learning method based on AI convergent evolution according to claim 1, characterized in that: The steps between step S3 and step S4 also include: S34: Expand the first behavior sample based on the behavior logic to generate multiple second behavior samples, and output the first behavior sample and the second behavior sample by permutation and combination. （1）； （2）； Wherein, A is the permutation formula, C is the combination formula, n is the number of samples in the second behavior, and m is the number of samples in the first behavior.

3. A thinking learning system based on AI convergent evolution, characterized by: include: A behavior learning module is configured to input historical designs of a current designer and historical designs of other designers of the same type as the current designer, extract design features from the historical designs, and generate a plurality of first behavior samples; a grammar learning module, configured to restrict the behavior logic of the first behavior sample based on a process tree; a behavior extension module, configured to extend the first behavior sample based on the behavior logic to generate a plurality of second behavior samples; a convergent evolution module, configured to perform convergent evolution on the first behavior sample and the second behavior sample, and output a final design sample; The convergent evolution module is also used to: Establish the convergent evolution function as follows: （3）； Wherein, N is the result of the current branch evolution iteration, k is the number of iterations, d is the degree of sample deviation, and L is the level, which includes the original evolution layer, the same evolution layer, and the differential evolution layer; The convergent evolution module is specifically used for: Substituting the original data, similar data and differential data into the original evolution layer, the similar evolution layer and the differential evolution layer in the L branch of the convergent evolution function in turn for convergent evolution, and obtaining the original evolution result, the similar evolution result and the differential evolution result respectively; Outputting a plurality of final design samples according to the original evolution result, the similar evolution result and the differential evolution result; The original data is the behaviors in the historical designs of the current designer included in the first behavior sample and the second behavior sample; The similar data are behaviors in the historical designs of the other designers included in the first behavior sample and the second behavior sample; The difference data is behaviors that are different between the historical designs of the current designer and the other designers, included in the first behavior sample and the second behavior sample.

4. The thinking learning system based on AI convergent evolution according to claim 3 is characterized in that: It also includes a logic parsing module, which is used to expand the first behavior sample based on the behavior logic, generate multiple second behavior samples, and output the first behavior sample and the second behavior sample by arranging and combining them: （1）； （2）； Wherein, A is the permutation formula, C is the combination formula, n is the number of samples in the second behavior, and m is the number of samples in the first behavior.

Images