CN107330715B - Method and device for selecting picture advertisement material - Google Patents
Method and device for selecting picture advertisement material Download PDFInfo
- Publication number
- CN107330715B CN107330715B CN201710400244.2A CN201710400244A CN107330715B CN 107330715 B CN107330715 B CN 107330715B CN 201710400244 A CN201710400244 A CN 201710400244A CN 107330715 B CN107330715 B CN 107330715B
- Authority
- CN
- China
- Prior art keywords
- advertisement
- score
- convolutional
- neural network
- picture
- 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
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
- G06Q30/0276—Advertisement creation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Finance (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Biophysics (AREA)
- General Business, Economics & Management (AREA)
- Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Marketing (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The embodiment of the invention provides a method and a device for selecting picture advertisement materials, which are beneficial to overcoming various defects in the aspect of selecting the picture advertisement materials in the prior art. The method comprises the following steps: storing a plurality of training data, wherein each training data comprises picture advertisement materials of the advertisement and a score when the advertisement is visited once, and the positive and negative directions of the score are determined according to whether the advertisement is converted after being visited this time; training the training data by using a convolutional neural network to obtain a model, wherein an input layer of the convolutional neural network is obtained according to the pixel value of the picture advertisement material, a convolutional kernel in the convolutional layer is preset, and the weight of each edge in the convolutional neural network and the convolutional kernel are updated according to the difference between the score output by the model and the score during training; and calculating the score of the specified picture advertisement material by using the model, and determining whether to select the specified picture advertisement material according to the score.
Description
Technical Field
The invention relates to the technical field of computers and software, in particular to a method and a device for selecting picture advertisement materials.
Background
The material (usually, pictures) is the basic material of the advertisement system, and the quality of the material directly determines the comprehensive benefits (clicks, conversion, etc. brought by the advertisement) of the advertisement platform and the advertiser. Therefore, reasonable selection of materials to be delivered by an advertiser is a very critical ring in an advertisement system, and optimization of material selection of the picture advertisement is expected, so that the materials are picture types hereinafter.
Advertising platforms such as Facebook, google, hundredth, Tencent, etc. in the industry provide some suggestions for optimizing picture advertising materials, such as: material definition, size, template reference, etc.
Currently, the common method for advertisers to select materials is as follows: and creating a plurality of groups of materials, firstly putting a certain budget for each material, and then putting the materials into an advertisement playing system. After certain data is accumulated, the effect (such as click rate, conversion rate, total amount of commodity transaction GMV and the like) of each material is observed, and the release of certain materials is suspended or increased according to the effect.
One process by which an advertiser may refine material is shown in fig. 1, with fig. 1 being a schematic illustration of one process for selecting advertising material according to the prior art. As shown in fig. 1, the process mainly includes the following steps:
(1) the advertiser creates material;
(2) the material is put on line;
(3) the advertisement playing system performs effect feedback (for example, indexes such as actual exposure number, click number, order placing number, GMV, actual advertisement cost and the like of the advertisement);
(4) the advertiser pauses the delivery of the low quality material and augments the delivery of the high quality material based on the effect of the feedback.
In the process of implementing the invention, the following defects are found in the prior art:
the material optimization suggestions provided by the advertisement platform are usually more comprehensive and fuzzy, and the quality of the advertisement materials cannot be reasonably and quantitatively evaluated, so that the optimization suggestions have no guiding significance for selecting the materials to be delivered to inexperienced advertisers.
The material selection methods commonly used by advertisers also have significant drawbacks:
from the advertiser perspective: (1) the selection period of the material is long, the accumulated amount of the advertisement playing system for playing the material is relied on, and the quality of the material cannot be accurately measured when the playing amount is insufficient; (2) online launch testing of some low quality material can result in loss of benefit.
From an advertising system perspective: (1) the advertiser puts in a large amount of low-quality materials, which increases the advertisement retrieval burden of the advertisement playing system; (2) frequent on-off-line of the material can increase the instability of the playing effect of the advertising system.
Disclosure of Invention
In view of this, embodiments of the present invention provide a method and an apparatus for selecting a picture advertisement material, which are helpful for overcoming various defects in the prior art in the aspect of selecting the picture advertisement material.
In order to achieve the above purpose, according to the embodiment of the present invention, the following technical solutions are proposed:
a method of selecting pictorial advertising material, comprising: storing a plurality of training data, wherein each training data comprises picture advertisement materials of the advertisement and a score when the advertisement is visited once, and the positive and negative directions of the score are determined according to whether the advertisement is converted after being visited this time; training the training data by using a convolutional neural network to obtain a model, wherein an input layer of the convolutional neural network is obtained according to the pixel value of the picture advertisement material, a convolutional kernel in the convolutional layer is preset, and the weight of each edge in the convolutional neural network and the convolutional kernel are updated according to the difference between the score output by the model and the score during training; and calculating the score of the specified picture advertisement material by using the model, and determining whether to select the specified picture advertisement material according to the score.
Optionally, the convolutional neural network further comprises a fully-connected layer connected to the convolutional layer.
Optionally, the step of updating the weights of the edges in the convolutional neural network and the convolutional kernels includes: and calculating the weight of each edge in the convolutional neural network and the adjustment amplitude of the convolutional kernel by using a gradient descent method, and then changing the weight of each edge in the convolutional neural network and the convolutional kernel according to the adjustment amplitude.
An apparatus for selecting a picture advertising material, comprising: the training data module is used for storing a plurality of training data, each training data comprises picture advertisement materials of the advertisement and a score when the advertisement is visited once, wherein the positive and negative directions of the score are determined according to whether the advertisement is converted after being visited this time; the model training module is used for training the training data by using a convolutional neural network to obtain a model, an input layer of the convolutional neural network is obtained according to the pixel value of the picture advertisement material, a convolutional kernel in the convolutional layer is preset, and the weight of each edge in the convolutional neural network and the convolutional kernel are updated according to the difference between the score output by the model and the score during training; and the material selection module is used for calculating the score of the specified picture advertisement material by using the model and determining whether to select the specified picture advertisement material according to the score.
Optionally, the convolutional neural network further comprises a fully-connected layer connected to the convolutional layer.
Optionally, the model training module is further configured to calculate a weight of each edge in the convolutional neural network and an adjustment magnitude of the convolutional kernel by using a gradient descent method, and then change the weight of each edge in the convolutional neural network and the convolutional kernel by the adjustment magnitude.
An apparatus for selecting a photo advertisement material, comprising an offline module and an online module, wherein: the offline module is to: storing a plurality of training data, wherein each training data comprises picture advertisement materials of the advertisement and a score when the advertisement is visited once, and the positive and negative directions of the score are determined according to whether the advertisement is converted after being visited this time; training the training data by using a convolutional neural network to obtain a model, wherein an input layer of the convolutional neural network is obtained according to the pixel value of the picture advertisement material, a convolutional kernel in the convolutional layer is preset, and the weight of each edge in the convolutional neural network and the convolutional kernel are updated according to the difference between the score output by the model and the score during training; the online module is used for: receiving a plurality of picture advertisement materials, calculating the score of each received picture advertisement material by using the model, selecting one or more picture advertisement materials according to the score, and then putting the selected picture advertisement materials into an advertisement library.
Optionally, the offline module is further configured to obtain a click/conversion history log of the advertisement from the advertisement playing system.
An electronic device, comprising: one or more processors; a storage device for storing one or more programs which, when executed by the one or more processors, cause the one or more processors to implement the method according to an embodiment of the present invention.
A computer-readable medium, on which a computer program is stored which, when being executed by a processor, carries out the method according to an embodiment of the invention.
According to the technical scheme of the embodiment of the invention, the model is obtained by using historical data training, and the model realizes a quality evaluation scheme for picture materials, so that the period of selection of the materials of an advertiser is shortened, and the dependence on an online playing module of a playing system is completely avoided; the advertiser does not need to put low-quality materials on line, so that the meaningless loss of benefits is avoided; and effective guarantee is brought for the effect stability of advertisement platform.
Drawings
The drawings are included to provide a better understanding of the invention and are not to be construed as unduly limiting the invention. Wherein:
FIG. 1 is a schematic illustration of a process for selecting advertising material according to the prior art;
FIG. 2 is a schematic diagram of the basic steps of a method of selecting pictorial advertising material in accordance with an embodiment of the present invention;
fig. 3 is a schematic diagram of one configuration of an apparatus for selecting a picture advertisement material according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of a system relating to advertisement placement, according to an embodiment of the present invention;
FIG. 5 is a schematic diagram of a task flow of an offline module according to an embodiment of the present invention;
FIG. 6 is a schematic diagram of one configuration of a convolutional neural network, according to an embodiment of the present invention;
FIG. 7 is a schematic diagram of a process of convolution calculation according to an embodiment of the present invention;
FIG. 8 is a schematic diagram of the manner in which an online module operates according to an embodiment of the present invention;
fig. 9 illustrates an exemplary system architecture of a method or apparatus for selecting photo advertising material to which an embodiment of the present invention may be applied;
fig. 10 is a hardware configuration diagram of an electronic device of a method of selecting a picture advertisement material according to an embodiment of the present invention.
Detailed Description
Exemplary embodiments of the present invention are described below with reference to the accompanying drawings, in which various details of embodiments of the invention are included to assist understanding, and which are to be considered exemplary only. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the invention. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.
In the embodiment of the invention, training data is formed based on the advertisement picture material and the click/conversion history log of the advertisement in the advertisement playing system, and the training data is used for training by utilizing a machine learning technology to obtain a model, wherein the model can score the picture advertisement material, and the higher the score is, the more possible the advertisement is converted when adopting the picture advertisement material. The basic steps of the method are shown in fig. 2, and fig. 2 is a schematic diagram of the basic steps of a method for selecting a picture advertisement material according to an embodiment of the present invention.
Step S21: a plurality of training data items are stored. Each training datum includes the picture advertisement material of the advertisement and the score of the advertisement when accessed once, wherein the positive and negative directions of the score are determined according to whether the advertisement is converted after being accessed. For example, if there is a conversion, the score is positive, and vice versa, the score is negative.
Step S22: the training data is trained using a convolutional neural network to obtain a model. The input layer of the convolutional neural network is obtained according to the pixel value of the picture advertisement material, the convolutional kernel in the convolutional layer is preset, and the weight of each edge in the convolutional neural network and the convolutional kernel are updated according to the difference between the score output by the model and the score during training.
The convolutional neural network may further include a fully-connected layer connected to the convolutional layer. When the weights of the edges and the convolution kernels in the convolutional neural network are updated, the weights of the edges and the adjustment amplitude of the convolution kernels in the convolutional neural network can be calculated by using a gradient descent method, and then the weights of the edges and the convolution kernels in the convolutional neural network are changed according to the adjustment amplitude.
Step S23: and calculating the score of the specified picture advertisement material by using the model, and determining whether to select the specified picture advertisement material according to the score.
The method can be implemented by using a computer, one architecture of the software is shown in fig. 3, and fig. 3 is a schematic diagram of a structure of the apparatus for selecting the picture advertisement material according to the embodiment of the present invention. In fig. 3, the apparatus 3 for selecting a picture advertisement material mainly includes a training data module, a model training module, and a material selection module.
The training data module is used for storing a plurality of training data, each training data comprises picture advertisement materials of the advertisement and a score when the advertisement is visited once, wherein the positive and negative directions of the score are determined according to whether the advertisement is converted after being visited this time. The model training module is used for training the training data by using a convolutional neural network to obtain a model, an input layer of the convolutional neural network is obtained according to the pixel value of the picture advertisement material, a convolutional kernel in the convolutional layer is preset, and the weight of each edge in the convolutional neural network and the convolutional kernel are updated according to the difference between the score output by the model and the score during training; the method can also be used for calculating the weight of each edge in the convolutional neural network and the adjustment amplitude of the convolutional kernel by using a gradient descent method, and then changing the weight of each edge in the convolutional neural network and the convolutional kernel according to the adjustment amplitude. The material selection module is used for calculating the score of the specified picture advertisement material by using the model and determining whether to select the specified picture advertisement material according to the score.
In a specific implementation, the system shown in fig. 4 can also be used, and fig. 4 is a schematic diagram of a system related to advertisement delivery according to an embodiment of the present invention, which includes existing materials 1 to N created by advertisers and an advertisement library, and according to an embodiment of the present invention, an offline module and an online module are added.
The off-line module is used for storing a plurality of training data, each training data comprises picture advertisement materials of the advertisement and a score when the advertisement is accessed once, wherein the positive and negative directions of the score are determined according to whether the advertisement is converted after being accessed this time; and training the training data by using a convolutional neural network to obtain a model, wherein an input layer of the convolutional neural network is obtained according to the pixel value of the picture advertisement material, a convolutional kernel in the convolutional layer is preset, and the weight of each edge in the convolutional neural network and the convolutional kernel are updated according to the difference between the score output by the model and the score during training. The online module is used for receiving a plurality of picture advertisement materials, then calculating scores of the received picture advertisement materials by using the model, selecting one or a plurality of picture advertisement materials according to the scores, and then putting the selected picture advertisement materials into an advertisement library.
It can therefore be seen that the main function of the offline module is to train to obtain a model (hereinafter referred to as a conversion model), while the main function of the online module is to use the model for material evaluation. In the embodiment of the present invention, first, the advertisement platform provides the above-mentioned device for selecting the picture advertisement material. Because the advertiser is more concerned about the conversion effect (such as ordering, purchasing and the like) of the advertisement, the evaluation of the material quality can be realized based on the conversion rate estimation technology; secondly, the advertiser creates a material, and the quality of the material is evaluated by using a material quality evaluation system. The grade represents the quality of the material, and the grade value can approximate the conversion rate brought by the material; and finally, the advertiser selects high-quality materials according to the material scores, the screened materials are put on line (enter an advertisement library), and finally, the advertisement playing system plays the materials and records logs.
Task flow of the offline module is shown in fig. 5, and fig. 5 is a schematic diagram of task flow of the offline module according to an embodiment of the present invention. According to fig. 5, the task flow of the offline module mainly includes: (1) the advertisement system accumulates logs; (2) extracting training data, and extracting information such as < material, whether to convert after clicking > and the like from a historical log (log of the last 30 days); (3) and automatically extracting and learning the features of the picture materials by using a Convolutional Neural Network (CNN) of a machine learning technology, carrying out model training, and generating an offline model capable of representing material conversion rate information. When the model is trained, a network structure (such as the number of network layers, network nodes of each layer, the size of a convolution kernel, an activation function and the like) is preset, and then the weights of the convolution kernel and the edges in the network are learned by utilizing a machine learning technology; (4) we again use the historical logs (2-day logs not participating in model training) to evaluate the model, and when the model reaches the expected standard, the model is output and pushed to the online module for online evaluation.
The structure of the CNN network used in the embodiment of the present invention is shown in fig. 6, and fig. 6 is a schematic diagram of one structure of the convolutional neural network according to the embodiment of the present invention. As shown in fig. 6, pixel values of the R channel, the G channel, and the B channel of the picture advertisement material are normalized to a value range [0,1] and then enter the CNN network, where the CNN network includes: input layer, convolution layer, pooling layer, full-link layer, and output layer. The convolutional layer and the pooling layer firstly learn the local spatial structure in the input image, then join the local information to the full-link layer, and the full-link layer learns more abstract global information containing the whole image. Thus, CNN networks have the ability to automatically mine features without the need for manual attempts at features. The details are as follows:
an input layer: RGB channels extracted from the picture material.
And (3) rolling layers: the input is derived from an input layer or pooling layer. The implementation principle of the convolution layer is as follows: firstly, selecting a channel of an input layer (or a convolutional layer) and a corresponding convolutional kernel (the value is generally updated gradually by training after being initialized randomly), then carrying out convolutional calculation on the convolutional kernel and any region (the size of which is consistent with that of the convolutional kernel) of the selected channel, and generating a convolutional result after all the calculation is finished. Referring to fig. 7, fig. 7 is a schematic diagram of a process of convolution calculation according to an embodiment of the present invention. As shown in fig. 7, a 4 × 4 region M' of the input layer is convolved with the convolution kernel to obtain the convolution layer result at the right end of the equation. For simplicity, the normalized value of a pixel in the figure is taken to be either 0 or 1.
The convolution result is calculated for each channel, and then the convolution results under a plurality of channels are superposed to generate a final convolution layer which is used as the input of the next layer of network.
And (3) rolling layers: in general, local pixels in a picture are closely related, and distant pixels are less correlated. Therefore, each neuron in the network only needs to sense the local part, and then the local information is integrated by the neurons at higher layers to obtain the global information.
A pooling layer: the main purpose of the pooling layer is to reduce the dimension of the network following the human visual system, e.g. the mean (or maximum) of the features over an area of the image can be calculated and this mean of the features used to replace all features in this area. These summary statistical features not only can greatly reduce feature dimensionality (compared to using all extracted features), but also improve model results (not easily overfit).
Full connection layer: and the full-connection layer further learns the high-level combination characteristics through the multilayer full-connection neural network according to the high-dimensional space image characteristics extracted by the convolution pooling layer, so that a final model inference result is finally obtained. Unlike the local connection and weight sharing of convolutional layers, the connection is maintained between each input node and each output node of the fully-connected layer, so that the position information of some local features is discarded, and a comprehensive model inference result is given from the global perspective.
An output layer: and (5) scoring the quality of the picture material.
In the training process of the conversion rate model, firstly, a CNN network structure is manually set, for example: 1 input layer (3 channels), a first convolutional layer (2 channels, and 6 convolutional kernels), a pooling layer (2 channels), a second convolutional layer (4 channels, 8 convolutional kernels), 1 full-link (4 nodes). The number of convolution kernels on a convolution layer is determined by input channels and output channels, for example, 6 convolution kernels of a first convolution layer are 3 input layer channels × 2 output channels.
The training process for this CNN network is then as follows: (1) training data is obtained, and the training data is composed of information such as < material, whether to convert after clicking > and the like. Wherein, after clicking, the conversion represents positive scoring (the value is 1), and after clicking, the conversion does not occur, which represents negative scoring (the value is 0); (2) regarding training data, taking the pixels of the picture as the input of a CNN network, performing operations such as convolution, pooling and the like, then accessing a full-connection layer, and finally outputting a score through an output layer; (3) comparing the output score (with the value range of [0,1]) with the real conversion behavior (with the conversion score of 1 and the non-conversion score of 0), and updating the weight of each edge in the network according to the difference between the pre-estimated value and the real conversion behavior; (4) the above process is repeated until the estimated value and the real value are within the expected range. (5) Model effect assessment using data not trained.
Next, a new CNN network structure may be tried, for example, with respect to the above manually set CNN network structure, the convolutional layers are reduced from two to one, and the fully-connected layers are changed from 4 nodes to 8 nodes, and then the above steps (1) to (5) are repeated, thereby obtaining another conversion rate model. In the case of continuously trying new CNN network structures, multiple conversion models can be obtained, from which the most effective conversion model can then be selected.
FIG. 8 is a schematic diagram of the manner in which an online module operates according to an embodiment of the present invention. In light of the foregoing, the primary function of the online module is to provide real-time material quality assessment services for the advertising system. After receiving a material scoring request submitted by an advertiser, the evaluation system can utilize the conversion rate model to estimate the conversion rate of the material in real time and finally output an evaluation score, and the specific calculation process is completely the same as the calculation mode of the score in the training process.
Fig. 9 illustrates an exemplary system architecture 900 for a method or apparatus for selecting photo advertising material to which embodiments of the present invention may be applied.
As shown in fig. 9, the system architecture 900 may include end devices 901, 902, 903, a network 904, and a server 905. Network 904 is the medium used to provide communication links between terminal devices 901, 902, 903 and server 905. Network 904 may include various connection types, such as wired, wireless communication links, or fiber optic cables, to name a few.
A user may use the terminal devices 901, 902, 903 to interact with a server 905 over a network 904 to receive or send messages and the like. The terminal devices 901, 902, 903 may have various communication client applications installed thereon, such as a shopping application, a web browser application, a search application, an instant messaging tool, a mailbox client, social platform software, and the like.
The terminal devices 901, 902, 903 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop portable computers, desktop computers, and the like.
The server 905 may be a server that provides various services, such as a background management server that supports shopping websites browsed by users using the terminal devices 901, 902, and 903. The background management server can analyze and process the received data such as the product information inquiry request and feed back the processing result to the terminal equipment.
It should be noted that the method for selecting picture advertisement material provided by the embodiment of the present invention may be executed by a server or a terminal device, and accordingly, the device for selecting picture advertisement material may be disposed in the server or the terminal device.
It should be understood that the number of terminal devices, networks, and servers in fig. 9 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for implementation.
Fig. 10 is a schematic diagram of a hardware configuration of an electronic device according to an embodiment of the present invention, in which the method of selecting a picture advertisement material is implemented. Referring now to FIG. 10, a block diagram of a computer system 1000 suitable for use with a terminal device implementing an embodiment of the invention is shown. The terminal device shown in fig. 10 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiments of the present invention.
As shown in fig. 10, the computer system 1000 includes a Central Processing Unit (CPU) that can perform various appropriate actions and processes according to a program stored in a Read Only Memory (ROM) or a program loaded from a storage section into a Random Access Memory (RAM). In the RAM, various programs and data necessary for system operation are also stored. The CPU, ROM, and RAM are connected to each other via a bus. An input/output (I/O) interface is also connected to the bus.
The following components are connected to the I/O interface: an input section including a keyboard, a mouse, and the like; an output section including a display such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and the like, and a speaker; a storage section including a hard disk and the like; and a communication section including a network interface card such as a LAN card, a modem, or the like. The communication section performs communication processing via a network such as the internet. The drive is also connected to the I/O interface as needed. A removable medium such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is mounted on the drive as necessary, so that a computer program read out therefrom is mounted into the storage section as necessary.
In particular, according to the embodiments of the present disclosure, the processes described above with reference to the flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program embodied on a computer readable medium, the computer program comprising program code for performing the method illustrated in the flow chart. In such an embodiment, the computer program may be downloaded and installed from a network via the communication section, and/or installed from a removable medium. The computer program performs the above-described functions defined in the system of the present invention when executed by a Central Processing Unit (CPU).
It should be noted that the computer readable medium shown in the present invention can be a computer readable signal medium or a computer readable storage medium or any combination of the two. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples of the computer readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the present invention, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In the present invention, however, a computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, fiber optic cable, RF, etc., or any suitable combination of the foregoing.
The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams or flowchart illustration, and combinations of blocks in the block diagrams or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
The modules described in the embodiments of the present invention may be implemented by software or hardware. The described modules may also be provided in a processor, which may be described as: a processor comprises a training data module, a model training module and a material selection module. Where the names of the units do not in some cases constitute a limitation on the units themselves, for example, the material selection module may also be described as "a module for calculating a score of a specified picture advertising material using the model, and determining whether to select the specified picture advertising material based on the score".
According to the technical scheme of the embodiment of the invention, the model is obtained by using historical data training, and the model realizes a quality evaluation scheme for picture materials, so that the period of selection of the materials of an advertiser is shortened, and the dependence on an online playing module of a playing system is completely avoided; the advertiser does not need to put low-quality materials on line, so that the meaningless loss of benefits is avoided; and effective guarantee is brought for the effect stability of advertisement platform.
The above-described embodiments should not be construed as limiting the scope of the invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions can occur, depending on design requirements and other factors. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. A method of selecting a pictorial advertising material, comprising:
based on the offline state, saving a plurality of training data, wherein each training data comprises picture advertisement materials of the advertisement and a score when the advertisement is accessed once, and the positive and negative directions of the score are determined according to whether the advertisement is converted after being accessed this time;
training the training data by using a convolutional neural network to obtain a model, wherein an input layer of the convolutional neural network is obtained according to the pixel value of the picture advertisement material, a convolutional kernel in the convolutional layer is preset, and the weight of each edge in the convolutional neural network and the convolutional kernel are updated according to the difference between the score output by the model and the score during training;
calculating the score of the specified picture advertisement material by using the model, and determining whether to select the specified picture advertisement material according to the score;
based on the online state, a plurality of picture advertisement materials are received, then the scores of the received picture advertisement materials are calculated by using the model, one or more picture advertisement materials are selected according to the scores, and then the selected picture advertisement materials are put into an advertisement library.
2. The method of claim 1, wherein the convolutional neural network further comprises a fully-connected layer connected to the convolutional layer.
3. The method of claim 1 or 2, wherein the step of updating the weights of the edges in the convolutional neural network and the convolutional kernel comprises: and calculating the weight of each edge in the convolutional neural network and the adjustment amplitude of the convolutional kernel by using a gradient descent method, and then changing the weight of each edge in the convolutional neural network and the convolutional kernel according to the adjustment amplitude.
4. An apparatus for selecting a picture advertising material, comprising:
the training data module is used for storing a plurality of training data based on the off-line state, wherein each training data comprises picture advertisement materials of the advertisement and a score when the advertisement is accessed once, and the positive and negative directions of the score are determined according to whether the advertisement is converted after being accessed this time;
the model training module is used for training the training data by using a convolutional neural network to obtain a model, an input layer of the convolutional neural network is obtained according to the pixel value of the picture advertisement material, a convolutional kernel in the convolutional layer is preset, and the weight of each edge in the convolutional neural network and the convolutional kernel are updated according to the difference between the score output by the model and the score during training;
the material selection module is used for calculating the score of the specified picture advertisement material by using the model and determining whether to select the specified picture advertisement material according to the score; based on the online state, a plurality of picture advertisement materials are received, then the scores of the received picture advertisement materials are calculated by using the model, one or more picture advertisement materials are selected according to the scores, and then the selected picture advertisement materials are put into an advertisement library.
5. The apparatus for selecting graphic advertising material as recited in claim 4, wherein the convolutional neural network further comprises a fully connected layer connected to the convolutional layer.
6. The apparatus of claim 4, wherein the model training module is further configured to calculate the weights of the edges in the convolutional neural network and the adjustment amplitude of the convolutional kernel using a gradient descent method, and then change the weights of the edges in the convolutional neural network and the convolutional kernel by the adjustment amplitude.
7. An apparatus for selecting a picture advertising material, comprising an offline module and an online module, wherein:
the offline module is to: storing a plurality of training data, wherein each training data comprises picture advertisement materials of the advertisement and a score when the advertisement is visited once, and the positive and negative directions of the score are determined according to whether the advertisement is converted after being visited this time; training the training data by using a convolutional neural network to obtain a model, wherein an input layer of the convolutional neural network is obtained according to the pixel value of the picture advertisement material, a convolutional kernel in the convolutional layer is preset, and the weight of each edge in the convolutional neural network and the convolutional kernel are updated according to the difference between the score output by the model and the score during training;
the online module is used for: receiving a plurality of picture advertisement materials, calculating the score of each received picture advertisement material by using the model, selecting one or more picture advertisement materials according to the score, and then putting the selected picture advertisement materials into an advertisement library.
8. The apparatus for selecting graphic advertisement material as claimed in claim 7, wherein the offline module is further configured to obtain a click/conversion history log of the advertisement from the advertisement playing system.
9. An electronic device, comprising:
one or more processors;
a storage device for storing one or more programs,
when executed by the one or more processors, cause the one or more processors to implement the method of any one of claims 1-3.
10. A computer-readable medium, on which a computer program is stored, which, when being executed by a processor, carries out the method according to any one of claims 1-3.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710400244.2A CN107330715B (en) | 2017-05-31 | 2017-05-31 | Method and device for selecting picture advertisement material |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710400244.2A CN107330715B (en) | 2017-05-31 | 2017-05-31 | Method and device for selecting picture advertisement material |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107330715A CN107330715A (en) | 2017-11-07 |
| CN107330715B true CN107330715B (en) | 2020-12-22 |
Family
ID=60193536
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710400244.2A Active CN107330715B (en) | 2017-05-31 | 2017-05-31 | Method and device for selecting picture advertisement material |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107330715B (en) |
Families Citing this family (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109784129A (en) * | 2017-11-14 | 2019-05-21 | 北京京东尚科信息技术有限公司 | Information output method and device |
| CN110097179B (en) * | 2018-01-29 | 2020-03-10 | 上海寒武纪信息科技有限公司 | Computer device, data processing method, and storage medium |
| CN109978581A (en) * | 2017-12-28 | 2019-07-05 | 百度在线网络技术（北京）有限公司 | Information launches optimization method, device, computer equipment and storage medium |
| CN110555124A (en) * | 2018-06-01 | 2019-12-10 | 北京京东尚科信息技术有限公司 | Picture selection method, system, medium and electronic device |
| CN109271604B (en) * | 2018-09-27 | 2023-05-23 | 三六零科技集团有限公司 | Advertisement layout method and device and computer equipment |
| CN109242927B (en) * | 2018-09-29 | 2024-02-02 | 三六零科技集团有限公司 | Advertisement template generation method and device and computer equipment |
| CN111242654B (en) * | 2018-11-28 | 2021-04-30 | 北京沃东天骏信息技术有限公司 | Method and system for generating advertisement picture |
| CN109815465B (en) * | 2018-12-19 | 2023-11-17 | 平安科技（深圳）有限公司 | Deep learning-based poster generation method and device and computer equipment |
| CN111797845A (en) * | 2019-03-19 | 2020-10-20 | 北京沃东天骏信息技术有限公司 | Picture processing method and device, storage medium and electronic equipment |
| CN113822689A (en) * | 2020-07-01 | 2021-12-21 | 北京沃东天骏信息技术有限公司 | Advertisement conversion rate estimation method and device, storage medium and electronic equipment |
| CN112598453B (en) * | 2020-12-29 | 2022-11-29 | 上海硬通网络科技有限公司 | Advertisement putting method and device and electronic equipment |
| CN112862516B (en) * | 2021-01-14 | 2024-03-12 | 北京达佳互联信息技术有限公司 | Resource release method and device, electronic equipment and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103544705A (en) * | 2013-10-25 | 2014-01-29 | 华南理工大学 | Image quality testing method based on deep convolutional neural network |
| US20150294370A1 (en) * | 2011-04-07 | 2015-10-15 | ntangle labs, inc. | Target Area Based Monetization Using Sensory Feedback |
| CN105046515A (en) * | 2015-06-26 | 2015-11-11 | 深圳市腾讯计算机系统有限公司 | Advertisement ordering method and device |
| CN106296690A (en) * | 2016-08-10 | 2017-01-04 | 北京小米移动软件有限公司 | The method for evaluating quality of picture material and device |
| CN106295682A (en) * | 2016-08-02 | 2017-01-04 | 厦门美图之家科技有限公司 | A kind of judge the method for the picture quality factor, device and calculating equipment |
| CN106599869A (en) * | 2016-12-22 | 2017-04-26 | 安徽大学 | Vehicle attribute identification method based on multi-task convolutional neural network |
-
2017
- 2017-05-31 CN CN201710400244.2A patent/CN107330715B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150294370A1 (en) * | 2011-04-07 | 2015-10-15 | ntangle labs, inc. | Target Area Based Monetization Using Sensory Feedback |
| CN103544705A (en) * | 2013-10-25 | 2014-01-29 | 华南理工大学 | Image quality testing method based on deep convolutional neural network |
| CN105046515A (en) * | 2015-06-26 | 2015-11-11 | 深圳市腾讯计算机系统有限公司 | Advertisement ordering method and device |
| CN106295682A (en) * | 2016-08-02 | 2017-01-04 | 厦门美图之家科技有限公司 | A kind of judge the method for the picture quality factor, device and calculating equipment |
| CN106296690A (en) * | 2016-08-10 | 2017-01-04 | 北京小米移动软件有限公司 | The method for evaluating quality of picture material and device |
| CN106599869A (en) * | 2016-12-22 | 2017-04-26 | 安徽大学 | Vehicle attribute identification method based on multi-task convolutional neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107330715A (en) | 2017-11-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107330715B (en) | Method and device for selecting picture advertisement material | |
| US20210256403A1 (en) | Recommendation method and apparatus | |
| US11132349B2 (en) | Update basis for updating digital content in a digital medium environment | |
| US10733262B2 (en) | Attribute control for updating digital content in a digital medium environment | |
| US10354184B1 (en) | Joint modeling of user behavior | |
| CN110266745B (en) | Information flow recommendation method, device, equipment and storage medium based on deep network | |
| CN114036398B (en) | Content recommendation and ranking model training method, device, equipment and storage medium | |
| CN112163676A (en) | Multitask service prediction model training method, device, equipment and storage medium | |
| US10937070B2 (en) | Collaborative filtering to generate recommendations | |
| CN111783810A (en) | Method and apparatus for determining attribute information of user | |
| CN112966701A (en) | Method and device for classifying objects | |
| CN113780321A (en) | Picture auditing method and system, electronic equipment and computer readable medium | |
| CN112231299B (en) | Method and device for dynamically adjusting feature library | |
| US10678800B2 (en) | Recommendation prediction based on preference elicitation | |
| US20240046922A1 (en) | Systems and methods for dynamically updating machine learning models that provide conversational responses | |
| US11645563B2 (en) | Data filtering with fuzzy attribute association | |
| US11049041B2 (en) | Online training and update of factorization machines using alternating least squares optimization | |
| CN112449217A (en) | Method and device for pushing video | |
| CN111784091B (en) | Method and device for processing information | |
| CN114493674A (en) | Advertisement click rate prediction model and method | |
| CN112418929A (en) | Data sharing method and device | |
| CN112308942A (en) | Method and apparatus for generating image | |
| CN111626805B (en) | Information display method and device | |
| CN111915339A (en) | Data processing method, device and equipment | |
| WO2022037689A1 (en) | Data form-based data processing method and machine learning application method |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |