CN109271604B - Advertisement layout method and device and computer equipment - Google Patents

Advertisement layout method and device and computer equipment Download PDF

Info

Publication number
CN109271604B
CN109271604B CN201811133550.5A CN201811133550A CN109271604B CN 109271604 B CN109271604 B CN 109271604B CN 201811133550 A CN201811133550 A CN 201811133550A CN 109271604 B CN109271604 B CN 109271604B
Authority
CN
China
Prior art keywords
advertisement
layout
size
model
training
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201811133550.5A
Other languages
Chinese (zh)
Other versions
CN109271604A (en
Inventor
徐芳芳
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
3600 Technology Group Co ltd
Original Assignee
3600 Technology Group Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 3600 Technology Group Co ltd filed Critical 3600 Technology Group Co ltd
Priority to CN201811133550.5A priority Critical patent/CN109271604B/en
Publication of CN109271604A publication Critical patent/CN109271604A/en
Application granted granted Critical
Publication of CN109271604B publication Critical patent/CN109271604B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0241—Advertisements
    • G06Q30/0277—Online advertisement

Abstract

The invention discloses an advertisement layout method, an advertisement layout device and computer equipment, which are applied to the field of Internet. The method comprises the following steps: based on element characteristics of advertisement elements and element characteristics formed by interaction of more than two advertisement elements, an initial layout model is built together, wherein advertisement elements laid out in a plurality of advertisement layout areas form banner advertisements; training an initial layout model by utilizing the original size advertisement template sample to obtain a banner advertisement layout model; a banner advertisement layout scheme is generated using the advertisement layout model. The invention solves the technical problem of low design efficiency of advertisement layout.

Description

Advertisement layout method and device and computer equipment
Technical Field
The present invention relates to the field of internet technologies, and in particular, to an advertisement layout method, an advertisement layout device, and a computer device.
Background
Banner advertisement (Banner Ad.) is the earliest form employed by web advertisements, and is the most common form of advertisement today. When a user clicks on a banner advertisement, it is typically possible to link to the advertiser's web page. While banner advertisement designs require placement and color matching. Layout design is the first step in the process, and greatly influences the quality of banner advertisements.
At present, the layout design of banner advertisements is mainly completed by manually utilizing design software such as PS (PS) and the like, is very low-efficiency and time-consuming, and cannot meet the advertisement putting requirement.
Disclosure of Invention
The embodiment of the invention provides an advertisement layout method, an advertisement layout device and computer equipment, so as to solve the technical problem of low layout design efficiency of banner advertisements in the prior art.
In a first aspect, an embodiment of the present invention provides an advertisement layout method, including:
based on element characteristics of advertisement elements and element characteristics formed by interaction of more than two advertisement elements, an initial layout model is built together, wherein advertisement elements laid out in a plurality of advertisement layout areas form banner advertisements;
training the initial layout model by utilizing a primary size advertisement template sample to obtain a banner advertisement layout model;
and generating a banner advertisement layout scheme by using the advertisement layout model.
Optionally, the establishing an initial layout model based on the element features of the advertisement elements and the element features formed by interaction of more than two advertisement elements includes:
abstracting each advertisement element used for layout in the advertisement layout area into a corresponding rectangular area;
And performing mathematical modeling by taking the position coordinates and the size information of each rectangular area as solutions to obtain the initial layout model.
Optionally, training the initial layout model by using the original size advertisement template sample to obtain a banner advertisement layout model includes:
constructing an advertisement layout measurement function aiming at the initial layout model;
and training the initial layout model by using the original size advertisement template sample until the energy function value of the advertisement layout measurement function is minimum, so as to obtain the banner advertisement layout model.
Optionally, the energy function value of the advertisement layout measurement function is a total energy value obtained by weighted summation of a plurality of energy values related to layout quality.
Optionally, training the initial layout model using the original size advertisement template sample until an energy function value of the advertisement layout metric function is minimum, to obtain the banner advertisement layout model, including:
initializing specific parameters of the advertisement layout measurement function, and putting the original-size advertisement template sample into a candidate solution;
taking the solution with the minimum total energy value of the advertisement layout measurement function in the candidate solutions as the current optimal solution, and carrying out reverse optimization processing based on the current optimal solution to obtain a new solution;
Updating the specific parameters based on a gradient descent method and judging whether a training termination condition is reached;
if the training termination condition is met, outputting the current optimal solution as the banner advertisement layout model; otherwise, updating the candidate solution based on the new solution, and returning to the step of taking the solution which minimizes the total energy value of the advertisement layout metric function in the candidate solution as the current optimal solution based on the updated candidate solution and the updated specific parameter.
Optionally, the determining whether the training termination condition is met includes:
judging whether the current function value of the objective function reaches a preset function value, if so, determining that the training termination condition is reached; the objective function is used for representing the difference between the total energy value of the advertisement layout template corresponding to the current optimal solution and the total energy value of the optimal layout scheme generated through an optimization algorithm.
Optionally, the step of generating the optimal layout scheme through an optimization algorithm includes:
performing annealing operation for more than one time based on the simulated annealing algorithm to generate advertisement layout proposals with preset quantity;
and determining the optimal layout scheme from the advertisement layout proposals with the preset quantity.
Optionally, the simulated annealing algorithm includes a plurality of annealing operators associated with the advertisement elements; in the step of generating a preset number of advertisement layout proposals based on performing the simulated annealing operation more than once, each advertisement layout proposal is generated by:
selecting one annealing operator from the annealing operators related to the advertisement elements aiming at the current proposal generating operation in the current annealing operation, wherein the current annealing operation comprises the proposal generating operation with preset times;
and executing the current proposal generating operation by using a simulated annealing algorithm containing the annealing operator selected at the current time to generate a corresponding advertisement layout proposal.
Optionally, after the generating the banner advertisement layout scheme using the advertisement layout model, the method further includes:
performing size changing treatment on the original size advertisement template sample to generate a new size advertisement template sample;
training the initial layout model based on the original size advertisement template sample and the new size advertisement template sample together to train a variable size layout model;
and generating a variable-size advertisement layout scheme by using the variable-size layout model.
In a second aspect, an embodiment of the present invention provides an advertisement layout apparatus, including:
the modeling unit is used for jointly establishing an initial layout model based on element characteristics of the advertisement elements and element characteristics formed by interaction of more than two advertisement elements, wherein the advertisement elements laid out in a plurality of advertisement layout areas form banner advertisements;
the first model training unit is used for training the initial layout model by utilizing the original size advertisement template sample to obtain a banner advertisement layout model;
and the first advertisement layout unit is used for generating a banner advertisement layout scheme by utilizing the advertisement layout model.
Optionally, the modeling unit includes:
an abstraction subunit, configured to abstract each advertisement element used for layout in the advertisement layout area into a corresponding rectangular area;
and the mathematical modeling sub-unit is used for performing mathematical modeling by taking the position coordinates and the size information of each rectangular area as solutions to obtain the initial layout model.
Optionally, the first model training unit includes:
a function construction subunit, configured to construct an advertisement layout metric function for the initial layout model;
and the sample training subunit is used for training the initial layout model by using the original-size advertisement template sample until the energy function value of the advertisement layout measurement function is minimum so as to obtain the banner advertisement layout model.
Optionally, the energy function value of the advertisement layout measurement function is a total energy value obtained by weighted summation of a plurality of energy values related to layout quality.
Optionally, the sample training subunit is specifically configured to:
initializing specific parameters of the advertisement layout measurement function, and putting the original-size advertisement template sample into a candidate solution;
taking the solution with the minimum total energy value of the advertisement layout measurement function in the candidate solutions as the current optimal solution, and carrying out reverse optimization processing based on the current optimal solution to obtain a new solution;
updating the specific parameters based on a gradient descent method and judging whether a training termination condition is reached;
if the training termination condition is met, outputting the current optimal solution as the banner advertisement layout model; otherwise, updating the candidate solution based on the new solution, and returning to the step of taking the solution which minimizes the total energy value of the advertisement layout metric function in the candidate solution as the current optimal solution based on the updated candidate solution and the updated specific parameter.
Optionally, the sample training subunit is specifically configured to:
judging whether the current function value of the objective function reaches a preset function value, if so, determining that the training termination condition is reached; the objective function is used for representing the difference between the total energy value of the advertisement layout template corresponding to the current optimal solution and the total energy value of the optimal layout scheme generated through an optimization algorithm.
Optionally, in the step of generating the optimal layout scheme by an optimization algorithm, the sample training subunit is specifically configured to:
performing annealing operation for more than one time based on the simulated annealing algorithm to generate advertisement layout proposals with preset quantity;
and determining the optimal layout scheme from the advertisement layout proposals with the preset quantity.
Optionally, the simulated annealing algorithm includes a plurality of annealing operators associated with the advertisement elements;
the sample training subunit is configured to, in the step of generating a preset number of advertisement layout proposals based on performing the annealing operation more than once based on the simulated annealing algorithm, generate each advertisement layout proposal by:
selecting one annealing operator from the annealing operators related to the advertisement elements aiming at the current proposal generating operation in the current annealing operation, wherein the current annealing operation comprises the proposal generating operation with preset times;
and executing the current proposal generating operation by using a simulated annealing algorithm containing the annealing operator selected at the current time to generate a corresponding advertisement layout proposal.
Optionally, the apparatus further includes:
the template processing unit is used for carrying out size changing processing on the original-size advertisement template sample to generate a new-size advertisement template sample;
The second model training unit is used for training the initial layout model based on the original size advertisement template sample and the new size advertisement template sample together so as to train a variable size layout model;
and a second advertisement layout unit for generating a variable-size advertisement layout scheme using the variable-size layout model.
In a third aspect, an embodiment of the present invention provides a computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the steps of the method according to any one of the possible implementations of the first aspect when the program is executed.
In a fourth aspect, embodiments of the present invention provide a computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of the method of any one of the possible implementations of the first aspect.
The technical scheme provided by the embodiment of the invention has at least the following technical effects or advantages:
the advertisement layout method, the advertisement layout device and the computer equipment provided by the embodiment of the invention jointly establish an initial layout model based on element characteristics of advertisement elements forming banner advertisements and element characteristics formed by interaction of more than two advertisement elements; training an initial layout model by utilizing the original size advertisement template sample to obtain a banner advertisement layout model; a banner advertisement layout scheme is generated using the advertisement layout model. Thereby realizing that the advertisement layout model for automatically generating the banner advertisement layout scheme is generated based on machine learning for automatically designing the advertisement layout, and therefore, the layout design efficiency of banner advertisements can be improved.
Drawings
Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to designate like parts throughout the figures. In the drawings:
FIG. 1 is a flow chart illustrating an advertisement placement method in an embodiment of the present invention;
FIG. 2 is a flow chart of an advertisement layout method in another embodiment of the invention
FIG. 3 is a schematic diagram showing the structure of an advertisement layout device according to an embodiment of the present invention;
fig. 4 shows a schematic structural diagram of a computer device in an embodiment of the present invention.
Detailed Description
In order to solve the technical problem of low layout design efficiency of banner advertisements in the prior art. The embodiment of the invention provides an advertisement layout method, an advertisement layout device and computer equipment, wherein the overall thought is as follows:
based on element characteristics of advertisement elements forming the banner advertisement and element characteristics formed by interaction of more than two advertisement elements, an initial layout model is built together; training an initial layout model by utilizing the original size advertisement template sample to obtain a banner advertisement layout model; a banner advertisement layout scheme is generated using the advertisement layout model. Based on the technical scheme, the method and the device realize that the advertisement layout model for automatically generating the banner advertisement layout scheme is generated based on machine learning and is used for automatically designing the advertisement layout, so that the layout design efficiency of banner advertisements can be improved.
In order to better understand the above technical solutions, the following detailed description will refer to the accompanying drawings and specific embodiments. The term "plurality" as used herein refers to "two or more" including "two" cases and "one" or "more" cases.
In a first aspect, an embodiment of the present invention provides an advertisement layout method for completing automatic advertisement layout design of banner advertisements. Referring to fig. 1, the advertisement layout method provided by the embodiment of the invention includes the following steps:
step S101, an initial layout model is built together based on element characteristics of advertisement elements and element characteristics formed by interaction of more than two advertisement elements, wherein the advertisement elements laid out in a plurality of advertisement layout areas form banner advertisements.
In the embodiment of the invention, the advertisement elements laid out in the layout area of the banner advertisement comprise various or all of the following five advertisement elements: merchandise pictures, brand logo, advertising themes, advertising subtitles, button labels (e.g., purchase immediately, see details, etc.). Of course, other advertisement elements may be supplemented for different service scenarios, such as: background patterns, decorations, and the like.
In an alternative embodiment, the initial layout model is built based on the following five element features of the advertisement elements and the element features of each two of the following five element features that are interactively formed together: commodity pictures, brand logo, advertising theme, advertising subheading, key features of button labels.
It should be noted that, the element features in the embodiments of the present invention are related to advertisement layout. Specifically, the element features of the advertisement element include the following three types:
first kind: distances between all ad elements and boundaries of the ad layout region.
Second kind: the size of all advertisement elements.
Third kind: all the advertisement elements are characterized by elements formed by interaction of two advertisement elements, and specifically: the relative positions of all advertisement elements.
Of course, other element features related to advertisement layout may be added during implementation.
In a specific embodiment, the initial layout model is built by: and abstracting each advertisement element used for layout in the advertisement layout area into a corresponding rectangular area, and carrying out mathematical modeling by taking the position coordinates and the size information of each rectangular area as solutions to obtain an initial layout model.
Specifically, each advertisement element is abstracted into a corresponding rectangular area, and an initial layout model for the banner advertisement is established with the upper left corner of the advertisement layout area (canvas of the banner advertisement) as the origin of coordinates and with the position coordinates and size information of the rectangular area of each advertisement element as the solution.
Given the example of an advertisement element j of layout scheme X, X (j, 0) is the abscissa of the advertisement element j, X (j, 1) is the ordinate of the advertisement element j, X (j, 2) is the width of the advertisement element j, and X (j, 3) is the height of the advertisement element j. Thus, the location coordinates of advertisement element j are: x (j, 0), X (j, 1); the size information of the advertisement element j is: x (j, 2), X (j, 3). With this rule, the position coordinates and size information of the rectangular area to which each advertisement element in the given layout scheme X is abstracted can be obtained. Thus, taking a given layout scheme X for laying out five advertisement elements as an example, the five advertisement element correspondence is abstracted into position coordinates and size information of five rectangular areas, and thus the initial layout model is established for the layout for five advertisement elements.
Step S102, training an initial layout model by utilizing the original size advertisement template sample to obtain a banner advertisement layout model.
In the embodiment of the present invention, if the layout quality metric function represents the advertisement layout quality, step S102 specifically includes:
and S1021, constructing an advertisement layout measurement function aiming at the initial layout model, wherein the advertisement layout measurement function is a total energy value obtained by carrying out weighted summation on a plurality of energy values related to the layout quality.
Defining a plurality of energy values, in an embodiment of the present invention, at least three energy values are included as follows:
1. distance energy value: the distances of all ad elements from the ad layout region boundaries are characterized.
2. Relative position energy value: the relative positions of all advertisement elements (i.e., the position of the advertisement element compared to another advertisement element).
3. Size energy value: all advertisement element sizes.
Further, more energy values may be added according to different problem complexities, where each energy value includes one or more terms.
In the embodiment of the invention, the constructed advertisement layout measurement function is specifically as follows:
Figure BDA0001814154650000081
wherein E is i Represents the i-th energy value, w i Representing the weight value, alpha, corresponding to the energy value of the i-th item i Controlling the smoothness of each energy value function, wherein θ is a specific parameter of the advertisement layout measurement function, namely: parameters that need to be solved. X represents a given layout scheme, and each energy value of the given layout scheme can be calculated based on the original size advertisement template sample. However, the specific parameter θ for this given layout scheme needs to be solved by an optimization algorithm. And when the characteristic parameter theta obtained by solving is optimal, the energy function value of the corresponding advertisement layout measurement function is minimum.
In the advertisement layout measurement function, the characteristic parameter theta is split into two parts [ w, alpha ]]W is as above i And alpha is i 。
Specifically, each advertisement element laid out in the advertisement layout area is divided into two types of advertisement elements: text elements related to text, image elements related to images. For example, in terms of commodity pictures, brand logo, advertisement theme, advertisement subtitle, and button label, the commodity pictures and brand logo are image elements, and the advertisement theme, advertisement subtitle, and button label are text elements.
Different types of advertisement elements correspond to different distance energy values, and specifically, the distance energy values include: distance energy value of text element
Figure BDA0001814154650000091
And distance energy value of picture element +.>
Figure BDA0001814154650000092
Distance energy value of text element
Figure BDA0001814154650000093
Is the average of the nearest distances of all text elements in the advertisement element from the boundary of the advertisement layout area.
Figure BDA0001814154650000094
Wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure BDA0001814154650000095
representing distance energy value from text element, +.>
Figure BDA0001814154650000096
And (3) representing the nearest distance between the ith text element and the boundary of the advertisement layout area, wherein n is the number of the text elements, and i epsilon (text) represents the value of i in the number of the text elements.
The distance energy value of the image element can be obtained based on the same way
Figure BDA0001814154650000097
The distance energy value of an image element is the average of the nearest distances of all image elements in the advertisement element from the boundary of the advertisement layout area.
Relative position energy value E dist In terms of definition: the average of the distances between all ad elements and the nearest ad element:
Figure BDA0001814154650000098
m is the number of all advertisement elements in the advertisement layout area, including: text element and image element, 1-S (min j∈(all) (d ij )；α dist ) Representing the distance between the ith advertisement element and the nearest advertisement element, i E (all) representing allThe number of advertisement elements is a value.
Specifically, the magnitude energy value includes: size energy value E of text element textSize Size energy value E of image element graphicSize
Size energy value E for text element textSize In other words, it can be defined as:
Figure BDA0001814154650000101
wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure BDA0001814154650000102
defined as->
Figure BDA0001814154650000103
Figure BDA0001814154650000104
For the height of the text element +.>
Figure BDA0001814154650000105
And h is the height of the advertisement layout area.
Likewise, the size energy value of the picture element may be calculated, wherein, when calculating the size energy value of the picture element,
Figure BDA0001814154650000106
defined as->
Figure BDA0001814154650000107
Figure BDA0001814154650000108
For the height of the picture element +.>
Figure BDA0001814154650000109
For the width of the image element, w and h are the width and height, respectively, of the advertisement layout area.
In the above calculation formula of energy values, each energy value is calculated based on the mapping function S (x; α) =arctan (xα)/arctan (α), then specifically:
When calculating the distance energy value:
Figure BDA00018141546500001010
when calculating the relative position energy value:
S(min j∈(all) (d ij )；α dist )＝arctan(min j∈(all) (d ij )α dist )/arctan(α dist )
when calculating the magnitude energy value:
Figure BDA00018141546500001011
the energy function value of the advertisement layout metric function is a total energy value obtained by weighted summation of a plurality of energy values related to the layout quality. Through the steps, each energy value can be obtained, and the total energy value can be obtained based on the weighted summation of each energy value. However, the optimal result of the specific parameter θ of the advertisement layout metric function is unknown, and the optimal result of the specific parameter θ corresponds to the actual layout scheme X that minimizes the total energy value of the advertisement layout metric function, and thus the corresponding actual layout scheme X is optimal, i.e., the optimal actual layout scheme X T 。
Step S1022, training the initial layout model by utilizing the original size advertisement template sample until the energy function value of the advertisement layout measurement function is minimum, and ending training to obtain the banner advertisement layout model.
In step S1022, the layout quality metric function measures position coordinates and size information of each advertisement element transformed with the original-size advertisement template sample, thereby measuring advertisement layout quality.
Wherein the smaller the energy function value of the layout quality metric function, the higher the advertisement layout quality is characterized. And then, transforming the position coordinates and the size information of each advertisement element by using the original size advertisement template sample until the energy function value of the layout quality measurement function is minimum, and ending training to obtain a trained banner advertisement layout model.
In an alternative embodiment, the process of training the initial layout model using the original size advertisement template sample until the energy function value of the advertisement layout metric function is minimum to obtain the banner advertisement layout model includes the following steps a1 to a4:
step a1, initializing specific parameters of an advertisement layout measurement function, and putting a full-size advertisement template sample into a candidate solution.
And a step a2, using the solution which minimizes the total energy value of the advertisement layout measurement function in the candidate solutions as the current optimal solution, and performing reverse optimization processing based on the current optimal solution to obtain a new solution.
And a3, updating specific parameters based on a gradient descent method and judging whether the training termination condition is reached.
Step a4, if the training termination condition is met, outputting the current optimal solution as a banner advertisement layout model; otherwise, updating the candidate solution based on the new solution, and returning to the step a2 based on the updated candidate solution and the updated specific parameters.
Specifically, the energy function value of the advertisement layout metric function is the total energy value E (X; θ) obtained by weighted summation of the energy values. In the implementation process, the real layout scheme X is optimal according to the specific parameter θ of the optimal result, and the total energy value E (X T The method comprises the steps of carrying out a first treatment on the surface of the θ) is minimum, X T Representing the optimal real layout scheme. But initially the value of θ that minimizes the total energy value is not known.
In order to obtain a value θ that minimizes the total energy value, an objective function is set in the embodiment of the present invention. According to the advertisement layout optimal corresponding to the θ value with the minimum total energy value, the objective function represents the difference between the total energy value of the optimal real layout scheme X and the total energy value corresponding to the optimal layout scheme solved by the optimization algorithm.
Figure BDA0001814154650000121
Wherein E (X) T The method comprises the steps of carrying out a first treatment on the surface of the θ) is the total energy value of the optimal real layout scheme X,
Figure BDA0001814154650000122
for initial values, λ is the regularization parameter, s determines which parameters need regularization. The optimal layout scheme obtained by the optimization algorithm is X S ＝min X E (X; θ). Given layout scheme X corresponding to θ value such that G (θ) =0 T For optimum, the training is ended at this time to obtain a banner advertisement layout model.
Iteratively updating the value of the specific parameter θ based on a gradient descent algorithm to continuously decrease the value G (θ) of the objective function: until a θ value required to make the objective function G (θ) =0 is obtained, the following is obtained: optimal results for the particular parameters.
Specifically, iteratively updating the value of the specific parameter θ based on the gradient descent algorithm specifically includes: deriving an objective function:
Figure BDA0001814154650000123
Step S103, generating a banner advertisement layout scheme by utilizing the advertisement layout model.
In an alternative embodiment, judging whether the current function value of the objective function reaches a preset function value, if so, determining that a training termination condition is reached; the objective function is used for representing the difference between the total energy value of the advertisement layout template corresponding to the current optimal solution and the total energy value of the optimal layout scheme generated by the optimization algorithm.
In the embodiment of the invention, the optimization algorithm can be a simulated annealing algorithm, and the current optimal layout scheme is obtained based on the simulated annealing algorithm.
The optimization algorithm used to solve the current optimal layout scheme may be a simulated annealing algorithm. In a specific embodiment. Specifically, the step of generating the optimal layout scheme based on the simulated annealing algorithm comprises the steps of b 1-b 2:
and b1, performing annealing operation for more than one time based on the simulated annealing algorithm to generate advertisement layout proposals with preset quantity.
The simulated annealing algorithm in embodiments of the present invention includes one or more annealing operators associated with the advertisement elements. Specifically, the annealing operator related to the advertisement element comprises one or more of the following: changing the position of an advertisement element, changing the size of an advertisement element, exchanging two advertisement elements. In the specific implementation process, other annealing operators can be extended to the simulated annealing algorithm according to the service scene, for example: changing the alignment between advertisement elements.
In the specific implementation process, the element positions are changed, specifically: by adding a gaussian distribution offset to the current position of the advertisement element. The changing element size is specifically: by adding a gaussian distribution offset at the current height.
In step b 1: and setting a proposal upper limit for each annealing operation respectively, wherein when the number of advertisement layout proposals generated based on the current annealing operation reaches the corresponding proposal upper limit, ending the current annealing operation, and carrying out the next annealing operation, and circulating until each annealing operation is finished so as to generate the preset number of advertisement layout proposals.
Specifically, each annealing operation includes a preset number of proposal generation operations. Each proposal generating operation is corresponding to one annealing operator, and the probability of which annealing operator is adopted by each proposal generating operation can be equal or unequal. For example, the annealing operator may be selected based on a roulette algorithm.
Each proposal generation operation generates a corresponding advertisement layout proposal. Selecting one annealing operator from a plurality of annealing operators related to advertisement elements aiming at the current proposal generating operation in the current annealing operation, wherein the current annealing operation comprises the proposal generating operation with preset times; and executing the current proposal generating operation by using a simulated annealing algorithm containing the annealing operator selected at the current time to generate a corresponding advertisement layout proposal.
Each proposal generating operation generates one advertisement layout proposal, so that the advertisement layout proposal is generated based on the proposal generating operation of the preset times, and the number of advertisement layout proposals generated by the annealing operation of the current time reaches the corresponding proposal upper limit. For example, if the upper limit of proposals set for the current annealing operation is p, p advertisement layout proposals are generated based on the proposal generation operation p times in the current annealing operation.
After step b1, step b2 is then performed to determine an optimal layout scheme from a preset number of advertisement layout schemes.
More specifically, the implementation process for generating the optimal layout scheme based on the simulated annealing algorithm comprises the following steps:
and step 1, initializing and setting a cooling process. Specifically, an initial solution is randomly generated as a candidate solution proposal and an initial annealing temperature. And putting a preset number of advertisement layout proposals into the candidate solutions.
And 2, selecting a solution which minimizes the energy value of the advertisement layout energy function from the candidate solution proposals.
And step 3, generating an advertisement layout proposal based on the annealing operator selected currently.
And step 4, judging whether the generated advertisement layout proposal is acceptable, if so, executing step s5, otherwise, executing step s6.
And 5, generating whether the total energy value of the new solution is lower than the total energy value corresponding to the candidate solution based on the candidate solution, if so, executing the step 8, otherwise, executing the step 7.
And 6, judging whether the number of the generated advertisement layout proposals reaches the upper limit of the proposal, if so, executing the step 9, otherwise, selecting the next annealing operator and returning to the step 3.
And 7, judging whether to accept the bad solution or not based on the Metropolis criterion, if so, executing the step 8, otherwise, executing the step 9.
And 8, adding the new solution into the candidate solution to obtain an updated candidate solution, and executing the step 8.
And 9, judging whether the optimized termination condition is met, if so, executing the step 10, otherwise, cooling and returning to the step 2.
And step 10, outputting a solution with the minimum total energy value of the current time as an optimal layout scheme.
Further, the embodiment of the invention also provides a technical scheme of the variable-size advertisement layout, and referring to fig. 2, the implementation process of the variable-size advertisement layout comprises the following steps:
s201, performing size changing processing on the original size advertisement template sample to generate a new size advertisement template sample.
Specifically, a variable-size layout metric function is built for the initial layout model, and the variable-size layout metric function can be described as:
Figure BDA0001814154650000141
Wherein X is p The definition of other parameters representing the original size advertisement template refers to the definition in the layout metric function, and is not repeated here for brevity of description.
The size layout measurement function comprises an energy value describing the original condition of the advertisement element, and also comprises the same energy value as that in the advertisement layout measurement function, namely: the energy values related to the layout quality are not described in detail for brevity of description. Specifically, the energy function value of the variable size layout metric function is a total energy value obtained by weighted summation of the distance energy value of the text element, the distance energy value of the image element, the relative position energy values of all the advertisement elements, the size energy value of the text element, the small energy value of the image element, and the energy value describing the original condition of the advertisement element.
Specifically, the energy value of the original condition of the advertisement element is described as follows:
Figure BDA0001814154650000151
/>
Figure BDA0001814154650000152
wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure BDA0001814154650000153
and->
Figure BDA0001814154650000154
New height dimension, original height dimension, ++i of the ith advertisement element, respectively>
Figure BDA0001814154650000155
And->
Figure BDA0001814154650000156
The relative position in the new advertisement layout and the relative position in the original advertisement layout are respectively in the center of the advertisement element.
S202, training an initial layout model based on the original size advertisement template sample and the new size advertisement template sample to train a variable size layout model.
Assume that there are 12 pairs of training sample data: the method comprises 12 original size advertisement template samples and 12 new size advertisement template samples, in the specific implementation process, the implementation of solving the parameter theta in the size layout measurement function can refer to the implementation of solving the specific parameter theta in the size layout measurement function, so that the final result of the parameter theta in the size layout measurement function is learned, and similarly, the optimal variable size layout scheme in the variable size layout optimization process is calculated according to the simulated annealing algorithm, so that the new size advertisement template sample is obtained.
S203, generating a variable-size advertisement layout scheme by using the variable-size layout model.
In a second aspect, an embodiment of the present invention provides an advertisement layout apparatus, as shown in fig. 3, where the advertisement layout apparatus provided by the embodiment of the present invention includes:
a modeling unit 301, configured to jointly establish an initial layout model based on element features of advertisement elements and element features formed by interaction of two or more advertisement elements, where advertisement elements laid out in a plurality of advertisement layout areas form banner advertisements;
a first model training unit 302, configured to train the initial layout model by using a full-size advertisement template sample, so as to obtain a banner advertisement layout model;
A first advertisement layout unit 303 for generating a banner advertisement layout scheme using the advertisement layout model.
In an alternative embodiment, the modeling unit 301 includes:
an abstraction subunit, configured to abstract each advertisement element used for layout in the advertisement layout area into a corresponding rectangular area;
and the mathematical modeling sub-unit is used for performing mathematical modeling by taking the position coordinates and the size information of each rectangular area as solutions to obtain the initial layout model.
In an alternative embodiment, the first model training unit 302 includes:
a function construction subunit, configured to construct an advertisement layout metric function for the initial layout model;
and the sample training subunit is used for training the initial layout model by using the original-size advertisement template sample until the energy function value of the advertisement layout measurement function is minimum so as to obtain the banner advertisement layout model.
In an alternative embodiment, the energy function value of the advertisement layout metric function is a total energy value obtained by weighted summation of a plurality of energy values related to layout quality.
In an alternative embodiment, the sample training subunit is specifically configured to:
Initializing specific parameters of the advertisement layout measurement function, and putting the original-size advertisement template sample into a candidate solution;
taking the solution with the minimum total energy value of the advertisement layout measurement function in the candidate solutions as the current optimal solution, and carrying out reverse optimization processing based on the current optimal solution to obtain a new solution;
updating the specific parameters based on a gradient descent method and judging whether a training termination condition is reached;
if the training termination condition is met, outputting the current optimal solution as the banner advertisement layout model; otherwise, updating the candidate solution based on the new solution, and returning to the step of taking the solution which minimizes the total energy value of the advertisement layout metric function in the candidate solution as the current optimal solution based on the updated candidate solution and the updated specific parameter.
In an alternative embodiment, the sample training subunit is specifically configured to:
judging whether the current function value of the objective function reaches a preset function value, if so, determining that the training termination condition is reached; the objective function is used for representing the difference between the total energy value of the advertisement layout template corresponding to the current optimal solution and the total energy value of the optimal layout scheme generated through an optimization algorithm.
In an alternative embodiment, the sample training subunit is specifically configured to, in the step of generating the optimal layout solution by using an optimization algorithm:
performing annealing operation for more than one time based on the simulated annealing algorithm to generate advertisement layout proposals with preset quantity;
and determining the optimal layout scheme from the advertisement layout proposals with the preset quantity.
In an alternative embodiment, the simulated annealing algorithm includes a plurality of annealing operators associated with the advertising element;
the sample training subunit is configured to, in the step of generating a preset number of advertisement layout proposals based on performing the annealing operation more than once based on the simulated annealing algorithm, generate each advertisement layout proposal by:
selecting one annealing operator from the annealing operators related to the advertisement elements aiming at the current proposal generating operation in the current annealing operation, wherein the current annealing operation comprises the proposal generating operation with preset times;
and executing the current proposal generating operation by using a simulated annealing algorithm containing the annealing operator selected at the current time to generate a corresponding advertisement layout proposal.
In an alternative embodiment, the apparatus further comprises:
The template processing unit is used for carrying out size changing processing on the original-size advertisement template sample to generate a new-size advertisement template sample;
the second model training unit is used for training the initial layout model based on the original size advertisement template sample and the new size advertisement template sample together so as to train a variable size layout model;
and a second advertisement layout unit for generating a variable-size advertisement layout scheme using the variable-size layout model.
In a third aspect, an embodiment of the present invention provides a computer device, as shown in fig. 4, including a memory 404, a processor 402, and a computer program stored in the memory 404 and executable on the processor 402, where the processor 402 implements the steps of any one of the foregoing embodiments of the advertisement layout method when the program is executed.
Where in FIG. 4 a bus architecture (represented by bus 400), bus 400 may comprise any number of interconnected buses and bridges, with bus 400 linking together various circuits, including one or more processors, represented by processor 402, and memory, represented by memory 404. Bus 400 may also link together various other circuits such as peripheral devices, voltage regulators, power management circuits, etc., as are well known in the art and, therefore, will not be described further herein. Bus interface 405 provides an interface between bus 400 and receiver 401 and transmitter 403. The receiver 401 and the transmitter 403 may be the same element, i.e. a transceiver, providing a means for communicating with various other apparatus over a transmission medium. The processor 402 is responsible for managing the bus 400 and general processing, while the memory 404 may be used to store data used by the processor 402 in performing operations.
According to a fourth aspect of the present invention there is provided a computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of any one of the preceding embodiments of an advertisement layout method.
One or more technical solutions provided in the embodiments of the present invention at least have the following technical effects or advantages:
the advertisement layout method, the advertisement layout device and the computer equipment provided by the embodiment of the invention jointly establish an initial layout model based on element characteristics of advertisement elements forming banner advertisements and element characteristics formed by interaction of more than two advertisement elements; training an initial layout model by utilizing the original size advertisement template sample to obtain a banner advertisement layout model; a banner advertisement layout scheme is generated using the advertisement layout model. Thereby realizing that the advertisement layout model for automatically generating the advertisement layout scheme is generated based on machine learning for automatically designing the advertisement layout, and therefore, the layout design efficiency of banner advertisements can be improved.
The algorithms and displays presented herein are not inherently related to any particular computer, virtual system, or other apparatus. Various general-purpose systems may also be used with the teachings herein. The required structure for a construction of such a system is apparent from the description above. In addition, the present invention is not directed to any particular programming language. It will be appreciated that the teachings of the present invention described herein may be implemented in a variety of programming languages, and the above description of specific languages is provided for disclosure of enablement and best mode of the present invention.
In the description provided herein, numerous specific details are set forth. However, it is understood that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures and techniques have not been shown in detail in order not to obscure an understanding of this description.
Similarly, it should be appreciated that in the foregoing description of exemplary embodiments of the invention, various features of the invention are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of streamlining the disclosure and aiding in the understanding of one or more of the various inventive aspects. However, the disclosed method should not be construed as reflecting the intention that: i.e., the claimed invention requires more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive aspects lie in less than all features of a single foregoing disclosed embodiment. Thus, the claims following the detailed description are hereby expressly incorporated into this detailed description, with each claim standing on its own as a separate embodiment of this invention.
Those skilled in the art will appreciate that the modules in the apparatus of the embodiments may be adaptively changed and disposed in one or more apparatuses different from the embodiments. The modules or units or components of the embodiments may be combined into one module or unit or component and, furthermore, they may be divided into a plurality of sub-modules or sub-units or sub-components. Any combination of all features disclosed in this specification (including any accompanying claims, abstract and drawings), and all of the processes or units of any method or apparatus so disclosed, may be used in combination, except insofar as at least some of such features and/or processes or units are mutually exclusive. Each feature disclosed in this specification (including any accompanying claims, abstract and drawings), may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise.
Furthermore, those skilled in the art will appreciate that while some embodiments herein include some features but not others included in other embodiments, combinations of features of different embodiments are meant to be within the scope of the invention and form different embodiments. For example, in the following claims, any of the claimed embodiments can be used in any combination.
Various component embodiments of the invention may be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art will appreciate that some or all of the functions of some or all of the components in a rich media advertisement making apparatus according to an embodiment of the present invention may be implemented in practice using a microprocessor or Digital Signal Processor (DSP). The present invention can also be implemented as an apparatus or device program (e.g., a computer program and a computer program product) for performing a portion or all of the methods described herein. Such a program embodying the present invention may be stored on a computer readable medium, or may have the form of one or more signals. Such signals may be downloaded from an internet website, provided on a carrier signal, or provided in any other form.
It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and that those skilled in the art will be able to design alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention may be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by one and the same item of hardware. The use of the words first, second, third, etc. do not denote any order. These words may be interpreted as names.
The invention discloses an advertisement layout method A1, which comprises the following steps:
based on element characteristics of advertisement elements and element characteristics formed by interaction of more than two advertisement elements, an initial layout model is built together, wherein advertisement elements laid out in a plurality of advertisement layout areas form banner advertisements;
Training the initial layout model by utilizing a primary size advertisement template sample to obtain a banner advertisement layout model;
and generating a banner advertisement layout scheme by using the advertisement layout model.
A2, the method of A1, wherein the establishing an initial layout model based on the element features of the advertisement elements and the element features formed by interaction of more than two advertisement elements includes:
abstracting each advertisement element used for layout in the advertisement layout area into a corresponding rectangular area;
and performing mathematical modeling by taking the position coordinates and the size information of each rectangular area as solutions to obtain the initial layout model.
A3, the method of A1 or A2, wherein training the initial layout model by using the original size advertisement template sample to obtain a banner advertisement layout model comprises the following steps:
constructing an advertisement layout measurement function aiming at the initial layout model;
and training the initial layout model by using the original size advertisement template sample until the energy function value of the advertisement layout measurement function is minimum, so as to obtain the banner advertisement layout model.
A4, the method as in A3, wherein the energy function value of the advertisement layout measurement function is a total energy value obtained by weighting and summing a plurality of energy values related to the layout quality.
A5, the method of A4, wherein training the initial layout model by using the original size advertisement template sample until the energy function value of the advertisement layout metric function is minimum, so as to obtain the banner advertisement layout model, comprises:
initializing specific parameters of the advertisement layout measurement function, and putting the original-size advertisement template sample into a candidate solution;
taking the solution with the minimum total energy value of the advertisement layout measurement function in the candidate solutions as the current optimal solution, and carrying out reverse optimization processing based on the current optimal solution to obtain a new solution;
updating the specific parameters based on a gradient descent method and judging whether a training termination condition is reached;
if the training termination condition is met, outputting the current optimal solution as the banner advertisement layout model; otherwise, updating the candidate solution based on the new solution, and returning to the step of taking the solution which minimizes the total energy value of the advertisement layout metric function in the candidate solution as the current optimal solution based on the updated candidate solution and the updated specific parameter.
The method of A6, the determining whether the training termination condition is reached, including:
judging whether the current function value of the objective function reaches a preset function value, if so, determining that the training termination condition is reached; the objective function is used for representing the difference between the total energy value of the advertisement layout template corresponding to the current optimal solution and the total energy value of the optimal layout scheme generated through an optimization algorithm.
A7, the method as described in A6, wherein the step of generating the optimal layout scheme through the optimization algorithm comprises the following steps:
performing annealing operation for more than one time based on the simulated annealing algorithm to generate advertisement layout proposals with preset quantity;
and determining the optimal layout scheme from the advertisement layout proposals with the preset quantity.
A8, the method of A7, wherein the simulated annealing algorithm comprises a plurality of annealing operators related to the advertisement elements; in the step of generating a preset number of advertisement layout proposals based on performing the simulated annealing operation more than once, each advertisement layout proposal is generated by:
selecting one annealing operator from the annealing operators related to the advertisement elements aiming at the current proposal generating operation in the current annealing operation, wherein the current annealing operation comprises the proposal generating operation with preset times;
and executing the current proposal generating operation by using a simulated annealing algorithm containing the annealing operator selected at the current time to generate a corresponding advertisement layout proposal.
A9, the method of any one of A1-A8, after the generating a banner advertisement layout scheme using the advertisement layout model, further comprises:
Performing size changing treatment on the original size advertisement template sample to generate a new size advertisement template sample;
training the initial layout model based on the original size advertisement template sample and the new size advertisement template sample together to train a variable size layout model;
and generating a variable-size advertisement layout scheme by using the variable-size layout model.
The invention discloses a B10, an advertisement layout device, comprising:
the modeling unit is used for jointly establishing an initial layout model based on element characteristics of the advertisement elements and element characteristics formed by interaction of more than two advertisement elements, wherein the advertisement elements laid out in a plurality of advertisement layout areas form banner advertisements;
the first model training unit is used for training the initial layout model by utilizing the original size advertisement template sample to obtain a banner advertisement layout model;
and the first advertisement layout unit is used for generating a banner advertisement layout scheme by utilizing the advertisement layout model.
B11, the apparatus of B10, the modeling unit comprising:
an abstraction subunit, configured to abstract each advertisement element used for layout in the advertisement layout area into a corresponding rectangular area;
And the mathematical modeling sub-unit is used for performing mathematical modeling by taking the position coordinates and the size information of each rectangular area as solutions to obtain the initial layout model.
B12, the apparatus of B10 or B11, the first model training unit comprising:
a function construction subunit, configured to construct an advertisement layout metric function for the initial layout model;
and the sample training subunit is used for training the initial layout model by using the original-size advertisement template sample until the energy function value of the advertisement layout measurement function is minimum so as to obtain the banner advertisement layout model.
B13, the apparatus of B12, wherein the energy function value of the advertisement placement metric function is a total energy value obtained by weighted summation of a plurality of energy values related to placement quality.
B14, the apparatus of B13, the sample training subunit being specifically configured to:
initializing specific parameters of the advertisement layout measurement function, and putting the original-size advertisement template sample into a candidate solution;
taking the solution with the minimum total energy value of the advertisement layout measurement function in the candidate solutions as the current optimal solution, and carrying out reverse optimization processing based on the current optimal solution to obtain a new solution;
Updating the specific parameters based on a gradient descent method and judging whether a training termination condition is reached;
if the training termination condition is met, outputting the current optimal solution as the banner advertisement layout model; otherwise, updating the candidate solution based on the new solution, and returning to the step of taking the solution which minimizes the total energy value of the advertisement layout metric function in the candidate solution as the current optimal solution based on the updated candidate solution and the updated specific parameter.
B15, the apparatus of B14, the sample training subunit being specifically configured to:
judging whether the current function value of the objective function reaches a preset function value, if so, determining that the training termination condition is reached; the objective function is used for representing the difference between the total energy value of the advertisement layout template corresponding to the current optimal solution and the total energy value of the optimal layout scheme generated through an optimization algorithm.
B16, the apparatus of B15, wherein the sample training subunit is specifically configured to:
performing annealing operation for more than one time based on the simulated annealing algorithm to generate advertisement layout proposals with preset quantity;
And determining the optimal layout scheme from the advertisement layout proposals with the preset quantity.
B17, the apparatus of B16, the simulated annealing algorithm comprising a plurality of annealing operators associated with advertising elements;
the sample training subunit is configured to, in the step of generating a preset number of advertisement layout proposals based on performing the annealing operation more than once based on the simulated annealing algorithm, generate each advertisement layout proposal by:
selecting one annealing operator from the annealing operators related to the advertisement elements aiming at the current proposal generating operation in the current annealing operation, wherein the current annealing operation comprises the proposal generating operation with preset times;
and executing the current proposal generating operation by using a simulated annealing algorithm containing the annealing operator selected at the current time to generate a corresponding advertisement layout proposal.
B18, the apparatus of any one of B10-B17, the apparatus further comprising:
the template processing unit is used for carrying out size changing processing on the original-size advertisement template sample to generate a new-size advertisement template sample;
the second model training unit is used for training the initial layout model based on the original size advertisement template sample and the new size advertisement template sample together so as to train a variable size layout model;
And a second advertisement layout unit for generating a variable-size advertisement layout scheme using the variable-size layout model.
The invention discloses a computer device, comprising a memory, a processor and a computer program stored on the memory and running on the processor, wherein the processor realizes the steps of the method of any one of A1-A9 when executing the program.
The invention discloses D20, a computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of the method of any of A1-A9.

Claims (14)

1. An advertising placement method, comprising:
based on element characteristics of advertisement elements and element characteristics formed by interaction of more than two advertisement elements, an initial layout model is built together, wherein advertisement elements laid out in a plurality of advertisement layout areas form banner advertisements;
training the initial layout model by utilizing a primary size advertisement template sample to obtain a banner advertisement layout model;
generating a banner advertisement layout scheme using the advertisement layout model;
training the initial layout model by using the original-size advertisement template sample to obtain a banner advertisement layout model, wherein the training comprises the following steps:
Constructing an advertisement layout measurement function aiming at the initial layout model;
training the initial layout model by using the original size advertisement template sample until the energy function value of the advertisement layout measurement function is minimum, so as to obtain the banner advertisement layout model;
the energy function value of the advertisement layout measurement function is a total energy value obtained by carrying out weighted summation on a plurality of energy values related to layout quality;
training the initial layout model using the original size advertisement template sample until an energy function value of the advertisement layout metric function is minimum to obtain the banner advertisement layout model, comprising:
initializing specific parameters of the advertisement layout measurement function, and putting the original-size advertisement template sample into a candidate solution;
taking the solution with the minimum total energy value of the advertisement layout measurement function in the candidate solutions as the current optimal solution, and carrying out reverse optimization processing based on the current optimal solution to obtain a new solution;
updating the specific parameters based on a gradient descent method and judging whether a training termination condition is reached;
if the training termination condition is met, outputting the current optimal solution as the banner advertisement layout model; otherwise, updating the candidate solution based on the new solution, and returning to the step of taking the solution which minimizes the total energy value of the advertisement layout metric function in the candidate solution as the current optimal solution based on the updated candidate solution and the updated specific parameter.
2. The method of claim 1, wherein the jointly establishing an initial layout model based on element features of the advertisement element and element features of the interaction of two or more advertisement elements comprises:
abstracting each advertisement element used for layout in the advertisement layout area into a corresponding rectangular area;
and performing mathematical modeling by taking the position coordinates and the size information of each rectangular area as solutions to obtain the initial layout model.
3. The method of claim 1, wherein said determining whether a training termination condition is reached comprises:
judging whether the current function value of the objective function reaches a preset function value, if so, determining that the training termination condition is reached; the objective function is used for representing the difference between the total energy value of the advertisement layout template corresponding to the current optimal solution and the total energy value of the optimal layout scheme generated through an optimization algorithm.
4. The method of claim 3, wherein the step of generating the optimal layout scheme by an optimization algorithm comprises:
performing annealing operation for more than one time based on the simulated annealing algorithm to generate advertisement layout proposals with preset quantity;
And determining the optimal layout scheme from the advertisement layout proposals with the preset quantity.
5. The method of claim 4, wherein the simulated annealing algorithm comprises a plurality of annealing operators associated with advertising elements; in the step of generating a preset number of advertisement layout proposals based on performing the simulated annealing operation more than once, each advertisement layout proposal is generated by:
selecting one annealing operator from the annealing operators related to the advertisement elements aiming at the current proposal generating operation in the current annealing operation, wherein the current annealing operation comprises the proposal generating operation with preset times;
and executing the current proposal generating operation by using a simulated annealing algorithm containing the annealing operator selected at the current time to generate a corresponding advertisement layout proposal.
6. The method of any of claims 1-5, further comprising, after said generating a banner advertisement layout scheme utilizing the advertisement layout model:
performing size changing treatment on the original size advertisement template sample to generate a new size advertisement template sample;
training the initial layout model based on the original size advertisement template sample and the new size advertisement template sample together to train a variable size layout model;
And generating a variable-size advertisement layout scheme by using the variable-size layout model.
7. An advertising placement device, comprising:
the modeling unit is used for jointly establishing an initial layout model based on element characteristics of the advertisement elements and element characteristics formed by interaction of more than two advertisement elements, wherein the advertisement elements laid out in a plurality of advertisement layout areas form banner advertisements;
the first model training unit is used for training the initial layout model by utilizing the original size advertisement template sample to obtain a banner advertisement layout model;
a first advertisement layout unit for generating a banner advertisement layout scheme using the advertisement layout model;
the first model training unit includes:
a function construction subunit, configured to construct an advertisement layout metric function for the initial layout model;
a sample training subunit, configured to train the initial layout model using the original-size advertisement template sample until an energy function value of the advertisement layout metric function is minimum, so as to obtain the banner advertisement layout model;
the energy function value of the advertisement layout measurement function is a total energy value obtained by carrying out weighted summation on a plurality of energy values related to layout quality;
The sample training subunit is specifically configured to:
initializing specific parameters of the advertisement layout measurement function, and putting the original-size advertisement template sample into a candidate solution;
taking the solution with the minimum total energy value of the advertisement layout measurement function in the candidate solutions as the current optimal solution, and carrying out reverse optimization processing based on the current optimal solution to obtain a new solution;
updating the specific parameters based on a gradient descent method and judging whether a training termination condition is reached;
if the training termination condition is met, outputting the current optimal solution as the banner advertisement layout model; otherwise, updating the candidate solution based on the new solution, and returning to the step of taking the solution which minimizes the total energy value of the advertisement layout metric function in the candidate solution as the current optimal solution based on the updated candidate solution and the updated specific parameter.
8. The apparatus of claim 7, wherein the modeling unit comprises:
an abstraction subunit, configured to abstract each advertisement element used for layout in the advertisement layout area into a corresponding rectangular area;
and the mathematical modeling sub-unit is used for performing mathematical modeling by taking the position coordinates and the size information of each rectangular area as solutions to obtain the initial layout model.
9. The apparatus of claim 7, wherein the sample training subunit is configured to:
judging whether the current function value of the objective function reaches a preset function value, if so, determining that the training termination condition is reached; the objective function is used for representing the difference between the total energy value of the advertisement layout template corresponding to the current optimal solution and the total energy value of the optimal layout scheme generated through an optimization algorithm.
10. The apparatus of claim 9, wherein the sample training subunit is configured, in the step of generating the optimal layout scheme by an optimization algorithm, to:
performing annealing operation for more than one time based on the simulated annealing algorithm to generate advertisement layout proposals with preset quantity;
and determining the optimal layout scheme from the advertisement layout proposals with the preset quantity.
11. The apparatus of claim 10, wherein the simulated annealing algorithm comprises a plurality of annealing operators associated with advertisement elements;
the sample training subunit is configured to, in the step of generating a preset number of advertisement layout proposals based on performing the annealing operation more than once based on the simulated annealing algorithm, generate each advertisement layout proposal by:
Selecting one annealing operator from the annealing operators related to the advertisement elements aiming at the current proposal generating operation in the current annealing operation, wherein the current annealing operation comprises the proposal generating operation with preset times;
and executing the current proposal generating operation by using a simulated annealing algorithm containing the annealing operator selected at the current time to generate a corresponding advertisement layout proposal.
12. The apparatus according to any one of claims 7-11, wherein the apparatus further comprises:
the template processing unit is used for carrying out size changing processing on the original-size advertisement template sample to generate a new-size advertisement template sample;
the second model training unit is used for training the initial layout model based on the original size advertisement template sample and the new size advertisement template sample together so as to train a variable size layout model;
and a second advertisement layout unit for generating a variable-size advertisement layout scheme using the variable-size layout model.
13. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the steps of the method of any of claims 1-6 when the program is executed.
14. A computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of the method of any of claims 1-6.
CN201811133550.5A 2018-09-27 2018-09-27 Advertisement layout method and device and computer equipment Active CN109271604B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811133550.5A CN109271604B (en) 2018-09-27 2018-09-27 Advertisement layout method and device and computer equipment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811133550.5A CN109271604B (en) 2018-09-27 2018-09-27 Advertisement layout method and device and computer equipment

Publications (2)

Publication Number Publication Date
CN109271604A CN109271604A (en) 2019-01-25
CN109271604B true CN109271604B (en) 2023-05-23

Family

ID=65199059

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811133550.5A Active CN109271604B (en) 2018-09-27 2018-09-27 Advertisement layout method and device and computer equipment

Country Status (1)

Country Link
CN (1) CN109271604B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112116681B (en) * 2019-06-19 2023-07-25 腾讯科技（深圳）有限公司 Image generation method, device, computer equipment and storage medium
CN110378432B (en) * 2019-07-24 2022-04-12 阿里巴巴（中国）有限公司 Picture generation method, device, medium and electronic equipment
CN110737963B (en) * 2019-12-20 2020-03-31 广东博智林机器人有限公司 Poster element layout method, system and computer readable storage medium
CN111353822A (en) * 2020-03-03 2020-06-30 广东博智林机器人有限公司 Image layout and model training method, device, equipment and storage medium

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103631865A (en) * 2013-11-01 2014-03-12 北京奇虎科技有限公司 Webpage generating method and device
CN105678581A (en) * 2016-01-01 2016-06-15 广州筷子信息科技有限公司 Method and system for automatic typesetting of advertisement
CN106485534A (en) * 2015-08-25 2017-03-08 阿里巴巴集团控股有限公司 The generation method of advertisement and device in webpage
CN107330715A (en) * 2017-05-31 2017-11-07 北京京东尚科信息技术有限公司 The method and apparatus for selecting display advertising material

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070255616A1 (en) * 2006-04-27 2007-11-01 Microsoft Corporation Techniques for authoring ads for dynamic layout environments

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103631865A (en) * 2013-11-01 2014-03-12 北京奇虎科技有限公司 Webpage generating method and device
CN106485534A (en) * 2015-08-25 2017-03-08 阿里巴巴集团控股有限公司 The generation method of advertisement and device in webpage
CN105678581A (en) * 2016-01-01 2016-06-15 广州筷子信息科技有限公司 Method and system for automatic typesetting of advertisement
CN107330715A (en) * 2017-05-31 2017-11-07 北京京东尚科信息技术有限公司 The method and apparatus for selecting display advertising material

Also Published As

Publication number Publication date
CN109271604A (en) 2019-01-25

Similar Documents

Publication Publication Date Title
CN109271604B (en) Advertisement layout method and device and computer equipment
CN102667811B (en) Alignment of objects in augmented reality
AU2016349518B2 (en) Edge-aware bilateral image processing
US20200202601A1 (en) Predicting Patch Displacement Maps Using A Neural Network
WO2019154035A1 (en) Method for implanting advertisements in video, and computer device
US20210409907A1 (en) Methods and devices for displaying a heat map and providing heat data
CN108076154A (en) Application message recommends method, apparatus and storage medium and server
CN108898185A (en) Method and apparatus for generating image recognition model
CN109255356A (en) A kind of character recognition method, device and computer readable storage medium
CN111310746B (en) Text line detection method, model training method, device, server and medium
US11887224B2 (en) Method, apparatus, and computer program for completing painting of image, and method, apparatus, and computer program for training artificial neural network
US11830110B2 (en) Authoring and optimization of accessible color themes
KR20200099271A (en) Method that provides and creates mosaic image based on image tag-word
CN111931782A (en) Semantic segmentation method, system, medium, and apparatus
CN112364916B (en) Image classification method based on transfer learning, related equipment and storage medium
CN107918936B (en) High frequency offset using tag tracking for block matching algorithms
CN113392702A (en) Target identification method based on self-adaptive image enhancement under low-light environment
JP6800901B2 (en) Object area identification device, object area identification method and program
CN111126493A (en) Deep learning model training method and device, electronic equipment and storage medium
CN116152368A (en) Font generation method, training method, device and equipment of font generation model
Ko et al. Superpixel Based ImageCut Using Object Detection
Luo et al. Illuminant estimation using pixels spatially close to the illuminant in the rg-chromaticity space
CN113902001A (en) Model training method and device, electronic equipment and storage medium
CN117011324A (en) Image processing method, device, electronic equipment and storage medium
KR20230036877A (en) system and method for matching real estate agent with client for real estate transaction

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
TA01 Transfer of patent application right
TA01 Transfer of patent application right

Effective date of registration: 20230505

Address after: 300450 No. 9-3-401, No. 39, Gaoxin 6th Road, Binhai Science Park, Binhai New Area, Tianjin

Applicant after: 3600 Technology Group Co.,Ltd.

Address before: 100088 room 112, block D, 28 new street, new street, Xicheng District, Beijing (Desheng Park)

Applicant before: BEIJING QIHOO TECHNOLOGY Co.,Ltd.

GR01 Patent grant
GR01 Patent grant