An advertisement delivery optimization method, system, device and medium

Abstract

The application discloses an advertisement optimization method, system, device and medium. The method comprises the following steps: obtaining an error value between a current advertisement effect and an expected effect; calculating a nonlinear PID control signal by using a nonlinear PID algorithm according to the error value; obtaining historical data of the effect of the advertisement, and predicting the error by using a long short-term memory network on the historical data to obtain error prediction data; converting the error prediction data by using a nonlinear function to obtain a feedforward compensation signal; and optimizing the advertisement according to the nonlinear PID control signal and the feedforward compensation signal. Compared with the related art, the application optimizes the PID algorithm by introducing a nonlinear function, and improves the response speed of the advertisement optimization system while improving the accuracy of the advertisement optimization system by combining the long short-term memory network (LSTM) for feedforward compensation prediction.

Description

TECHNICAL FIELD

[0001] The application relates to the field of advertisement optimization, and in particular to an advertisement optimization method, system, device and medium. BACKGROUND

[0002] With the rapid development of Internet advertising, advertisement delivery has become one of the important means of enterprise marketing. In advertisement delivery, how to accurately control the delivery amount, improve the advertising effect and ensure the cost control has become an important challenge faced by the advertising platform. Especially in the dynamic market environment, the advertising system needs to be able to respond to the changes in the external environment in real time, such as market competition, user demand fluctuations, advertising effect changes, etc., in order to optimize the advertising delivery strategy.

[0003] At present, the advertisement delivery system generally adopts an adjustment strategy based on historical data, and adjusts the advertisement delivery amount through a simple proportional, integral and differential (PID) control method. However, this method mainly relies on the feedback of past errors and cannot effectively predict future changes, resulting in a lag in response and insufficient adaptability to dynamic changes and complex patterns of the system. Therefore, there is an urgent need for a new advertisement optimization method to improve the accuracy and flexibility of advertisement delivery. SUMMARY

[0004] The technical problem to be solved by the application is to provide an advertisement optimization method, system, device and medium. The present application optimizes the PID algorithm by introducing a nonlinear function, and combines LSTM (Long Short-Term Memory Network) for feedforward compensation prediction, thereby improving the accuracy and response speed of the advertisement delivery system.

[0005] To solve the above technical problems, the present application provides the following technical solutions:

[0006] In a first aspect, the application provides an advertisement optimization method, comprising:

[0007] obtaining an error value between a current advertising effect and an expected effect;

[0008] calculating a nonlinear PID control signal according to the error value using a nonlinear PID algorithm;

[0009] obtaining historical data of the effect of advertisement delivery, and predicting the error using a long short-term memory network to obtain error prediction data;

[0010] converting the error prediction data through a nonlinear function to obtain a feedforward compensation signal;

[0011] The advertising delivery is optimized based on the nonlinear PID control signal and the feedforward compensation signal. Further, the nonlinear PID algorithm uses the following calculation formula:

[0012] f(e(t))=α1·tanh(β1e(t))

[0013]

[0014] h(e(t)-e(t-1))=α3·tanh(β3(e(t)-e(t-1)))

[0015] Wherein, α1 and β1 are the control parameters of the proportional term, α1 controls the upper and lower bounds of the contribution of the proportional term to the control signal, and β1 controls the trend of the proportional term as the proportion changes; α2 and β2 are the control parameters of the integral term, α2 controls the upper and lower bounds of the contribution of the integral term to the control signal, and β2 controls the trend of the integral term as the integral changes; α3 and β3 are the control parameters of the derivative term, α3 controls the upper and lower bounds of the contribution of the derivative term to the control signal, and β3 controls the trend of the derivative term as the derivative changes.

[0016] Furthermore, the Long Short-Term Memory network is calculated using the following formula:

[0017] f t =σ(W f ·[h t-1 ,e t ]+b f )

[0018] i t =σ(W i ·[h t-1 ,e t ]+b i )

[0019]

[0020] o t =σ(W o ·[h t-1 ,e t ]+b o )

[0021] h t =o t ·tanh(C t )

[0022]

[0023] Among them, W f This is the weight matrix of the forget gate; σ is the sigmoid activation function, ht-1 It is the hidden state from the previous moment, e t b represents the error at the current time. f It is the bias term of the forget gate; W i Here is the weight matrix of the input gate, σ is the sigmoid activation function, and b... i It is the bias term of the input gate; W C This is the weight matrix of the candidate memory units, tanh is the hyperbolic tangent function, and b C Bias for candidate memory cells; f t It is the output of the forget gate, C t-1 It is the cell state at the previous moment. It is the output of the candidate memory unit, C t It is the current cell state; o t It is the activation value of the output gate, W. o It is the weight matrix of the output gate, b o It is the bias term of the output gate; h t It is the hidden state at the current moment, tanh(C t ) is the activation value of the current cell state; This represents the error value for predicting the future.

[0024] Furthermore, the error prediction data is transformed using a nonlinear function to obtain a feedforward compensation signal, calculated using the following formula:

[0025] l(FF(t))=α4·tanh(β4·FF(t))

[0026] Wherein, α4 ​​and β4 are the control parameters for the contribution of the feedforward compensation signal.

[0027] Furthermore, the optimization of advertising delivery based on the nonlinear PID control signal and the feedforward compensation signal is performed using the following calculation formula:

[0028]

[0029] Where u(t) is the control signal.

[0030] Furthermore, the control parameters are adaptively adjusted, including adjusting the control parameters using a gradient descent method.

[0031] Furthermore, the control parameters are adjusted using gradient descent, and the calculation formula is as follows:

[0032]

[0033] Where y(t) represents the actual effect of the advertising campaign, y *(t) represents the desired effect, η is the learning rate, and θ represents all parameters to be optimized.

[0034] Secondly, this application also provides an advertising delivery optimization system, including:

[0035] The acquisition module is used to obtain the error value between the current advertising effect and the expected effect;

[0036] The PID calculation module is used to calculate the nonlinear PID control signal based on the error value using a nonlinear PID algorithm.

[0037] The prediction module is used to obtain historical data on the effectiveness of advertising, and to use a long short-term memory network to predict the error of the historical data to obtain error prediction data.

[0038] The conversion module is used to convert the error prediction data through a nonlinear function to obtain a feedforward compensation signal;

[0039] The optimization module is used to optimize the ad delivery based on the nonlinear PID control signal and the feedforward compensation signal.

[0040] Thirdly, this application also provides a computer electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the advertising delivery optimization method described in any one of the above.

[0041] Fourthly, this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the video generation method described in any one of the above descriptions.

[0042] The advertising optimization method, system, device, and medium provided in this application have the following beneficial effects:

[0043] 1. This application effectively addresses the complexity and dynamism of the advertising market by precisely controlling advertising delivery strategies through real-time adjustments. Nonlinear PID control uses a nonlinear function (such as tanh) to limit the control signal range, ensuring that the delivery volume is adjusted within a reasonable range and avoiding over-adjustment. The feedforward compensation section uses LSTM to predict future error changes and make adjustments in advance, reducing the lag effect caused by error accumulation and further improving the accuracy and timeliness of advertising delivery.

[0044] 2. This application combines the error feedback mechanism of the PID algorithm with the feedforward compensation mechanism, comprehensively considering real-time error, historical error, error change trends, and future predictions, thereby optimizing the advertising purchase decision-making process and adapting to demand fluctuations under different market conditions. Through this innovative control mechanism, the advertising delivery system can achieve more efficient and accurate advertising effect optimization, improving advertising conversion rates and market competitiveness. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 This is a flowchart illustrating an advertising placement optimization method according to an embodiment of this application;

[0047] Figure 2 This is a schematic diagram of the structure of an advertising delivery optimization system according to an embodiment of this application;

[0048] Figure 3 This is a schematic diagram of the structure of a computer electronic device according to an embodiment of this application. Detailed Implementation

[0049] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0050] It should be noted that when an element is said to be "fixed" to another element, it can be directly on the other element or there may be an intervening element. When an element is said to be "connected" to another element, it can be directly connected to the other element or there may be an intervening element. Conversely, when an element is said to be "directly" on another element, there is no intervening element. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0051] In this application, unless otherwise expressly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0052] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0053] The terminology used in one or more embodiments of this application is for the purpose of describing particular embodiments only and is not intended to limit the scope of one or more embodiments of this application. The singular forms “a,” “the,” and “the” used in one or more embodiments of this application are also intended to include the plural forms unless the context clearly indicates otherwise.

[0054] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein in the template description is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0055] It should be understood that although the terms first, second, etc., may be used to describe various information in one or more embodiments of this application, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first may also be referred to as second without departing from the scope of one or more embodiments of this application, and similarly, second may also be referred to as first. Depending on the context, the word "if" as used herein may be interpreted as "when" or "when".

[0056] Currently, common optimization algorithms in existing advertising delivery systems include the classic PID control algorithm. The PID control algorithm dynamically adjusts the amount of advertising delivered by calculating the proportional (P), integral (I), and derivative (D) of the error. Specifically:

[0057] Proportional control (P): Adjusts the amount of advertising based on the current error, reflecting the deviation of the current system state.

[0058] Integral control (I): Adjusts advertising volume based on the cumulative effect of historical errors to reflect long-term trends.

[0059] Differential control (D): Predicts future error trends and makes adjustments based on the change in error between the current time step and the previous time step.

[0060] However, existing PID control methods have some limitations. First, they rely solely on current and historical errors for adjustment, failing to effectively predict future error changes. Therefore, the system often reacts insufficiently to sudden market and trend changes. Second, existing PID control methods are generally linear and lack adaptability to complex nonlinear systems, resulting in poor control performance when facing dynamic, complex, and highly nonlinear advertising markets.

[0061] To compensate for the shortcomings of PID control, some systems introduce feedforward control or machine learning algorithms (such as LSTM, neural networks, etc.) to predict future error changes. By predicting future trends in advance, adjustments can be made before errors occur, reducing system lag.

[0062] Deficiencies of existing technology:

[0063] Although existing technologies include PID control and feedforward control, they still have some significant drawbacks and limitations:

[0064] Response lag: Traditional PID control methods rely solely on feedback from historical data (current error, cumulative error, and error changes), which cannot respond in advance to future market fluctuations, resulting in a slow response of the control system and an inability to adjust advertising volume in a timely manner.

[0065] Poor nonlinear adaptability: Traditional PID algorithms are primarily linear control, which cannot well adapt to the complex, dynamic, and nonlinear advertising market environment. The advertising market is affected by various factors, such as changes in user behavior and fluctuations in market competition, and traditional PID control is insufficient in its ability to adapt to these complex nonlinear changes.

[0066] Ineffective handling of large-scale data and complex patterns: Existing ad optimization systems often rely on rule settings or simple predictive models, making it difficult to handle large amounts of complex data and potential patterns. For example, bidding strategies, changes in user interests, and competitors' ad adjustments can all affect ad performance, but existing technologies struggle to fully capture these changes, resulting in inaccurate adjustments to ad placement strategies.

[0067] Insufficient predictive capability: Although some methods attempt to incorporate feedforward control, the feedforward compensation in these methods largely relies on simple rules or manually set empirical formulas, lacking accurate prediction of complex future trends. Therefore, the system struggles to accurately predict future error changes and thus cannot make precise adjustments in advance.

[0068] The technical solution of this application and how it solves the above-mentioned technical problems will be described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes will not be repeated in some embodiments. The embodiments of this application will be described below with reference to the accompanying drawings.

[0069] Please refer to Figure 1 The advertising optimization method provided in this application includes at least the following steps:

[0070] S10. Obtain the error value between the current advertising effect and the expected effect.

[0071] Specifically, in this implementation, firstly, it is necessary to calculate the error based on the difference between the current advertising effect and the expected effect. Assume the actual advertising effect is y(t), and the expected effect is... The error e(t) can then be expressed as:

[0072]

[0073] This error e(t) will be used for subsequent PID control and feedforward compensation calculations.

[0074] S20. Based on the error value, a nonlinear PID algorithm is used to calculate and obtain a nonlinear PID control signal.

[0075] Specifically, in this embodiment, the error value is calculated using a nonlinear PID control algorithm to obtain the initial control signal (i.e., the nonlinear PID control signal). Traditional PID control includes a proportional term (f(e(t))) and an integral term. The three parts are the differential project (h(e(t)-e(t-1))). In this embodiment, all three parts are processed by nonlinear functions to improve control accuracy and flexibility.

[0076] In this scheme, we choose to use the tanh function for nonlinear mapping. Specifically, for each control term f(), g(), and h(), the nonlinear transformation is performed using the following formula:

[0077] f(e(t))=α1·tanh(β1e(t))

[0078]

[0079] h(e(t)-e(t-1))=α3·tanh(β3(e(t)-e(t-1)))

[0080] Wherein, α1 and β1 are the control parameters of the proportional term, α1 controls the upper and lower bounds of the contribution of the proportional term to the control signal, and β1 controls the trend of the proportional term as the proportion changes; α2 and β2 are the control parameters of the integral term, α2 controls the upper and lower bounds of the contribution of the integral term to the control signal, and β2 controls the trend of the integral term as the integral changes; α3 and β3 are the control parameters of the derivative term, α3 controls the upper and lower bounds of the contribution of the derivative term to the control signal, and β3 controls the trend of the derivative term as the derivative changes.

[0081] Understandably, existing PID control methods typically use proportional (err), integral (sum_err), and derivative (der_err) terms to calculate the overall impact factor of advertising placement. However, when any of these three terms exhibits large variations, these excessive changes can lead to significant shifts in the "contribution" of that term to the impact factor, making it difficult to assign a definite parameter value. In this embodiment, a tanh function is introduced for scaling, and parameters are added within the tanh function to adjust its changing trend. Once the range of each of the above terms is defined, the overall range is determined.

[0082] For example, when the proportional term (err) is 2, the integral term (sum_err) is 500, and the derivative term (der_err) is 500000, a relatively small weighting coefficient is usually assigned to the derivative term to suppress its influence. However, this "small" coefficient is only reasonable relative to the current numerical range; if der_err reaches 500000000 in subsequent input data, the originally set weights will not be able to effectively control its influence, potentially leading to a risk of "influence factor explosion" in the adjustment system.

[0083] To address this issue, this embodiment employs the hyperbolic tangent function (tanh) to normalize and scale der_err. The advantage of this method is that even with extremely large input values, the tanh function effectively suppresses the amplitude while maintaining relative differentiation between inputs—for example, when der_err = 500000000, the scaled output will still be larger than when der_err = 500000, thus avoiding the risk of runaway due to excessively large absolute values ​​without sacrificing the relative influence of the differential terms.

[0084] S30. Obtain historical data on the effectiveness of advertising campaigns, and use a long short-term memory network to perform error prediction on the historical data to obtain error prediction data.

[0085] Specifically, to reduce the lag in error feedback, a feedforward compensation mechanism is introduced in this embodiment, using LSTM (Long Short-Term Memory) network for error prediction. LSTM can learn the long-term dependencies in historical error sequences, thereby predicting errors at future moments and helping the system make adjustments in advance.

[0086] It should be noted that the input of LSTM is the error sequence e(t-1), e(t-2), ..., e(tn). Through the three gating structures of the network (input gate, forget gate, and output gate), the LSTM model can store and update information at each time step.

[0087] In one embodiment of this application, the LSTM calculates the output using the following formula:

[0088] 1. Forget Gate: Controls whether past memories are retained.

[0089] f t =σ(W f ·[h t-1 ,e t ]+b f )

[0090] Among them, W f This is the weight matrix of the forget gate, σ is the sigmoid activation function, and h... t-1 It is the hidden state from the previous moment, e t b represents the error at the current time. f It is the bias term of the forget gate.

[0091] 2. Input gate: Determines the impact of the current error on the model state.

[0092] i t =σ(W i ·[h t-1 ,e t ]+b i )

[0093] Among them, W i Here, σ is the weight matrix of the input gate, σ is the si gmoid activation function, and h... t-1 It is the hidden state from the previous moment, e t b represents the error at the current time. i It is the bias term of the input gate.

[0094] 3. Candidate memory units: These determine which new information will affect the state of the unit.

[0095]

[0096] Among them, W CHere, tanh is the weight matrix of the candidate memory units, and h is the hyperbolic tangent function. t-1 It is the hidden state from the previous moment, e t b represents the error at the current time. C The bias is used to select candidate memory cells.

[0097] 4. Memory update:

[0098]

[0099] Among them, f t It is the output of the forget gate, C t-1 It is the cell state at the previous moment. It is the output of the candidate memory unit, C t It represents the current state of the cell, indicating the updated memory under the control of the forget gate and the input gate.

[0100] 5. Output gate: Controls the output at the current moment.

[0101] o t =σ(W o ·[h t-1 ,e t ]+b o )

[0102] Among them, o t σ is the activation value of the output gate, σ is the sigmoid activation function, and W is the activation value of the output gate. o It is the weight matrix of the output gate, h t-1 It is the hidden state from the previous moment, e t b represents the error at the current time. o It is the bias term of the output gate.

[0103] 6. Final output:

[0104] h t =o t ·tanh(C t )

[0105] Among them, h t It is the hidden state at the current moment, tanh(C t ) is the activation value of the current cell state.

[0106] Through the above calculations, LSTM can predict future errors based on past error sequences. LSTM output prediction value This will be used as a feedforward compensation signal:

[0107]

[0108] S40. The error prediction data is transformed using a nonlinear function to obtain a feedforward compensation signal.

[0109] Specifically, to ensure that the feedforward compensation signal FF(t) is within a reasonable range and within reasonable upper and lower bounds, and to prevent excessive fluctuations, in this embodiment, the feedforward compensation signal needs to be further processed by a nonlinear function to obtain the final feedforward compensation signal:

[0110] l(FF(t))=α4·tanh(β4·FF(t))

[0111] Wherein, α4 ​​and β4 are the control parameters for the contribution of the feedforward compensation signal.

[0112] S50. Optimize the advertising delivery based on the nonlinear PID control signal and the feedforward compensation signal.

[0113] In one embodiment of this application, the optimization of advertising delivery based on the nonlinear PID control signal and the feedforward compensation signal is performed using the following calculation formula:

[0114]

[0115] Among them, u(t) is the control signal, which is the final delivery control strategy.

[0116] Specifically, in this embodiment, the advertising delivery is optimized by combining nonlinear PID control signals and feedforward compensation signals.

[0117] It should be noted that, before calculating the final control signal, in this embodiment, it is necessary to reasonably set the initial parameters and dynamically adjust each parameter, including the parameters α1, α2, α3, β1, β2, β3 of the nonlinear PID control and the parameters α4, β4 of the feedforward compensation term, as detailed below:

[0118] 1. Initial parameter settings

[0119] The initial values ​​of the parameters can be set empirically based on historical data, for example:

[0120] α i = Initial gain coefficient, β i =Nonlinear adjustment coefficient

[0121] The selection of these parameters determines the magnitude, speed, and sensitivity to errors of the system response.

[0122] 2. Parameter adaptive adjustment

[0123] To adapt to the dynamic changes in the advertising market, we use the gradient descent optimization method to dynamically adjust parameters based on the system's operating status.

[0124] Parameter updates based on gradient descent

[0125] The loss function is calculated here, defined as the difference between the actual effect y(t) and the expected value. Sum of squared errors:

[0126]

[0127] L is the calculated loss value.

[0128] Update parameters using gradient descent:

[0129]

[0130] Where η is the learning rate and θ represents all parameters to be optimized.

[0131] Understandably, during ad delivery, the system needs to monitor ad performance in real time and adjust control signals based on new error feedback. Through continuous error calculation, PID control, feedforward compensation, and optimization adjustments, the system can maintain the alignment of ad delivery volume with market demand, thereby achieving continuous optimization of ad delivery.

[0132] The advertising optimization method provided in this application has the following beneficial effects:

[0133] 1. This application effectively addresses the complexity and dynamism of the advertising market by precisely controlling advertising delivery strategies through real-time adjustments. Nonlinear PID control uses a nonlinear function (such as tanh) to limit the control signal range, ensuring that the delivery volume is adjusted within a reasonable range and avoiding over-adjustment. The feedforward compensation section uses LSTM to predict future error changes and make adjustments in advance, reducing the lag effect caused by error accumulation and further improving the accuracy and timeliness of advertising delivery.

[0134] 2. This application combines the error feedback mechanism of the PID algorithm with the feedforward compensation mechanism, comprehensively considering real-time error, historical error, error change trends, and future predictions, thereby optimizing the advertising purchase decision-making process and adapting to demand fluctuations under different market conditions. Through this innovative control mechanism, the advertising delivery system can achieve more efficient and accurate advertising effect optimization, improving advertising conversion rates and market competitiveness.

[0135] Please see Figure 2 This application also provides an advertising delivery optimization system 200, including:

[0136] The acquisition module 201 is used to acquire the error value between the current advertising effect and the expected effect;

[0137] The PID calculation module 202 is used to calculate the nonlinear PID control signal based on the error value using a nonlinear PID algorithm.

[0138] The prediction module 203 is used to obtain historical data on the effectiveness of advertising, and to use a long short-term memory network to perform error prediction on the historical data to obtain error prediction data.

[0139] The conversion module 204 is used to convert the error prediction data through a nonlinear function to obtain a feedforward compensation signal;

[0140] The optimization module 205 is used to optimize the advertising delivery based on the nonlinear PID control signal and the feedforward compensation signal.

[0141] Please see Figure 3 This application also provides a computer electronic device 300, including a memory 303 and a processor 302. The memory 303 stores a computer program, and the processor executes the computer program to implement the steps of the advertising placement optimization method described above.

[0142] Specifically, the electronic device 300 includes: a transceiver 301, a bus interface, and a processor 302, wherein the processor 302 is used to obtain the error value between the current advertising effect and the expected effect;

[0143] Based on the error value, a nonlinear PID algorithm is used to calculate and obtain the nonlinear PID control signal;

[0144] Historical data on the effectiveness of advertising campaigns are obtained, and error prediction data is obtained by using a long short-term memory network on the historical data.

[0145] The error prediction data is transformed using a nonlinear function to obtain a feedforward compensation signal;

[0146] The advertising delivery is optimized based on the aforementioned nonlinear PID control signal and feedforward compensation signal.

[0147] In this embodiment of the application, the electronic device 300 further includes a memory 303. Figure 3In this context, the bus architecture can include any number of interconnected buses and bridges, specifically linking various circuits together, represented by one or more processors (processor 302) and memory (memory 303). The bus architecture can also link various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. The bus interface provides an interface. The transceiver 301 can be multiple elements, including transmitters and receivers, providing a unit for communicating with various other devices over a transmission medium. The processor 302 is responsible for managing the bus architecture and general processing, and the memory 303 can store data used by the processor 302 during operation.

[0148] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the video generation method described in any one of the above-described embodiments.

[0149] In this embodiment, the computer-readable storage medium can be a non-volatile storage medium or a volatile storage medium. For example, the computer storage medium may include, but is not limited to, various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0150] In all examples shown and described herein, any specific values ​​should be interpreted as merely exemplary and not as limitations; therefore, other examples of exemplary embodiments may have different values.

[0151] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0152] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can also be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings show the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that, in alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and combinations of blocks in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0153] In addition, the functional modules or units in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0154] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a terminal device (which may be a smartphone, personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application.

[0155] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. An optimization method for ad placement, characterized in that, include: Obtain the error value between the current advertising effect and the expected effect; Based on the error value, a nonlinear PID algorithm is used to calculate and obtain the nonlinear PID control signal. The calculation formula is as follows: ； in, For the control parameters of the proportional term, The upper and lower bounds of the contribution of the proportional term to the control signal. Control the trend of the proportional term as the proportion changes; , For the control parameters of the integral term, The upper and lower bounds of the contribution of the control integral term to the control signal. Control the trend of the integral term as the integral changes; , These are the control parameters for the differential term. The upper and lower bounds of the contribution of the control derivative term to the control signal. Control the trend of the differential term as the differential changes; Historical data on the effectiveness of advertising campaigns are obtained, and error prediction data is obtained by using a long short-term memory network on the historical data. The error prediction data is transformed using a nonlinear function to obtain the feedforward compensation signal, calculated as follows: ； in, These are the control parameters for the feedforward compensation term; Based on the aforementioned nonlinear PID control signal and feedforward compensation signal, the advertising delivery is optimized using the following calculation formula: ； in, Control signals.

2. The method for optimizing advertising placement according to claim 1, characterized in that, The Long Short-Term Memory (LSTM) network is calculated using the following formula: ； ； ； ； ； ； ； in, It is the weight matrix of the forget gate; It is the sigmoid activation function. It was the hidden state from a previous moment. The error at the current moment, It is the bias term of the forget gate; It is the weight matrix of the input gate. It is the sigmoid activation function. It is the bias term of the input gate; It is the weight matrix of the candidate memory units. It is the hyperbolic tangent function. Bias for candidate memory cells; It is the output of the forget gate. It is the cell state at the previous moment. It is the output of the candidate memory unit. It represents the current state of the cell. It is the activation value of the output gate. It is the weight matrix of the output gate. It is the bias term of the output gate; It is the hidden state at the current moment. It is the activation value of the current cell state; This represents the error value for predicting the future.

3. The method for optimizing advertising placement according to claim 1 or 2, characterized in that, The control parameters are adaptively adjusted, including adjusting the control parameters using a gradient descent method.

4. The method for optimizing advertising placement according to claim 3, characterized in that, The control parameters are adjusted using gradient descent, and the calculation formula is as follows: ； in, For the actual effect of advertising, For the desired effect, For learning rate, This represents all parameters that need to be optimized.

5. An optimization system for ad delivery, characterized in that, include: The acquisition module is used to obtain the error value between the current advertising effect and the expected effect; The PID calculation module is used to calculate the nonlinear PID control signal based on the error value using a nonlinear PID algorithm. The calculation formula is as follows: ； in, For the control parameters of the proportional term, The upper and lower bounds of the contribution of the proportional term to the control signal. Control the trend of the proportional term as the proportion changes; , For the control parameters of the integral term, The upper and lower bounds of the contribution of the control integral term to the control signal. Control the trend of the integral term as the integral changes; , These are the control parameters for the differential term. The upper and lower bounds of the contribution of the control derivative term to the control signal. Control the trend of the differential term as the differential changes; The prediction module is used to obtain historical data on the effectiveness of advertising, and to use a long short-term memory network to predict the error of the historical data to obtain error prediction data. The conversion module is used to convert the error prediction data using a nonlinear function to obtain a feedforward compensation signal. The calculation formula is as follows: ； in, These are the control parameters for the feedforward compensation term; The optimization module is used to optimize the ad delivery based on the nonlinear PID control signal and the feedforward compensation signal. The calculation formula is as follows: ； in, Control signals.

6. A computer electronic device, characterized in that, The device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the advertising delivery optimization method according to any one of claims 1-4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the advertising delivery optimization method according to any one of claims 1-4.

Images