Big data analysis and prediction-based strategy dynamic optimization system and method

Abstract

The invention discloses a strategy dynamic optimization system and method based on big data analysis and prediction, and relates to the technical field of data analysis and processing, and the system comprises a data collection layer which is used for building a data source; the data processing layer is used for performing data feature extraction after performing data cleaning on the data source; the extreme scene simulation layer is used for executing coping strategy training in an adversarial environment; the competition optimization layer is used for configuring a competition agent and updating an adaptive strategy set after adaptive evaluation of strategy quality, performing game optimization and generating a game optimization result; and the strategy dynamic optimization layer is used for receiving the game optimization result and establishing a dynamic response strategy. The technical problems that in the prior art, analysis deviation is caused by poor data quality, a static strategy is difficult to adapt to a complex and changeable market environment and cannot effectively cope with dynamic games of different competitive roles, and the strategy adaptability and flexibility are insufficient are solved, and the technical effect of improving the decision-making flexibility and adaptability of the fluctuating market environment is achieved.

Description

Technical Field

[0001] This application relates to the technical field of data analysis and processing, and specifically relates to a strategy dynamic optimization system and method based on big data analysis and prediction. Background Art

[0002] The market environment is changing rapidly, and various trading activities, information dissemination, and interaction methods have undergone huge changes. With the popularization of the Internet, trading data has shown explosive growth, covering various online and offline trading scenarios, and containing key information such as market supply and demand, price fluctuations, and market hotspots. For example, news data reflects the macroeconomic situation, industry dynamics, and emergencies in real time, and has a profound impact on the market trend; social data shows consumers' preferences, emotions, and group behavior trends, providing a new perspective for understanding market demand; search data can directly reflect users' focus of attention and potential needs. However, there are problems with uneven data quality in the vast amount of data, such as data missing, errors, duplicates, etc. Directly using the original data will lead to deviation in the analysis results. At the same time, in many highly volatile markets (such as cryptocurrency, luxury goods, NFT, creative content, futures trading, etc.), traditional time series models cannot stably predict, and the strategy formulation method is difficult to adapt to the complex and changing market environment. Static strategies cannot cope with the changing dynamic game environment, resulting in insufficient flexibility and adaptability of the strategies.

[0003] Therefore, in the current related technologies, there are technical problems that poor data quality leads to analysis deviation, static strategies are difficult to adapt to the complex and changing market environment and cannot effectively cope with the dynamic game of different competitive roles, resulting in insufficient adaptability and flexibility of the strategies. Summary of the Invention

[0004] This application provides a strategy dynamic optimization system and method based on big data analysis and prediction, solves the technical problems in the prior art that poor data quality leads to analysis deviation, static strategies are difficult to adapt to the complex and changing market environment and cannot effectively cope with the dynamic game of different competitive roles, resulting in insufficient adaptability and flexibility of the strategies, and achieves the technical effect of improving the decision-making flexibility and adaptability in a volatile market environment.

[0005] This application provides a strategy dynamic optimization system based on big data analysis and prediction. The system includes: a data collection layer for collecting transaction data, news data, social data, search data, public data, and private domain user retention data to establish a data source; a data processing layer for performing data cleaning on the data source, then extracting data features, identifying abnormal patterns according to the data feature extraction results, and constructing a strategy training environment; an extreme scenario simulation layer for performing coping strategy training in an adversarial environment after injecting random noise into the strategy training environment to establish an adaptation strategy set; a competition optimization layer for configuring competition agents according to the data source, where the competition agents include a leader, a follower, and an adversary, and after performing an adaptation evaluation of the strategy quality on the adaptation strategy set, updating the adaptation strategy set, and using the updated adaptation strategy set and the competition agents to perform game optimization to generate a game optimization result; a strategy dynamic optimization layer for receiving the game optimization result and establishing a dynamic response strategy.

[0006] In a possible implementation, the strategy dynamic optimization system based on big data analysis and prediction also performs the following processing: establishing an extreme scenario data set, and using a generative adversarial network to expand the extreme scenario data set to generate an updated extreme scenario data set; self-checking the injection of the random noise within a preset period, and performing scenario coverage evaluation according to the self-checking result and the updated extreme scenario data set to establish an injection constraint for the random noise; configuring a time continuous perturbation constraint and establishing risk factors, where the risk factors include liquidity risk and credit risk; injecting random noise into the strategy training environment according to the injection constraint of the random noise, the time continuous perturbation constraint, and the risk factors.

[0007] In a possible implementation, the strategy dynamic optimization system based on big data analysis and prediction also performs the following processing: monitoring indicators of a strategy optimization entity to establish a key indicator set, where the key indicator set includes return rate, risk exposure, liquidity, and resource utilization efficiency; establishing the current state of the strategy optimization entity according to the key indicator set; using the current state and the updated adaptation strategy set as input data to perform game optimization with the competition agents.

[0008] In a possible implementation, the strategy dynamic optimization system based on big data analysis and prediction also performs the following processing: recording the revenue data, risk data, and decision-making performance data of all roles during the game process to establish a feedback signal; using the updated adaptation strategy set as an initial strategy set, and using the feedback signal as an optimization constraint to perform iterative update of the initial strategy set; completing game optimization according to the iterative update result.

[0009] In a possible implementation, the policy dynamic optimization system based on big data analysis and prediction further performs the following processing: calculating fitness values in an initial policy set, screening the initial policy set using the fitness values, and establishing excellent individual identifiers; after reconstructing iteration probabilities according to the excellent individual identifiers, performing individual selection of the initial policy set; performing gene exchange based on the individual selection result, and updating the parameter contribution degree of the individual selection result using the optimization constraint; after randomly perturbing the parameters of the individual selection result, completing one iteration update, where the current individual selection result is the individual selection result after gene exchange and adjustment of the parameter contribution degree.

[0010] In a possible implementation, the policy dynamic optimization system based on big data analysis and prediction further performs the following processing: configuring a quality screening threshold, and screening adaptive strategies for the adaptation evaluation result according to the quality screening threshold; after removing the adaptive strategies that do not meet the quality screening threshold, updating the adaptive strategy set with the remaining adaptive strategies.

[0011] In a possible implementation, the policy dynamic optimization system based on big data analysis and prediction further performs the following processing: extracting the data source after data cleaning into structured data and unstructured data; performing cross-modal feature fusion on the structured data and unstructured data using a deep learning model; performing adaptive feature selection on the cross-modal feature fusion result to complete data feature extraction.

[0012] In a possible implementation, the policy dynamic optimization system based on big data analysis and prediction further performs the following processing: obtaining the behavior pattern database of the competing agent, performing backward reasoning according to the behavior pattern database, and predicting the potential countermeasures of the competing agent; establishing a mapping game strategy according to the potential countermeasures; performing compensation for the game optimization result according to the mapping game strategy.

[0013] In a possible implementation, the policy dynamic optimization system based on big data analysis and prediction further performs the following processing: an early warning layer, which is used to analyze the dynamic response policy, generate a risk early warning signal, and issue the risk early warning signal for risk warning.

[0014] The present application also provides a method for dynamically optimizing strategies based on big data analysis and prediction, including: collecting transaction data, news data, social data, search data, public data, and private domain user retention data to establish a data source; after cleaning the data source, performing data feature extraction, identifying abnormal patterns based on the results of data feature extraction, and constructing a strategy training environment; after injecting random noise into the strategy training environment, performing coping strategy training in an adversarial environment to establish an adaptation strategy set; configuring a competition agent according to the data source, where the competition agent includes a leader, a follower, and an adversary, and after performing an adaptation evaluation of the strategy quality on the adaptation strategy set, updating the adaptation strategy set, and using the updated adaptation strategy set and the competition agent to perform game optimization to generate a game optimization result; receiving the game optimization result and establishing a dynamic response strategy.

[0015] It is intended to solve the technical problems in the prior art, such as analysis deviation caused by poor data quality, static strategies being difficult to adapt to complex and changeable market environments and unable to effectively cope with the dynamic games of different competitive roles, resulting in insufficient strategy adaptability and flexibility, and achieve the technical effect of enhancing the decision-making flexibility and adaptability in volatile market environments through the system and method for dynamically optimizing strategies based on big data analysis and prediction proposed in this application, including a data acquisition layer for establishing a data source; a data processing layer for performing data cleaning on the data source and then performing data feature extraction; an extreme scenario simulation layer for performing coping strategy training in an adversarial environment; a competition optimization layer for configuring a competition agent and performing an adaptation evaluation of the strategy quality, then updating the adaptation strategy set and performing game optimization to generate a game optimization result; and a strategy dynamic optimization layer for receiving the game optimization result and establishing a dynamic response strategy. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings of the embodiments of the present disclosure will be briefly introduced below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the operations before or below do not necessarily need to be executed precisely in sequence. On the contrary, according to needs, they can be executed in reverse order or simultaneously. At the same time, other operations can also be added to these processes, or one or several operations can be removed from these processes.

[0017] Figure 1 It is a schematic structural diagram of a system for dynamically optimizing strategies based on big data analysis and prediction provided by an embodiment of the present application.

[0018] Figure 2 It is a schematic flowchart of a method for dynamically optimizing strategies based on big data analysis and prediction provided by an embodiment of the present application.

[0019] Description of the attached drawing reference numerals: Data acquisition layer 10, data processing layer 20, extreme scenario simulation layer 30, competition optimization layer 40, policy dynamic optimization layer 50. Specific implementation manners

[0020] The above description is only an overview of the technical solution of this application. In order to be able to understand the technical means of this application more clearly, it can be implemented according to the content of the description. And in order to make the above and other purposes, features and advantages of this application more obvious and understandable, the specific implementation manners of this application are hereby given.

[0021] In order to make the purpose, technical solution and advantages of this application clearer, the following will further describe this application in detail with reference to the attached drawings. The described embodiments should not be regarded as limitations of this application. All other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of this application.

[0022] In the following description, "some embodiments" are involved, which describe a subset of all possible embodiments. However, it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict. The terms "first\second" involved are only used to distinguish similar objects and do not represent a specific order for the objects. The terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, system, product or server including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or modules not clearly listed or inherent to these processes, products or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application.

[0023] The embodiments of this application provide a policy dynamic optimization system based on big data analysis and prediction, as Figure 1 shown. The system includes: A data acquisition layer 10, which is used to collect transaction data, news data, social data, search data, public data, and private domain user retention data to establish a data source.

[0024] Preferably, the data collection layer serves as the front end of the data processing system, and is mainly used to collect various types of data and establish data sources, including transaction data, news data, social data, search data, public data, and private domain user retention data. Specifically, transaction data refers to data generated in the process of various commercial transactions, covering various online and offline transaction scenarios, which may include transaction time, transaction amount, information on transaction goods or services, identity information of both parties to the transaction, transaction location and other data. For example, order data on e-commerce platforms includes the name, price, quantity, payment method and merchant information of the goods purchased by consumers; transaction data of offline physical stores are recorded through the cash register system, including transaction time, product details, payment method, etc.; by analyzing transaction data, we can understand the market supply and demand relationship, price fluctuation trends, consumer purchasing preferences and consumption capacity, etc. News data refers to news information content released by various news media, including text, pictures, videos and other forms, which reflect the dynamic information of various fields in real time, mainly including news title, text, release time, news source, keywords, etc. For example, economic news may contain information such as macroeconomic policy adjustments, industry development trends, corporate mergers and acquisitions, and restructuring; science and technology news may report on new technology research and development breakthroughs, the rise of emerging industries, etc.; it helps companies and investors to timely understand the macroeconomic situation, industry trends and the impact of emergencies on the market.

[0025] Preferably, social data refers to various data generated by users on social platforms, mainly including users' personal profiles, posted content (such as text, pictures, videos, etc.), social relationships (friend lists, following and being followed relationships, etc.), interaction behaviors (liking, commenting, sharing), and other behavioral data, as well as information such as the user's location, used device, and login time. For example, content such as users' evaluations of product usage experiences, views on a certain brand, and discussions on popular topics posted on social media platforms can help enterprises deeply understand consumers' preferences, needs, emotions, and behavioral trends of social groups. Search data refers to relevant data such as keywords entered by users in search engines or other search tools, search time, and search result click behaviors, mainly including query terms entered by users, page views of search results, link information clicked by users, etc. For example, when a user enters "smartphone recommendation" in a search engine, the search engine will record this query term and the relevant search result pages subsequently clicked by the user, all of which belong to search data, which can reflect users' focus of attention and potential needs, and help enterprises understand market hotspots and consumers' information search habits, so as to optimize product information display and adjust advertising placement strategies. Public data refers to data released by public institutions or other public channels, such as economic data (such as GDP, inflation rate, unemployment rate, etc.), industry reports, census data, etc. Public data has authority and macroscopicality, which is helpful for overall market analysis. Private domain user retention data refers to user data in the private domain traffic pool owned by an enterprise or organization, including registration information, login data, browsing records, consumption history, retention time, user feedback, etc. of users on the enterprise platform (such as official website, APP). Private domain user retention data helps to understand the characteristics and behaviors of the enterprise's own user group. Finally, various collected data are integrated to establish a data source to ensure the efficient storage, quick query, and secure management of data.

[0026] The data processing layer 20 is used to perform data cleaning on the data source, then execute data feature extraction, identify abnormal patterns according to the data feature extraction results, and construct a policy training environment.

[0027] Preferably, the data processing layer further processes the collected data sources, including cleaning the data sources and performing feature extraction to obtain the data feature extraction results. Specifically, since the collected data may have problems such as incompleteness, errors, duplications, or inconsistencies, data cleaning aims to improve data quality and ensure the accuracy and reliability of subsequent analysis results. It usually includes handling missing values, such as deleting the records where missing values are located, filling with the mean or median according to the characteristics of the data; removing duplicate data to ensure the uniqueness of the data; detecting and correcting incorrect data, for example, checking the rationality of the data through data validation rules and correcting or marking the obviously incorrect data; handling inconsistent data, unifying the data format, encoding, and unit, etc., to make the data consistent and comparable.

[0028] Preferably, then feature extraction is performed on the cleaned data to obtain the key feature information that can reflect the data features and patterns, and the original data is transformed into a form that is easier to understand and analyze. Specifically, different feature extraction methods are used for different types of data. For transaction data, statistical features of transaction frequency and transaction amount (such as mean, standard deviation, maximum value, minimum value, etc.), commodity category distribution, etc. may be extracted; for news data, keyword frequency, news topic category, sentiment tendency (judging the positive, negative, or neutral sentiment of the news content through sentiment analysis technology), etc. may be extracted; for social data, user social activity (such as the frequency of posting content, number of interactions, etc.), social influence (measured by the number of followers, likes, and comments, etc.), topic preference, etc. may be extracted; for search data, the frequency of search keywords, search time distribution, click-through rate of search results, etc. may be extracted; and these features are used as the data feature extraction results.

[0029] Preferably, based on the data feature extraction results, abnormal patterns are identified, that is, abnormal data points or data sets different from the normal patterns in the data are found, so as to timely detect abnormal fluctuations in the market, abnormal behaviors of users, etc. Specifically, based on the results of data feature extraction, various anomaly detection algorithms and models are used to identify abnormal patterns. For example, statistical methods are used to set thresholds, and data points exceeding the thresholds of the normal range are regarded as abnormal; or clustering algorithms in machine learning are used to divide the data into different clusters, and data points in isolated clusters may be considered as abnormal points; deep learning models can also be adopted to learn the patterns of normal data and perform anomaly detection on data with large differences from the learned patterns. Finally, a strategy training environment is constructed, which can simulate real scenarios to adapt to various situations in actual applications. Specifically, according to data features and business requirements, various parameters and rules of the environment are determined. For example, in the financial trading strategy training environment, parameters such as the change range of asset prices, transaction costs, and risk indicators are set according to historical trading data and market fluctuations; in the marketing decision-making strategy training environment, different marketing strategy options (such as advertising channels, promotion activity forms, etc.) and corresponding effect evaluation indicators are set according to the behavioral characteristics of consumers and market competition situations; at the same time, the cleaned data and extracted features are used as input data for the environment to be trained to achieve the expected goal, and finally a strategy training environment is obtained.

[0030] The extreme scenario simulation layer 30 is used to perform coping strategy training in an adversarial environment after injecting random noise into the strategy training environment, and establish an adaptation strategy set.

[0031] Preferably, the extreme scenario simulation layer is mainly used to simulate extreme situations that may occur in reality, such as many high-volatility markets (such as cryptocurrency, luxury goods, NFT, creative content, futures trading, etc.). By injecting random noise into the strategy training environment, an adversarial environment is constructed, and then an adaptation strategy set is trained and established. Specifically, there are various uncertainties and unexpected situations in the real scenario. By injecting random noise into the strategy training environment to simulate unpredictable factors, the trained strategies can cope with various complex and volatile market environments. Specifically, according to different application scenarios and data characteristics, the type and distribution of the noise are determined. For example, in financial market simulation, random noise conforming to a certain probability distribution (such as normal distribution, t-distribution, etc.) can be generated according to the fluctuation situation of historical data and superimposed on data such as asset prices and trading volumes; in user behavior simulation, random noise can be added to data such as the user's click frequency and purchase time interval to simulate the uncertainty of user behavior. The intensity and range of the noise need to be adjusted according to the actual situation, ensuring that extreme situations can be simulated without making the noise too large to cause the data to lose its authenticity.

[0032] Preferably, the response strategy training in an adversarial environment is then performed, that is, in an adversarial environment with noise, the model or strategy is trained to be able to handle various possible challenges and changes. Specifically, using reinforcement learning, game theory, etc., the agent (such as a trading strategy model, a marketing decision-making model, etc.) continuously conducts trial-and-error learning in the adversarial environment. For example, in a financial trading scenario, the agent makes buying and selling decisions based on the market data with injected noise, and obtains reward or punishment feedback through interaction with the environment, and then continuously adjusts the strategy to maximize the long-term cumulative reward, so as to learn to make optimal decisions in the adversarial environment. During the training process, multiple agents can also be introduced to compete or cooperate, further enhancing the adaptability and robustness of the strategy. For example, in a simulated market competition scenario, different agents represent different market participants, and through mutual competition and learning, they continuously optimize their market strategies; and then obtain multiple adaptation strategies and analyze and screen these strategies, remove the strategies with poor effects, and retain the strategies that perform well in various extreme scenarios to form an adaptation strategy set. For example, in marketing decision-making, the adaptation strategy set may include various marketing strategy combinations for different market competition intensities and different consumer demand fluctuations, such as adopting a strategy of low-price promotion combined with precise advertising placement when the market competition is fierce, and adopting a strategy of increasing the added value of products and strengthening brand promotion efforts when consumer demand is strong, etc.

[0033] The competition optimization layer 40 is used to configure competition agents according to the data source. The competition agents include a leader, a follower, and an adversary. After performing an adaptation evaluation of the strategy quality on the adaptation strategy set, the adaptation strategy set is updated, and the updated adaptation strategy set and the competition agents are used for game optimization to generate a game optimization result.

[0034] Preferably, the competition optimization layer is mainly used to evaluate and optimize the adaptation strategy set by configuring competition agents, so as to generate better game optimization results. Specifically, competition agents are configured according to the data source to simulate the behaviors and decisions of different roles in the competition environment, that is, corresponding attributes and behavior rules are set for different competition agents according to the information contained in the data source, such as market dynamics, user behavior patterns, competitor data, etc. Among them, competition agents are virtual roles with autonomous decision-making and behavior capabilities in the competition environment, including leaders, followers, and adversaries. Leaders usually have strong market influence and resource advantages and may adopt strategies of proactive innovation and leading market trends; followers tend to observe the actions of leaders and adjust their strategies according to their decisions to obtain a certain market share; adversaries aim to challenge leaders or other competitors and adopt aggressive strategies to try to break the existing market pattern. For example, in the e-commerce market, large e-commerce platforms can act as leaders, attracting users by launching new service models or large-scale promotional activities; small e-commerce platforms, as followers, will imitate some successful practices of large platforms and conduct refined operations for specific user groups; while some emerging e-commerce platforms may act as adversaries, competing for market share by offering unique products or lower prices.

[0035] Preferably, an adaptation evaluation of the strategy quality of the adaptation strategy set is carried out to determine the effectiveness and adaptability of the adaptation strategy set in different competition scenarios, and find out the areas that need to be improved and optimized. Specifically, multiple evaluation indicators are used to evaluate each strategy in the adaptation strategy set. Among them, the evaluation indicators can include the profitability of the strategy (such as expected profit, market share growth, etc.), riskiness (such as the probability of strategy failure, potential losses, etc.), adaptability (the ability to adapt to different market conditions and competitor behaviors), etc. The strategy is applied to the simulated competition scenario, and its effect is observed and compared with the preset goals and other strategies for analysis. For example, for a marketing promotion strategy, evaluate its sales growth, customer acquisition cost, and impact on brand image under different market competition intensities to judge the quality and adaptability of the strategy. Then, according to the evaluation results, the strategies in the adaptation strategy set are adjusted and optimized to enable it to better handle various competition situations, including modifying or eliminating the strategies with poor performance in the adaptation strategy set based on the adaptation evaluation results of strategy quality, and at the same time introducing new strategies or improving existing strategies. For example, if it is found that a certain marketing strategy has poor effect in the face of fierce competition, it may be adjusted, such as changing the advertising channel or adjusting the product pricing strategy; or according to the newly emerging trends in the market and the new movements of competitors, introducing new marketing means, such as social media marketing, word-of-mouth marketing, etc., to enrich and optimize the adaptation strategy set.

[0036] Preferably, the updated adaptation strategy set and competitive agents are used for game optimization, that is, through simulated competitive scenarios, competitive agents with different roles interact according to the strategies to find the optimal strategy combination. Specifically, each competitive agent (dominant, follower, antagonist) selects a suitable strategy based on its own goals, current environmental information, and the updated adaptation strategy set, and then conducts interactions among the competitive agents. For example, after the dominant agent launches a new product, it may affect the market shares of the follower and the antagonist. The follower's price reduction may cause price adjustments by the dominant agent and the antagonist. Each competitive agent obtains feedback based on the interaction results, such as an increase or decrease in revenue, a change in market share, etc. Based on the feedback information, the competitive agent learns and judges the effectiveness of the current strategy, and then adjusts the strategy according to the learning results. After multiple games and strategy adjustments, the competitive agents gradually converge to obtain a strategy combination, which is the result of game optimization, in order to achieve better results in the competitive environment.

[0037] The strategy dynamic optimization layer 50 is used to receive the game optimization result and establish a dynamic response strategy.

[0038] Preferably, based on the game optimization result, the strategy dynamic optimization layer establishes a dynamic response strategy according to the continuously changing environment and competitive situation to achieve better adaptability and competitiveness. Specifically, the strategy dynamic optimization layer collects the game optimization result, including the relatively optimal strategy combination formed by different competitive agents (dominant, follower, antagonist) after the game, as well as relevant evaluation indicators and feedback information, such as the revenue situation of each agent, the change in market share, the stability of the strategy, etc. Then, it analyzes and evaluates the game optimization result to understand the performance and potential problems of the current strategy in different aspects. For example, it analyzes the effectiveness and adaptability of the strategy in coping with market demand fluctuations, technological changes, etc., and finds possible weak links or areas for improvement. Then, it continuously monitors the changes in the external environment, including market trends, technological developments, changes in consumer preferences, etc. According to the environmental changes and the analysis of the game optimization result, it dynamically adjusts and optimizes the existing strategy, which may include reconfiguring the strategies of different competitive agents. For example, it adjusts the innovation direction of the dominant agent, the imitation speed and scope of the follower, and the attack strength and key areas of the antagonist, and formulates multiple alternative strategy plans, and establishes corresponding trigger mechanisms so that when specific environmental conditions occur, it can quickly switch to the appropriate strategy to ensure that the dynamic response strategy has sufficient flexibility and adaptability to quickly respond to various emergencies and uncertainties. Furthermore, the strategy can be continuously optimized as the environment changes, maintaining an advantageous position in the competitive environment.

[0039] Furthermore, the specific configuration of the extreme scenario simulation layer 30 further includes establishing an extreme scenario dataset, and using a generative adversarial network to augment the extreme scenario dataset to generate an updated extreme scenario dataset; performing self-check on the injection of the random noise within a preset period, evaluating the scenario coverage based on the self-check result and the updated extreme scenario dataset, and establishing an injection constraint for the random noise; configuring a time-continuous perturbation constraint, and establishing risk factors, where the risk factors include liquidity risk and credit risk; injecting random noise into the policy training environment according to the injection constraint of the random noise, the time-continuous perturbation constraint, and the risk factors.

[0040] Preferably, collect and sort out data under various extreme conditions from historical event records and databases, such as extreme volatility in the financial market (such as stock market crashes, large exchange rate fluctuations), to form an initial extreme scenario dataset, and then use a generative adversarial network to augment the initial extreme scenario dataset, that is, by learning the characteristics and distribution laws of the original data, generate new, similar but not exactly the same extreme scenario data, increasing the diversity and scale of the dataset to obtain an updated extreme scenario dataset, so as to better cover various possible extreme situations. Then, according to a preset time period, check the process of injecting random noise into the policy training environment, which may include checking whether the generation of the noise conforms to the predetermined rules and distributions, whether the intensity of the noise is within a reasonable range, and whether the injection process is normal, and whether there are errors or abnormal conditions, etc., to obtain the self-check result of the injection.

[0041] Preferably, combining the self-check result of the random noise injection and the augmented updated extreme scenario dataset, evaluate whether the current random noise injection can fully cover various extreme scenarios, including checking whether the training environment after injecting the noise can simulate various extreme situations included in the updated extreme scenario dataset, and whether there are some scenarios that are not effectively covered. Through the scenario coverage evaluation, understand the effect and deficiencies of the current random noise injection, and then based on the scenario coverage evaluation result, determine the constraint conditions for the random noise injection. If it is found that some extreme scenarios are not well covered, it is necessary to adjust the generation method, intensity range or injection frequency of the noise, etc., to ensure that the random noise injection can more comprehensively cover various extreme scenarios and enable the policy training environment to more realistically simulate the actual situation under extreme conditions.

[0042] Preferably, a time - continuous perturbation constraint is configured, that is, when injecting random noise into the policy training environment, it is necessary to ensure that the noise changes continuously over time, conforming to the dynamic change law in the actual situation. When injecting random noise into the policy training environment, it is necessary to ensure that the noise changes continuously over time, conforming to the dynamic change law in the actual situation. For example, the changes in market conditions or environmental factors often occur gradually. Through the time - continuous perturbation constraint, the injected random noise can more realistically simulate this continuous change characteristic; then determine the risk factors related to policy training, specifically including liquidity risk and credit risk. Among them, liquidity risk reflects the ease of an asset being liquidated at a reasonable price in the short term, and credit risk refers to the possibility of a counterparty defaulting or the credit status deteriorating; consider the impact of risk factors on the policy so that the trained policy can better cope with various risks in the actual scenario.

[0043] Preferably, according to the injection constraint of random noise, time - continuous perturbation constraint, and risk factors, random noise is injected into the policy training environment. During the injection process, it is necessary to ensure that the generation and injection of noise meet the requirements of the injection constraint, ensure temporal continuity, and can reflect the influence of factors such as liquidity risk and credit risk. The policy training environment can then simulate a more complex and realistic scenario, enabling the trained policy to have better adaptability in the face of various uncertainties and risks. By continuously injecting random noise and conducting policy training, the model can better learn how to make optimal decisions in different extreme scenarios and risk conditions.

[0044] Furthermore, the specific configuration of the competition optimization layer 40 also includes monitoring the indicators of the policy optimization entity, establishing a key indicator set, where the key indicator set includes return rate, risk exposure, liquidity, and resource utilization efficiency; establishing the current state of the policy optimization entity according to the key indicator set; using the current state and the updated adaptation policy set as input data to execute the game optimization with the competition agent.

[0045] Preferably, relevant data of the strategy optimization entity (such as investment portfolio, enterprise operation strategy, etc.) are monitored for indicators, that is, by collecting and analyzing a large amount of monitoring data, information on the performance of the strategy optimization entity in different aspects is obtained, and indicators crucial for the evaluation and decision-making of the strategy optimization entity are selected to form a key indicator set, including return rate, risk exposure, liquidity, and resource utilization efficiency. Specifically, the return rate measures the profit level of the strategy optimization entity within a certain period, such as investment return rate, return on assets, etc.; risk exposure assesses the various risk levels faced by the strategy optimization entity, such as market risk, credit risk, etc.; liquidity indicates the ability of the strategy optimization entity's assets to be liquidated and meet capital requirements, such as the turnover speed of assets, cash reserve situation, etc.; resource utilization efficiency examines the utilization of resources (such as funds, manpower, equipment, etc.) by the strategy optimization entity, such as the use efficiency of funds, work efficiency of employees, etc., to determine whether resources are reasonably allocated.

[0046] Preferably, each indicator is quantified and analyzed based on the selected key indicator set to establish the overall state of the strategy optimization entity at the current moment. For example, the evaluation results of the return rate, risk exposure, liquidity, and resource utilization efficiency are integrated to form a description that can comprehensively reflect the current operation status of the strategy optimization entity; the current state data of the strategy optimization entity and the updated adaptation strategy set are provided as input data to the game optimization model, and the strategy optimization entity plays games with competitive agents (dominators, followers, adversaries, etc.). Specifically, the competitive agent selects strategies for interaction based on its own goals and understanding of the environment, based on the input current state and adaptation strategy set of the strategy optimization entity. Through multiple strategy selections and game processes, the competitive agent continuously adjusts its strategy to achieve the optimal game result. For example, in the financial investment scenario, the strategy optimization entity (such as an investment institution) plays games with other competitive agents (such as other investment institutions or market participants) according to its current asset status (reflected by indicators such as return rate and risk exposure) and the updated investment strategy set, and realizes the optimal allocation of assets and maximization of returns in the competitive environment by continuously adjusting the investment strategy.

[0047] Furthermore, the specific configuration of the competition optimization layer 40 also includes recording the revenue data, risk data, and decision-making performance data of all roles during the game process to establish a feedback signal; using the updated adaptation strategy set as the initial strategy set, and using the feedback signal as the optimization constraint to perform iterative updates of the initial strategy set; and completing game optimization according to the iterative update results.

[0048] Preferably, during the game between the strategy optimization entity and the competitive agents (dominant players, followers, adversaries, etc.), relevant data of all participating roles are recorded in detail, including revenue data, risk data, and decision-making performance data. Among them, the revenue data includes the revenue obtained by each role at each stage of the game. For example, in a business competition game, the enterprise's sales revenue, profit, etc.; in a financial investment game, the investor's investment return, etc. The risk data includes the risk status faced by each role, such as market risk, credit risk, etc. For example, the value at risk (VaR) of an investment portfolio, the probability of corporate debt default, etc. The decision-making performance data includes the decisions made by each role during the game and the actual execution effects of these decisions. For example, the timing of the decision, the strategy selected by the decision, the change in market share and product sales volume after the decision, etc. Integrating the recorded revenue data, risk data, and decision-making performance data to obtain a feedback signal can comprehensively reflect the comprehensive performance of each role during the game and the overall situation of the game, such as judging which strategies are effective, which decisions have problems, and the gap between the current game state and the expected goal, etc.

[0049] Preferably, the updated set of adaptation strategies is used as the initial strategy set, and the established feedback signal is used as a constraint condition during the iterative update process to guide the optimization direction of the strategy. Specifically, an optimization algorithm (such as a genetic algorithm, a reinforcement learning algorithm, etc.) is used to iteratively update the initial strategy set. In each iteration, the strategies in the strategy set are adjusted, mutated, or combined according to the feedback signal to generate new strategies, and then the newly generated strategies are evaluated to determine whether they are superior to the strategies in the original strategy set. If the new strategy is better, it is incorporated into the updated strategy set; otherwise, the original strategy is retained. The strategy set is continuously optimized through multiple iterations to obtain the iterative update result and complete the game optimization based on the iterative update result, that is, the most suitable strategy combination for the current game environment and the goals of each role is selected from the evaluated strategy set, which can enable each role to maximize its own interests or achieve other predetermined goals as much as possible during the competition or cooperation game process.

[0050] Furthermore, the specific configuration of the competition optimization layer 40 also includes calculating the fitness values in the initial strategy set, using the fitness values to screen the initial strategy set, and establishing excellent individual identifiers; after reconstructing the iteration probability according to the excellent individual identifiers, performing individual selection of the initial strategy set; performing gene exchange according to the individual selection result, and updating the parameter contribution degree of the individual selection result using the optimization constraint; after randomly perturbing the parameters of the individual selection result, completing one iteration update, where the current individual selection result is the individual selection result after gene exchange and adjustment of the parameter contribution degree.

[0051] Preferably, calculate the fitness values in the initial strategy set, that is, quantify each evaluation index according to revenue data, risk data, decision-making performance data, etc., assign weights according to their importance, and then perform weighted calculation on the advantages and disadvantages of each strategy in the initial strategy set, that is, obtain the fitness value of each strategy; then screen the initial strategy set according to the fitness values, retain the strategies with higher fitness values, and eliminate the strategies with lower fitness values to remove the strategies that obviously do not meet the optimization goal, narrow the scope of the strategy set, and then identify the retained strategies (excellent individuals) after screening; then reconstruct the selection probability of each strategy in the subsequent iteration process according to the excellent individual identification. Generally speaking, excellent individuals with higher fitness values are given higher iteration probabilities; based on the reconstructed iteration probabilities, use random sampling to select individuals (i.e., strategies) from the initial strategy set, and select a certain number of strategies as the basis for the next round of iteration.

[0052] Preferably, perform a gene exchange operation on the strategies in the individual selection result, that is, exchange and combine some gene segments of different strategies to generate new strategies. Here, a gene refers to a parameter in a strategy, similar to gene recombination in biological evolution. The purpose is to create more excellent strategies by combining the advantages of different strategies. For example, among two investment strategies, one focuses on short-term revenue and the other focuses on risk control. Through gene exchange, a new strategy that considers both short-term revenue and risk control is generated; then use the optimization constraints (i.e., the constraint conditions based on feedback signals) to update the contribution degrees of each parameter in the individual selection result. Here, the parameter contribution degree reflects the importance of each parameter in the strategy for achieving the optimization goal. By analyzing the performance of the strategy under the optimization constraints, adjust the contribution degrees of each parameter so that in subsequent iterations, the parameters that contribute more to the optimization goal can be adjusted more. Finally, randomly perturb the parameters of the individual selection result (i.e., the new strategy) after gene exchange and adjusting the parameter contribution degrees, that is, randomly change the parameter values of the strategy within a certain range to explore more possibilities in the strategy space and avoid the algorithm falling into a local optimal solution. For example, randomly increase or decrease the investment ratio parameter in the investment strategy and observe the performance of the strategy under the new parameters, thereby completing an iterative update of the initial strategy set. Compared with the initial strategy set, the fitness and the degree of approaching the optimal solution are improved.

[0053] Furthermore, the specific configuration of the competition optimization layer 40 further includes configuring a quality screening threshold, and screening the adaptive strategies of the adaptation evaluation results according to the quality screening threshold; after eliminating the adaptive strategies that do not meet the quality screening threshold, update the adaptive strategy set with the retained adaptive strategies.

[0054] Preferably, the quality screening threshold is a standard set according to actual business requirements and is used to measure the quality of adaptation strategies. Each adaptation strategy is evaluated by comparing its various indicators with the pre-set quality screening threshold. For example, for an investment strategy, if the set annual return rate threshold is 10% and the risk volatility threshold is 15%, then the actual annual return rate and risk volatility of each strategy are checked. If the annual return rate of a certain strategy is lower than 10% or the risk volatility is higher than 15%, it is considered not to meet the quality screening threshold. After comparison with the threshold, the adaptation strategies that do not meet the requirements are removed from the original strategy set, and only those strategies whose indicators all reach or exceed the quality screening threshold are retained to form an updated adaptation strategy set, improving the quality and effectiveness of the overall strategy.

[0055] Furthermore, the specific configuration of the competition optimization layer 40 further includes obtaining the behavior pattern database of the competition agent, performing backward reasoning based on the behavior pattern database to predict the potential coping strategies of the competition agent; establishing a mapping game strategy according to the potential coping strategies; and compensating the game optimization result according to the mapping game strategy.

[0056] Preferably, the behavior pattern database of the competition agent is established by collecting and organizing various behavior data of the competition agent in previous game processes, including information such as its decision-making choices, action methods, and reaction times in different situations. Backward reasoning is performed based on the behavior pattern database, that is, starting from the past behavior results of the competition agent and using the data in the behavior pattern database to analyze the motivation, goal, and decision-making rules behind its behavior. For example, if it is found that a competitor suddenly lowered the product price in a certain market area and obtained a higher market share, the possible motivation can be speculated by analyzing the relevant data in the database, such as to exclude other competitors, expand the market share, or to clear inventory, etc. Then, combined with the current game situation and information, the possible coping strategies that the competition agent may adopt in the future are predicted.

[0057] Preferably, conduct an in-depth analysis of the potential countermeasures of the predicted competing agents, understand the strategy objectives, advantages, disadvantages, and possible impacts on oneself, and then combine them with one's own game objectives and resources according to the analysis results of the potential countermeasures to establish a mapping game strategy, that is, determine what game strategy one should adopt to respond to each potential countermeasure. For example, if a competitor adopts a low-price strategy, one can choose to respond by optimizing product costs, providing differentiated value-added services, strengthening brand marketing, etc., so as to maintain market share and brand image while maintaining profits, and then clarify the specific actions and decision-making plans that one should take in different competitive situations in response to different potential countermeasures of competitors; finally, compensate for the game optimization results according to the mapping game strategy, which may include adjusting strategies, optimizing resource allocation, strengthening actions in certain aspects, etc. Through continuous evaluation and compensation, the game results can be gradually optimized, and one's competitiveness and benefits in the game can be improved.

[0058] Further, the specific configuration of the data processing layer 20 also includes extracting the data source after data cleaning into structured data and unstructured data; using a deep learning model to perform cross-modal feature fusion on the structured data and unstructured data; and performing adaptive feature selection on the cross-modal feature fusion result to complete data feature extraction.

[0059] Preferably, perform data cleaning on the data source, such as handling missing values, correcting incorrect data, removing duplicate data, etc., to improve data quality, and then divide the cleaned data source into structured data and unstructured data according to the data format and structural characteristics. Structured data refers to data with a clear structure and fixed format, such as tabular data in a database, where each column has a specific data type and meaning; unstructured data has no fixed structure, such as text, images, audio, video, etc. Then, input the structured data and unstructured data into the corresponding deep learning models for feature extraction respectively, and fuse the extracted features of different modalities. Among them, deep learning models (such as convolutional neural networks, recurrent neural networks, or long short-term memory networks) can automatically extract valuable features from a large amount of data. For example, for a dataset containing images and text descriptions, first use a convolutional neural network (CNN) to extract visual features of the images, such as color, texture, shape, etc., and then use a long short-term memory network (LSTM) to extract semantic features of the text, such as the meaning of words, the structure of sentences, etc., and splice the two feature vectors to form cross-modal features. Finally, according to the characteristics of the data and the task requirements, automatically select the most representative data features, which can better reflect the essential information of the original data and have higher quality and efficiency.

[0060] Further, the strategy dynamic optimization system based on big data analysis and prediction further includes an early warning layer for parsing the dynamic response strategy, generating a risk early warning signal, and giving a risk early warning for the risk early warning signal.

[0061] Preferably, the early warning layer parses the dynamic response strategy, that is, understands various rules, conditions, and corresponding operation instructions included in the strategy, and generates a corresponding risk early warning signal according to the current actual operation situation to prompt potential risks, including key information about the risks, such as risk types, risk levels, possible influence ranges, etc. Finally, the risk early warning signal is reported so as to obtain risk information in a timely manner and take corresponding measures to reduce the losses that may be brought by the risks, thereby ensuring that the strategy becomes stronger in an extremely volatile environment.

[0062] In the above, reference is made to Figure 1 The strategy dynamic optimization system based on big data analysis and prediction according to an embodiment of the present invention is described in detail. Next, reference will be made to Figure 2 Describe the strategy dynamic optimization method based on big data analysis and prediction according to an embodiment of the present invention. The strategy dynamic optimization method based on big data analysis and prediction, as Figure 2 shown, the method includes: collecting transaction data, news data, social data, search data, public data, and private domain user retention data to establish a data source; after cleaning the data source, performing data feature extraction, identifying abnormal patterns according to the data feature extraction results, and constructing a strategy training environment; after injecting random noise into the strategy training environment, performing coping strategy training in an adversarial environment to establish an adaptation strategy set; configuring a competition intelligent agent according to the data source, the competition intelligent agent includes a leader, a follower, and an adversary, after performing an adaptation evaluation of the strategy quality on the adaptation strategy set, updating the adaptation strategy set, and using the updated adaptation strategy set and the competition intelligent agent for game optimization to generate a game optimization result; receiving the game optimization result and establishing a dynamic response strategy.

[0063] In a possible implementation manner, the strategy dynamic optimization method based on big data analysis and prediction further includes: establishing an extreme scenario data set, and using a generative adversarial network to expand the extreme scenario data set to generate an updated extreme scenario data set; performing self-check on the injection of the random noise within a preset period, performing a scenario coverage evaluation according to the self-check result and the updated extreme scenario data set, and establishing an injection constraint for the random noise; configuring a time continuous perturbation constraint and establishing risk factors, the risk factors include liquidity risk and credit risk; injecting random noise into the strategy training environment according to the injection constraint of the random noise, the time continuous perturbation constraint, and the risk factors.

[0064] In a possible implementation, the method for dynamically optimizing a strategy based on big data analysis and prediction further includes: monitoring indicators of the strategy optimization subject, establishing a set of key indicators, where the set of key indicators includes yield, risk exposure, liquidity, and resource utilization efficiency; establishing the current state of the strategy optimization subject according to the set of key indicators; using the current state and the updated set of adaptation strategies as input data to perform game optimization with the competing intelligent agent.

[0065] In a possible implementation, the method for dynamically optimizing a strategy based on big data analysis and prediction further includes: recording the revenue data, risk data, and decision-making performance data of all roles during the game process to establish a feedback signal; using the updated set of adaptation strategies as the initial strategy set, and using the feedback signal as the optimization constraint to perform iterative update of the initial strategy set; completing game optimization according to the iterative update result.

[0066] In a possible implementation, the method for dynamically optimizing a strategy based on big data analysis and prediction further includes: calculating the fitness values in the initial strategy set, using the fitness values to screen the initial strategy set, and establishing excellent individual identifiers; after reconstructing the iterative probability according to the excellent individual identifiers, performing individual selection of the initial strategy set; performing gene exchange according to the individual selection result, and updating the parameter contribution degree of the individual selection result using the optimization constraint; after randomly perturbing the parameters of the individual selection result, completing one iteration update, where the current individual selection result is the individual selection result after gene exchange and adjustment of the parameter contribution degree.

[0067] In a possible implementation, the method for dynamically optimizing a strategy based on big data analysis and prediction further includes: configuring a quality screening threshold, and screening the adaptation strategies of the adaptation evaluation results according to the quality screening threshold; after removing the adaptation strategies that do not meet the quality screening threshold, updating the set of adaptation strategies with the remaining adaptation strategies.

[0068] In a possible implementation, the method for dynamically optimizing a strategy based on big data analysis and prediction further includes: extracting the data source after data cleaning into structured data and unstructured data; using a deep learning model to perform cross-modal feature fusion of the structured data and unstructured data; performing adaptive feature selection on the cross-modal feature fusion result to complete data feature extraction.

[0069] In a possible implementation, the method for dynamically optimizing a strategy based on big data analysis and prediction further includes: obtaining the behavior pattern database of the competing intelligent agent, performing backward reasoning according to the behavior pattern database, and predicting the potential coping strategies of the competing intelligent agent; establishing a mapping game strategy according to the potential coping strategies; performing compensation for the game optimization result according to the mapping game strategy.

[0070] In a possible implementation, the method for dynamically optimizing a policy based on big data analysis and prediction further includes: a warning layer, configured to parse the dynamic response policy, generate a risk warning signal, and issue the risk warning signal for risk warning.

[0071] The system for dynamically optimizing a policy based on big data analysis and prediction provided by the embodiments of the present invention can execute the method for dynamically optimizing a policy based on big data analysis and prediction provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects for executing the method.

[0072] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, however, any number of different modules can be used and run on a user terminal and / or a server. The included respective units and modules are only divided according to functional logic, but are not limited to the above division as long as the corresponding functions can be implemented; in addition, the specific names of the respective functional units are only for facilitating mutual distinction and do not limit the protection scope of the present invention.

[0073] The above specific implementation manners do not constitute a limitation on the protection scope of the present application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principle of the present application shall be included within the protection scope of the present application.

Claims

1. A strategy dynamic optimization system based on big data analysis and prediction, characterized in that, The system includes: A data acquisition layer for collecting transaction data, news data, social data, search data, public data, and private domain user retention data to establish a data source; A data processing layer for performing data cleaning on the data source, then extracting data features, identifying abnormal patterns based on the data feature extraction results, and constructing a policy training environment; An extreme scenario simulation layer for performing coping strategy training in an adversarial environment after injecting random noise into the policy training environment to establish an adaptation strategy set; A competition optimization layer for configuring competition agents according to the data source, where the competition agents include a leader, a follower, and an adversary. After performing an adaptation evaluation of the strategy quality on the adaptation strategy set, updating the adaptation strategy set, and using the updated adaptation strategy set and the competition agents to perform game optimization to generate a game optimization result; A policy dynamic optimization layer for receiving the game optimization result and establishing a dynamic response policy.

2. The strategy dynamic optimization system based on big data analysis and prediction according to claim 1, characterized in that The extreme scenario simulation layer is further used for: Establishing an extreme scenario dataset and using a generative adversarial network to expand the extreme scenario dataset to generate an updated extreme scenario dataset; Performing self-check on the injection of the random noise within a preset period, performing scenario coverage evaluation according to the self-check result and the updated extreme scenario dataset, and establishing an injection constraint for the random noise; Configuring a time-continuous perturbation constraint and establishing risk factors, where the risk factors include liquidity risk and credit risk; Performing random noise injection into the policy training environment according to the injection constraint of the random noise, the time-continuous perturbation constraint, and the risk factors.

3. The strategy dynamic optimization system based on big data analysis prediction according to claim 1, characterized in that The competition optimization layer is further used for: Monitoring indicators of a policy optimization entity, establishing a key indicator set, where the key indicator set includes return rate, risk exposure, liquidity, and resource utilization efficiency; Establishing the current state of the policy optimization entity according to the key indicator set; Using the current state and the updated adaptation strategy set as input data to perform game optimization with the competition agents.

4. The strategy dynamic optimization system based on big data analysis prediction according to claim 3, characterized in that, In the competition optimization layer, using the current state and the updated adaptation strategy set as input data to perform game optimization with the competition agents includes: Recording the revenue data, risk data, and decision-making performance data of all roles during the game process to establish a feedback signal; Using the updated adaptation strategy set as the initial strategy set and using the feedback signal as an optimization constraint to perform iterative update of the initial strategy set; Completing game optimization according to the iterative update result.

5. The strategy dynamic optimization system based on big data analysis prediction according to claim 4, characterized in that, In the competition optimization layer, using the feedback signal as an optimization constraint to perform iterative update of the initial strategy set includes: Calculating the fitness values in the initial strategy set, screening the initial strategy set using the fitness values, and establishing excellent individual identifiers; Reconstructing the iteration probability according to the excellent individual identifiers and then performing individual selection of the initial strategy set; Performing gene exchange according to the individual selection result and updating the parameter contribution degree of the individual selection result using the optimization constraint; After randomly perturbing the parameters of the individual selection result, completing one iteration update, where the current individual selection result is the individual selection result after gene exchange and adjustment of the parameter contribution degree.

6. The strategy dynamic optimization system based on big data analysis and prediction according to claim 1, characterized in that In the competition optimization layer, after performing an adaptation evaluation on the quality of the adaptation strategy set, the adaptation strategy set is updated, including: Configuring a quality screening threshold, and screening adaptation strategies based on the adaptation evaluation results according to the quality screening threshold; After removing the adaptation strategies that do not meet the quality screening threshold, the adaptation strategy set is updated with the remaining adaptation strategies.

7. The strategy dynamic optimization system based on big data analysis and prediction according to claim 1, characterized in that In the data processing layer, after cleaning the data source, data feature extraction is performed, including: Extracting the data source after data cleaning into structured data and unstructured data; Using a deep learning model for cross-modal feature fusion of structured data and unstructured data; Performing adaptive feature selection on the cross-modal feature fusion results to complete data feature extraction.

8. The strategy dynamic optimization system based on big data analysis and prediction according to claim 1, characterized in that The competition optimization layer is further used for: Obtaining the behavior pattern database of the competition agent, and performing backward reasoning based on the behavior pattern database to predict the potential coping strategies of the competition agent; Establishing a mapping game strategy according to the potential coping strategies; Compensating the game optimization result according to the mapping game strategy.

9. The strategy dynamic optimization system based on big data analysis and prediction according to claim 1, characterized in that, The system further includes: An early warning layer, which is used to analyze the dynamic response strategy, generate a risk early warning signal, and issue the risk early warning signal for risk early warning.

10. A method for dynamically optimizing strategies based on big data analysis and prediction, characterized in that The method is applied to the strategy dynamic optimization system based on big data analysis prediction according to any one of claims 1-9. The method includes: Collecting transaction data, news data, social data, search data, public data, and private domain user retention data to establish a data source; After cleaning the data source, performing data feature extraction, identifying abnormal patterns according to the data feature extraction results, and constructing a strategy training environment; After injecting random noise into the strategy training environment, performing coping strategy training in an adversarial environment to establish an adaptation strategy set; Configuring competition agents according to the data source. The competition agents include a leader, a follower, and an adversary. After performing an adaptation evaluation on the quality of the adaptation strategy set, the adaptation strategy set is updated, and game optimization is performed using the updated adaptation strategy set and the competition agents to generate a game optimization result; Receiving the game optimization result and establishing a dynamic response strategy.

Images