Real-time credit scoring system based on deep learning

Abstract

The invention relates to the technical field of credit scoring, in particular to a real-time credit scoring system based on deep learning, which comprises a data collection module, a data preprocessing module, a deep learning model module and a real-time scoring module, and is characterized in that the data preprocessing module is connected with the data collection module, and the deep learning model module is connected with the data preprocessing module; the real-time scoring module is connected with the deep learning model module; an advanced deep learning algorithm is introduced to process real-time data from a plurality of heterogeneous data sources, such as financial transaction records, social media behaviors, online consumption habits and the like of users, so that a comprehensive, dynamic and real-time updated credit score is generated. According to the system, multiple innovations are carried out in the aspects of the flexibility of an algorithm, the universality of data processing and the real-time performance of risk prediction, and a more comprehensive and efficient credit risk assessment tool is provided for financial institutions.

Description

Technical Field

[0001] The present invention relates to the technical field of credit scoring, and in particular to a real-time credit scoring system based on deep learning. Background Art

[0002] Credit scoring systems have become a critical decision-making tool for financial institutions and businesses. However, traditional credit scoring models often rely on static historical data, such as users' credit card repayment records, loan history, and bank account activity. This historical data-based assessment approach has several significant limitations. First, data lags prevent credit scores from reflecting users' latest credit status in real time. This lag can lead to poor decision-making, especially in a rapidly changing economic environment. Second, traditional credit scoring models are often based on simple linear regression or decision tree algorithms. While easy to implement, these models perform poorly when dealing with complex, multidimensional data. In modern society, a user's credit status is influenced by multiple factors, which may have complex nonlinear relationships. Traditional scoring models struggle to effectively capture these relationships, resulting in inaccurate scoring results. With the development of big data technology, the sources of user behavior data have become more diverse, expanding beyond financial data to include social media behavior, online shopping habits, and geolocation data. The introduction of these new types of data provides a more comprehensive perspective for credit scoring, but also increases the complexity of data processing, making traditional models difficult to address.

[0003] In recent years, the rise of deep learning technology has brought new opportunities for the development of credit scoring systems. By simulating the structure of the human brain's neural networks, deep learning can effectively process high-dimensional and nonlinear data, particularly excelling in areas such as image recognition, natural language processing, and financial analysis. In the field of credit scoring, deep learning can generate more accurate and dynamic credit scores by learning complex patterns from massive amounts of historical data and analyzing users' latest behavioral data in real time.

[0004] In the future, with the further development of deep learning algorithms, especially the application of cutting-edge technologies such as self-supervised learning, generative adversarial networks (GANs), and graph neural networks (GNNs), credit scoring systems are expected to achieve a higher level of intelligence. This will not only improve the accuracy and real-time performance of credit scoring, but also make the system more adaptable and robust, allowing it to better cope with complex and volatile economic environments. In short, real-time credit scoring systems based on deep learning are a significant improvement over traditional credit scoring models. They can effectively compensate for data lags and model limitations, provide financial institutions and enterprises with more accurate and dynamic credit risk assessments, and enhance the efficiency and security of overall financial services. Summary of the Invention

[0005] The purpose of this invention is to provide a real-time credit scoring system based on deep learning to address the various limitations of traditional credit scoring methods and achieve more accurate and dynamic credit risk assessment.

[0006] To achieve the above objectives, the present invention provides a real-time credit scoring system based on deep learning, comprising a data collection module, a data preprocessing module, a deep learning model module, and a real-time scoring module, wherein the data preprocessing module is connected to the data collection module, the deep learning model module is connected to the data preprocessing module, and the real-time scoring module is connected to the deep learning model module;

[0007] The data collection module is used to collect multi-dimensional data of users in real time using stream processing technology and store the data in a data warehouse through a distributed database;

[0008] The data preprocessing module is used to clean and normalize the collected data and automatically generate higher-level features based on the collected data using automatic feature generation technology. After further dimensionality reduction, the features are used as input for the deep learning model;

[0009] The deep learning model module is used to process the pre-processed data using deep learning technology and generate an accurate credit score;

[0010] The real-time scoring module is used to convert the output of the deep learning model module into an actionable credit score and respond to changes in user behavior in real time.

[0011] The data collection module includes a multi-source heterogeneous data real-time collection unit and a data flow management storage unit, and the data flow management storage unit is connected to the multi-source heterogeneous data real-time collection unit and the data preprocessing module respectively;

[0012] The multi-source heterogeneous data real-time collection unit is used to establish connections with API interfaces of multiple data sources to collect multi-dimensional data of users in real time;

[0013] The data stream management storage unit is used to introduce stream processing technology to process data streams, and at the same time store the collected data in a data warehouse through a distributed database.

[0014] The multi-source heterogeneous data real-time collection unit includes a financial data source subunit, an e-commerce platform data subunit, a social network behavior data subunit, a geographic location information subunit, and a device usage data subunit, and the financial data source subunit, the e-commerce platform data subunit, the social network behavior data subunit, the geographic location information subunit, and the device usage data subunit are respectively connected to the data flow management and storage unit;

[0015] The financial data source subunit is used to establish a real-time connection with financial institutions to collect users' bank transaction records, credit card usage, loan application records, and repayment history;

[0016] The e-commerce platform data sub-unit is used to obtain users' consumption records on the e-commerce platform in real time through the API interface, including shopping habits, purchase frequency, payment methods and consumption amounts;

[0017] The social network behavior data subunit is used to collect user behavior on social media platforms;

[0018] The geographic location information subunit is used to obtain the user's location information through the geographic location service of the user equipment;

[0019] The device usage data subunit is used to collect the usage of the user's device, including usage frequency, device type, operating system version and network connection mode.

[0020] Wherein, the data preprocessing module includes a data cleaning and normalization unit and an advanced feature engineering unit, the data cleaning and normalization unit is connected to the data flow management and storage unit, and the advanced feature engineering unit is connected to the data cleaning and normalization unit;

[0021] The data cleaning and normalization unit is used to clean the collected raw data, delete invalid data, correct erroneous records, and format and normalize the cleaned data;

[0022] The advanced feature engineering unit is used to adopt automatic feature generation technology to automatically generate higher-level features based on the normalized data, and further perform dimensionality reduction on the features before using them as input to the deep learning model.

[0023] The deep learning model module includes a hybrid deep neural network architecture and an adaptive learning dynamic update unit, the hybrid deep neural network architecture is connected to the advanced feature engineering unit, and the adaptive learning dynamic update unit is connected to the hybrid deep neural network architecture;

[0024] The hybrid deep neural network architecture is used to integrate deep learning technologies such as convolutional neural networks, long short-term memory networks, graph neural networks, and transformer models based on self-attention mechanisms to form a model system. In particular, a hierarchical attention mechanism is used to integrate different networks in the following ways:

[0025] Convolutional neural networks are used to extract local pattern features of user behavior data;

[0026] Long short-term memory networks are used to capture the time series characteristics of user behavior;

[0027] Graph neural networks are used to model user social relationship networks;

[0028] Transformer model for global feature interaction and attention allocation;

[0029] Through the multimodal feature fusion layer, the outputs of different networks are adaptively weighted and fused;

[0030] The adaptive learning dynamic update unit is used to perform online learning, incremental training and reinforcement learning on the model to ensure that the model maintains high accuracy and stability in a constantly changing data environment.

[0031] The real-time scoring module includes an efficient real-time scoring engine and a risk warning feedback unit, wherein the efficient real-time scoring engine is connected to the hybrid deep neural network architecture, and the risk warning feedback unit is connected to the efficient real-time scoring engine;

[0032] The efficient real-time scoring engine is used to monitor changes in user behavior data in real time, process the changing data stream, and generate credit scores in real time. It uses a stream processing architecture based on in-memory computing, combined with a lightweight neural network inference engine, to achieve millisecond-level response, with an average response time of less than 50ms and 99.9% of requests completed within 100ms.

[0033] The risk warning feedback unit automatically triggers a risk warning mechanism when a user's credit score falls below a set threshold. It uses a dynamic threshold adjustment algorithm to automatically adjust the risk threshold based on real-time market data and user behavior patterns, increasing warning accuracy by 30% and reducing false alarm rates by 45%.

[0034] The real-time credit scoring system based on deep learning further includes a model optimization monitoring module, which is connected to the hybrid deep neural network architecture;

[0035] The model optimization monitoring module is used to automatically adjust the hyperparameters and structure of the deep learning model, and to monitor the scoring accuracy and computational efficiency of the model in real time.

[0036] The model optimization and monitoring module includes an automated model optimization unit and a model performance monitoring and anomaly detection unit, and the automated model optimization unit and the model performance monitoring and anomaly detection unit are respectively connected to the hybrid deep neural network architecture;

[0037] The automated model optimization unit is used to integrate reinforcement learning algorithms and automatically adjust the hyperparameters and structure of deep learning models. The automated hyperparameter search algorithm based on Bayesian optimization, combined with the online adjustment strategy of reinforcement learning, improves model optimization efficiency by 60% and model performance by 15%.

[0038] The model performance monitoring anomaly detection unit is used to monitor the scoring accuracy and computational efficiency of the model in real time. If abnormal fluctuations in the model score are detected, anomaly detection and analysis are automatically performed, and corresponding corrective measures are taken. Among them, an anomaly detection algorithm based on time series analysis is used, combined with an adaptive threshold adjustment mechanism, and the anomaly detection accuracy rate reaches 95%, with an average detection time of less than 1 second.

[0039] The present invention provides a real-time credit scoring system based on deep learning. The deep learning model module adopts an advanced deep learning algorithm and combines a low-latency scoring architecture and a dynamic feature update mechanism to achieve a credit scoring response time of milliseconds. This real-time processing capability greatly improves the timeliness of credit assessment, enabling financial institutions to obtain the latest credit scores instantly and make timely decisions. This is particularly important in a rapidly changing economic environment, helping institutions reduce risks and seize market opportunities. The data collection module greatly enriches the basic data for credit scoring by collecting and integrating user information from multiple heterogeneous data sources such as financial transactions, social networks, geographic locations, and device usage. This diverse data collection method, combined with the powerful processing power of the deep learning model, enables the system to capture subtle changes in user behavior and reflect them in the score. Compared with traditional credit scoring systems that rely solely on financial data, the present invention significantly improves the accuracy and comprehensiveness of the score and can more accurately reflect the user's true credit status. The real-time scoring module ensures that the credit scoring model can continuously adapt to changes in user behavior through adaptive learning and dynamic update mechanisms. By using online learning and incremental training techniques, the model can be updated in real time when new data arrives, avoiding the lag and inadaptability of traditional models. In addition, the system continuously optimizes the model's predictive capabilities through reinforcement learning and closed-loop feedback mechanisms, further improving the reliability and robustness of the scoring. This adaptability enables the system to maintain efficient operation in different market environments and economic cycles. The real-time scoring module has a built-in intelligent risk warning system that can automatically trigger an alarm when a user's credit score falls below a specific threshold, issuing real-time reminders to financial institutions or users themselves. Through the adaptive threshold adjustment mechanism, the system can dynamically adjust the risk threshold based on historical data and real-time feedback to optimize the warning effect. This function not only helps financial institutions detect potential risks in advance, but also helps them take timely risk control measures to reduce possible losses. The automated model optimization unit integrates reinforcement learning algorithms to regularly adjust the hyperparameters and structure of the deep learning model, ensuring the model remains efficient over the long term. Simultaneously, the model performance monitoring and anomaly detection unit monitors the model's scoring accuracy and computational efficiency in real time, automatically taking corrective measures when abnormal fluctuations occur. These optimization and monitoring capabilities enhance the system's stability and sustainability, enabling financial institutions to rely on it for long-term credit risk management. This addresses the various limitations of traditional credit scoring methods and enables more accurate and dynamic credit risk assessments. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art.

[0041] Figure 1This is a principle block diagram of a real-time credit scoring system based on deep learning according to the first embodiment of the present invention.

[0042] Figure 2 It is a principle block diagram of the data collection module and the data preprocessing module of the first embodiment of the present invention.

[0043] Figure 3 This is a principle block diagram of the deep learning model module and the real-time scoring module of the first embodiment of the present invention.

[0044] Figure 4 It is a principle block diagram of the model optimization monitoring module of the second embodiment of the present invention.

[0045] In the figure: 101-data collection module, 102-data preprocessing module, 103-risk warning feedback unit, 104-deep learning model module, 105-real-time scoring module, 106-multi-source heterogeneous data real-time collection unit, 107-data stream management and storage unit, 108-financial data source sub-unit, 109-e-commerce platform data sub-unit, 110-social network behavior data sub-unit, 111-geographic location information sub-unit, 112-device usage data sub-unit, 113-data cleaning and normalization unit, 114-advanced feature engineering unit, 115-hybrid deep neural network architecture, 116-adaptive learning dynamic update unit, 117-efficient real-time scoring engine, 201-model optimization and monitoring module, 202-automatic model optimization unit, 203-model performance monitoring and anomaly detection unit. DETAILED DESCRIPTION

[0046] The embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, but should not be understood as limiting the present invention.

[0047] The first embodiment of this application is:

[0048] See also Figures 1 to 3 ,in, Figure 1 This is a principle block diagram of a real-time credit scoring system based on deep learning according to the first embodiment of the present invention. Figure 2 It is a principle block diagram of the data collection module 101 and the data preprocessing module 102 according to the first embodiment of the present invention. Figure 3 It is a principle block diagram of the deep learning model module 104 and the real-time scoring module 105 of the first embodiment of the present invention.

[0049] The present invention provides a real-time credit scoring system based on deep learning, comprising a data collection module 101, a data preprocessing module 102, a deep learning model module 104, and a real-time scoring module 105. The data collection module 101 comprises a multi-source heterogeneous data real-time collection unit 106 and a data stream management and storage unit 107. The multi-source heterogeneous data real-time collection unit 106 comprises a financial data source subunit 108, an e-commerce platform data subunit 109, a social network behavior data subunit 110, a geographic location information subunit 111, and a device usage data subunit 112. The data preprocessing module 102 comprises a data cleaning and normalization unit 113 and an advanced feature engineering unit 114. The deep learning model module 104 comprises a hybrid deep neural network architecture 115 and an adaptive learning dynamic update unit 116. The real-time scoring module 105 comprises an efficient real-time scoring engine 117 and a risk warning feedback unit 103.

[0050] For this specific embodiment, the data preprocessing module 102 is connected to the data collection module 101, the deep learning model module 104 is connected to the data preprocessing module 102, and the real-time scoring module 105 is connected to the deep learning model module 104;

[0051] The data collection module 101 is used to collect multi-dimensional data of users in real time using stream processing technology and store the data in a data warehouse through a distributed database;

[0052] The data preprocessing module 102 is used to clean and normalize the collected data, and use automatic feature generation technology to automatically generate higher-level features based on the collected data. After further dimensionality reduction, the features are used as input for the deep learning model;

[0053] The deep learning model module 104 is used to process the pre-processed data using deep learning technology and generate an accurate credit score;

[0054] The real-time scoring module 105 is used to convert the output of the deep learning model module 104 into an actionable credit score and respond to changes in user behavior in real time.

[0055] The data collection module 101 uses stream processing technology to collect multi-dimensional user data from various data sources in real time. This data is then stored in a distributed database in a data warehouse for subsequent processing and analysis, ensuring data scalability and accessibility. The data preprocessing module 102 cleans the raw data stored in the data warehouse to improve data quality. It normalizes the cleaned data to enable comparison and analysis of data of different dimensions and ranges on the same scale. It then utilizes automatic feature generation technology to automatically generate higher-level features based on the raw data. These features can more comprehensively reflect the user's credit status and improve the model's predictive capabilities. The generated features are then further subjected to dimensionality reduction processing, such as PCA and LDA, to reduce the model's computational complexity and improve its generalization capabilities. Ultimately, the processed feature data serves as input for the deep learning model. The deep learning model module 104 uses deep learning technology to train and learn the preprocessed data to capture the complex relationship between the user's credit status and data features. By continuously iterating and optimizing model parameters, the model's predictive accuracy and stability are improved. Ultimately, the deep learning model is able to generate accurate credit scores based on the input feature data. The real-time scoring module 105 converts the output of the deep learning model, i.e., the predicted credit score, into an actionable credit score result and displays it to the user or related system in real time. It responds to changes in user behavior in real time, dynamically adjusts the credit score based on the user's latest data, and can continuously monitor user behavior and trigger data collection, preprocessing, and scoring processes in real time.

[0056] The data flow management storage unit 107 is connected to the multi-source heterogeneous data real-time collection unit 106 and the data pre-processing module 102 respectively;

[0057] The multi-source heterogeneous data real-time collection unit 106 is used to establish connections with API interfaces of multiple data sources to collect multi-dimensional data of users in real time;

[0058] The data stream management storage unit 107 is used to introduce stream processing technology to process data streams, and at the same time store the collected data in a data warehouse through a distributed database.

[0059] The multi-source heterogeneous data real-time collection unit 106 establishes connections with API interfaces of multiple data sources to collect multi-dimensional data of users in real time, including but not limited to bank transaction records, credit card usage, e-commerce platform consumption records, social network behavior data, geographic location information, and user device usage data. These data sources are not limited to the financial field but also cover users' daily behavior, thus providing a more comprehensive foundation for credit scoring. To ensure the validity and timeliness of data, the data stream management and storage unit 107 introduces stream processing technology, which can handle high-throughput data streams. At the same time, the data is stored in the data warehouse through a distributed database to ensure data security and traceability.

[0060] Secondly, the financial data source subunit 108, the e-commerce platform data subunit 109, the social network behavior data subunit 110, the geographic location information subunit 111, and the device usage data subunit 112 are respectively connected to the data flow management storage unit 107;

[0061] The financial data source subunit 108 is used to establish a real-time connection with financial institutions to collect users' bank transaction records, credit card usage, loan application records, and repayment history;

[0062] The e-commerce platform data subunit 109 is used to obtain the user's consumption records on the e-commerce platform in real time through the API interface, including shopping habits, purchase frequency, payment method and consumption amount;

[0063] The social network behavior data subunit 110 is used to collect user behavior on the social media platform;

[0064] The geographic location information subunit 111 is used to obtain the user's location information through the geographic location service of the user equipment;

[0065] The device usage data subunit 112 is used to collect usage information of user devices, including usage frequency, device type, operating system version, and network connection mode.

[0066] The financial data source subunit 108 establishes real-time connections with financial institutions such as banks, credit card companies, and lending institutions to collect information including, but not limited to, users' bank transaction records, credit card usage, loan application records, and repayment history. This data is an important foundation for credit scoring and reflects users' financial health and borrowing behavior. The e-commerce platform data subunit 109 uses an API interface to obtain real-time user consumption records on major e-commerce platforms, including shopping habits, purchase frequency, payment methods, and spending amounts. This data can reflect users' spending power and consumption patterns, thereby providing additional credit information. The social network behavior data subunit 110 collects user behavior on social media platforms, such as posts, comments, likes, and interactions within social networks. Social behavior can provide deep insights into user behavior patterns, such as the user's social activity and the strength of their relationships with other users. These characteristics can help assess the user's risk profile. The geographic location information subunit 111 obtains user location information through the geolocation service of user devices, such as smartphones. This information can be used to analyze user behavior patterns, such as travel patterns, living areas, and frequented locations, further enriching the basis for credit scoring. The device usage data subunit 112 collects the usage of user devices, including usage frequency, device type, operating system version, network connection mode, etc. These data can reflect the user's technology usage level and preferences, thereby indirectly providing some clues to their economic activities.

[0067] At the same time, the data cleaning and normalization unit 113 is connected to the data flow management storage unit 107, and the advanced feature engineering unit 114 is connected to the data cleaning and normalization unit 113;

[0068] The data cleaning and normalization unit 113 is used to clean the collected raw data, delete invalid data, correct erroneous records, and format and normalize the cleaned data;

[0069] The advanced feature engineering unit 114 is used to adopt automatic feature generation technology to automatically generate higher-level features based on the normalized data, and further perform dimensionality reduction on the features as input to the deep learning model.

[0070] Because data formats and quality may vary from source to source, the data cleaning and normalization unit 113 first cleans the collected raw data, deleting invalid data and correcting erroneous records. Subsequently, the data is formatted and normalized to ensure that data from different sources can be uniformly input into the deep learning model. The advanced feature engineering unit 114, building on traditional feature engineering, introduces automatic feature generation technology to automatically generate higher-level features based on user behavior data, such as user consumption patterns, social network activity, and location change frequency. These features, after further dimensionality reduction processing, serve as input to the deep learning model, significantly improving the accuracy of credit scoring.

[0071] In addition, the hybrid deep neural network architecture 115 is connected to the advanced feature engineering unit 114, and the adaptive learning dynamic update unit 116 is connected to the hybrid deep neural network architecture 115;

[0072] The hybrid deep neural network architecture 115 is used to integrate deep learning technologies of convolutional neural networks, long short-term memory networks, graph neural networks, and transformer models based on self-attention mechanisms to form a model system, wherein a hierarchical attention mechanism is used to integrate different networks in the following manner:

[0073] Convolutional neural networks are used to extract local pattern features of user behavior data;

[0074] Long short-term memory networks are used to capture the time series characteristics of user behavior;

[0075] Graph neural networks are used to model user social relationship networks;

[0076] Transformer model for global feature interaction and attention allocation;

[0077] Through the multimodal feature fusion layer, the outputs of different networks are adaptively weighted and fused;

[0078] The adaptive learning dynamic updating unit 116 is used to perform online learning, incremental training and reinforcement learning on the model to ensure that the model maintains high accuracy and stability in a constantly changing data environment.

[0079] In order to fully capture the multi-dimensional characteristics of user credit risk, the hybrid deep neural network architecture 115 integrates a variety of advanced deep learning technologies such as convolutional neural networks (CNN), long short-term memory networks (LSTM), graph neural networks (GNN), and transformer models (Transformer) based on self-attention mechanisms to form a complex and efficient model system. Among them, convolutional neural networks (CNN): are mainly used to process time series data of user behavior, such as bank transaction records, changes in consumption habits, etc. Traditional credit scoring systems often ignore the characteristics of the time dimension, while CNN can capture the potential time series patterns in the data through convolution operations. CNN extracts local features in the time series through a one-dimensional convolution layer, and then performs feature dimensionality reduction through a pooling layer. For the input user behavior time series data X = {x1, x2,…, x T}, the features extracted by CNN are expressed as:

[0080] h t =ReLU(W c *x t:t+k-1 +b c )

[0081] Among them, w c and b c are the weight and bias of the convolution kernel respectively, * represents the convolution operation, h t is the feature vector of the tth time step. Finally, CNN inputs these feature vectors into the fully connected layer to generate high-level time series features.

[0082] Long Short-Term Memory (LSTM): This network is used to capture long-term dependencies in user behavior and is suitable for processing historical behavior data with a long span, such as a user's long-term repayment history and large-scale spending behavior. LSTM effectively prevents the gradient vanishing problem through its gating mechanism, allowing the model to learn long-term dependencies. For a given input sequence X = {x1, x2, ..., x T}, the state of the LSTM unit is updated as follows:

[0083] f t =σ(W f ·[h t-1 ,x t ]+b f )

[0084] i t =σ(W i ·[h t-1 ,x t ]+b i )

[0085] C t =f t⊙C t-1 +i t ⊙tanh(W C ·[h t-1 ,x t ]+b C )

[0086] o t =σ(W o ·[h t-1 ,x t ]+b o )

[0087] h t =o t ⊙tanh(C t )

[0088] Among them, f t 、i t and o t are the activation values ​​of the forget gate, input gate, and output gate respectively, C t is the cell state, h t is the output hidden state. LSTM captures the long-term dependency characteristics of user behavior through these state update equations.

[0089] Graph Neural Networks (GNNs) are mainly used to analyze the relationships and influence of users in social networks. Social relationships have an important impact on the credit risk assessment of users. GNNs can mine potential risk information by modeling the connection between users and their social networks. GNNs update the representation of nodes by aggregating information from neighboring nodes. For a social network graph The representation of each node v is obtained by aggregating its neighboring nodes:

[0090]

[0091] in, is the representation of node v at the kth layer, is the set of neighbor nodes of node v, AGGREGATE is the aggregation function, W (k) and b (k) is a trainable parameter. GNN generates feature representations that reflect the user’s global information in the social network through multi-layer aggregation.

[0092] The Transformer model is used to capture complex global dependencies in data, and performs particularly well when processing highly heterogeneous user behavior data. Compared with traditional sequence models, the Transformer can process input data in parallel through the self-attention mechanism, greatly improving efficiency. The self-attention mechanism in the Transformer generates weighted feature representations by calculating the correlation between different positions in the input sequence. For the input sequence X = {x1, x2, ..., x T}, the self-attention is calculated as follows:

[0093]

[0094] Among them, Q, K, and V are query, key, and value matrices respectively, and d k is the dimension of the key. Through the multi-head attention mechanism, the model can simultaneously capture multiple correlation patterns in the sequence and generate feature representations that contain global context information.

[0095] The adaptive learning dynamic update unit 116 includes multiple technologies such as online learning, incremental training and reinforcement learning to ensure that the model maintains high accuracy and stability in a constantly changing data environment. Among them, online learning allows the model to be updated in real time when new data arrives, without waiting for a large amount of data to be accumulated for batch training. For each newly arrived data point (x t ,y t ), the model is based on the current loss function Update parameters:

[0096]

[0097] Where η is the learning rate, is the gradient of the loss function with respect to the model parameters. Through online updates, the model can quickly adapt to dynamic changes in data, ensuring real-time scoring.

[0098] Incremental training allows the model to be further trained with incremental data while maintaining the previously learned results. Compared to full retraining, incremental training is more economical in terms of computing resources and avoids the forgetting effect of the model. Incremental training updates the model parameters based on the existing model parameters θ0 using new data:

[0099]

[0100] Among them, X old and X new Represent the feature matrices of old data and new data respectively.

[0101] Reinforcement Learning (RL) guides the model to update parameters through a reward mechanism, which is particularly suitable for long-term risk management in credit scoring. The model obtains a reward r at each time step t by interacting with the environment (i.e., user behavior). t , and update the policy π through the policy gradient method θ :

[0102]

[0103] Here, γ is the discount factor, representing the degree of decay of future rewards. Through reinforcement learning, the model can better assess users' long-term credit risk. In summary, the deep learning model module 104 integrates multiple advanced deep learning technologies and adaptive learning mechanisms to achieve a comprehensive and efficient assessment of user credit risk. This module not only captures complex user behavior patterns but also offers dynamic update capabilities, ensuring that credit scores remain highly accurate and timely.

[0104] Finally, the efficient real-time scoring engine 117 is connected to the hybrid deep neural network architecture 115, and the risk warning feedback unit 103 is connected to the efficient real-time scoring engine 117;

[0105] The efficient real-time scoring engine 117 is used to monitor changes in user behavior data in real time, process the changing data stream, and generate credit scores in real time. It uses a stream processing architecture based on in-memory computing, combined with a lightweight neural network inference engine, to achieve millisecond-level response, with an average response time of less than 50ms and 99.9% of requests completed within 100ms;

[0106] The risk warning feedback unit 103 automatically triggers a risk warning mechanism when a user's credit score falls below a set threshold. A dynamic threshold adjustment algorithm is used to automatically adjust the risk threshold based on real-time market data and user behavior patterns, thereby increasing the warning accuracy by 30% and reducing the false alarm rate by 45%.

[0107] To achieve millisecond-level response times, the efficient real-time scoring engine 117 utilizes a low-latency architecture that combines in-memory computing and parallel processing technologies. Core technologies include stream processing frameworks such as Apache Flink and Apache Spark Streaming, which efficiently process large-scale data streams and generate credit scores in real time. The efficient real-time scoring engine 117 leverages in-memory data caching to avoid frequent disk I / O operations, significantly reducing scoring latency. The efficient real-time scoring engine 117 not only relies on pre-trained weights from deep learning models but also dynamically updates input features. This functionality is supported by a feature extraction and update subsystem that monitors changes in user behavior data in real time and immediately reflects these changes to the scoring engine. For example, the system dynamically updates features such as user spending frequency and borrowing behavior to ensure that scores are always based on the latest data. The efficient real-time scoring engine 117 incorporates a multi-level parallel computing mechanism, utilizing GPU acceleration and distributed computing technologies to handle large numbers of concurrent requests. Each credit score request is processed through a parallel pipeline, completing the entire process from data acquisition, feature extraction, model inference, to score output in an extremely short time. In addition, the efficient real-time scoring engine 117 also adopts a design based on a microservice architecture, and each scoring service can be independently deployed and expanded to ensure high availability and scalability of the system.

[0108] Formula: For a given user feature vector X={x1,x2,…,x n}, the scoring engine calculates the credit score through the deep learning model f(X;θ):

[0109] Score=f(X;θ)=σ(W2·ReLU(W1·X+b1)·+b2)

[0110] Where σ is an activation function (such as Sigmoid), and W1, W2, b1, and b2 are model parameters. The efficient real-time scoring engine 117 uses these formulas to efficiently calculate credit scores in a real-time environment. The scoring results are presented to financial institutions via a user-friendly user interface (UI) or API. The UI design includes charts and data tables, displaying multi-dimensional credit score information, such as the overall score, risk distribution, and behavioral influencing factors. The API supports multiple integration methods, allowing financial institutions to seamlessly integrate the scoring system into their existing risk control platforms for real-time decision support. When a user's credit score falls below a set threshold, the risk warning feedback unit 103 automatically triggers the risk warning mechanism. The risk warning feedback unit 103 relies on a combination of a rules engine and a deep learning model. The rules engine is based on simple threshold judgments, while the deep learning model can capture more complex risk patterns. For example, the system can monitor a user's multiple short-term borrowing and lending activities and large consumer expenditures. If these behaviors, combined, may indicate high credit risk, the system will immediately generate a warning. The risk warning feedback unit 103 not only uses static scoring thresholds but also incorporates an adaptive threshold adjustment mechanism. Reinforcement learning algorithms (such as Q-learning) enable continuous adjustment of risk thresholds based on historical data and user feedback to maximize financial institutions' profits and minimize losses. Specifically, the risk warning feedback unit 103 analyzes the results of each warning. If the warning is effective, it lowers the threshold for the corresponding condition to more sensitively capture risk signals. If the warning is ineffective, it raises the threshold to reduce false positives. Warning information is sent to relevant parties through multiple channels, including SMS, email, in-app notifications, and API push notifications. The notifications are comprehensive, including the risk score, behavioral indicators that triggered the warning, and recommended countermeasures, ensuring that financial institutions can respond quickly to risks. The real-time scoring module 105 includes a closed-loop feedback system to collect feedback from users and financial institutions on the scoring results. This feedback data is used through reinforcement learning and adaptive learning mechanisms to further optimize the model's predictive capabilities. For example, if a certain scoring result demonstrates high accuracy in real-world applications, the system will use feedback mechanisms to increase the weight of this feature; otherwise, its influence will be reduced. The model optimization process uses techniques such as Bayesian optimization and genetic algorithms to dynamically adjust model hyperparameters to ensure continuous performance improvement. The risk warning feedback unit 103 also provides risk prediction and trend analysis capabilities, predicting future credit risk trends by analyzing time series data of user behavior. This function utilizes LSTM or transformer models for time series prediction, generating future credit score curves to help financial institutions take proactive measures.

[0111] The real-time credit scoring system based on deep learning in this embodiment uses the real-time collection of multi-source heterogeneous data by the multi-source heterogeneous data real-time collection unit 106. The data preprocessing module 102 performs efficient data preprocessing, inputs the data into the deep learning model module 104 for complex deep learning model processing, and the real-time scoring module 105 performs real-time scoring and risk warning. This system achieves a comprehensive, efficient, and real-time assessment of user credit risk. Furthermore, a closed-loop feedback mechanism continuously optimizes the model to ensure the accuracy and stability of the scoring system. This system addresses the various limitations of traditional credit scoring methods and achieves more accurate and dynamic credit risk assessment.

[0112] The second embodiment of this application is:

[0113] Based on the first embodiment, please refer to Figure 4 ,in, Figure 4 It is a principle block diagram of the model optimization monitoring module 201 according to the second embodiment of the present invention.

[0114] The deep learning-based real-time credit scoring system of this embodiment further includes a model optimization monitoring module 201 , which includes an automated model optimization unit 202 and a model performance monitoring anomaly detection unit 203 .

[0115] For this specific embodiment, the model optimization monitoring module 201 is connected to the hybrid deep neural network architecture 115;

[0116] The model optimization monitoring module 201 is used to automatically adjust the hyperparameters and structure of the deep learning model, and to monitor the scoring accuracy and computational efficiency of the model in real time.

[0117] The automated model optimization unit 202 and the model performance monitoring anomaly detection unit 203 are respectively connected to the hybrid deep neural network architecture 115;

[0118] The automated model optimization unit 202 is used to integrate a reinforcement learning algorithm to automatically adjust the hyperparameters and structure of the deep learning model. The automated hyperparameter search algorithm based on Bayesian optimization is used in combination with an online adjustment strategy for reinforcement learning to improve model optimization efficiency by 60% and model performance by 15%.

[0119] The model performance monitoring anomaly detection unit 203 is used to monitor the scoring accuracy and computational efficiency of the model in real time. If abnormal fluctuations in the model score are detected, anomaly detection and analysis are automatically performed, and corresponding corrective measures are taken. Among them, an anomaly detection algorithm based on time series analysis is used, combined with an adaptive threshold adjustment mechanism, and the anomaly detection accuracy rate reaches 95%, with an average detection time of less than 1 second.

[0120] The automated model optimization unit 202 automatically adjusts the hyperparameters and structure of the deep learning model by integrating a reinforcement learning algorithm to continuously improve the model's performance. Regular optimization training is performed based on scoring accuracy and model error rate to ensure that the model remains efficient in the long term. To ensure model stability, the model performance monitoring anomaly detection unit 203 monitors the model's scoring accuracy and computational efficiency in real time. If abnormal fluctuations in the model score are detected, anomaly detection and analysis are automatically performed, and appropriate corrective measures are taken.

[0121] Using a real-time credit scoring system based on deep learning in this embodiment, the automated model optimization unit 202 is designed. By integrating a reinforcement learning algorithm, it regularly adjusts the hyperparameters and structure of the deep learning model to ensure that the model remains efficient during long-term operation. Furthermore, the model performance monitoring and anomaly detection unit 203 is equipped to monitor the model's scoring accuracy and computational efficiency in real time and automatically take corrective measures when abnormal fluctuations occur. These optimization and monitoring functions improve the stability and sustainability of the system, enabling financial institutions to rely on the system for long-term credit risk management.

[0122] The performance indicators of the real-time credit scoring system based on deep learning in actual application are shown in the following table:

[0123] index The present invention Traditional methods Improvement Response time 50ms 500ms 90％ Prediction accuracy 92.5％ 78.3％ 18.1％ AUC 0.956 0.812 17.7％ Feature generation efficiency 1200 records / second 850 records / second 41.2％ Model training speed 3.5 hours 10.2 hours 65.7％ Anomaly detection accuracy 95.2％ 82.5％ 15.4％

[0124] The above disclosure is merely one or more preferred embodiments of the present application and is not intended to limit the scope of the present application. A person skilled in the art will understand that all or part of the processes of the above embodiments and equivalent changes made in accordance with the claims of the present application are still within the scope of the present application.

Claims

1. A real-time credit scoring system based on deep learning, characterized by: It includes a data collection module, a data preprocessing module, a deep learning model module and a real-time scoring module, wherein the data preprocessing module is connected to the data collection module, the deep learning model module is connected to the data preprocessing module, and the real-time scoring module is connected to the deep learning model module; The data collection module is used to collect multi-dimensional data of users in real time using stream processing technology and store the data in a data warehouse through a distributed database; The data preprocessing module is used to clean and normalize the collected data and automatically generate higher-level features based on the collected data using automatic feature generation technology. After further dimensionality reduction, the features are used as input for the deep learning model; The deep learning model module is used to process the pre-processed data using deep learning technology and generate an accurate credit score; The real-time scoring module is used to convert the output of the deep learning model module into an actionable credit score and respond to changes in user behavior in real time.

2. The real-time credit scoring system based on deep learning according to claim 1, characterized in that: The data collection module includes a multi-source heterogeneous data real-time collection unit and a data flow management storage unit, and the data flow management storage unit is connected to the multi-source heterogeneous data real-time collection unit and the data preprocessing module respectively; The multi-source heterogeneous data real-time collection unit is used to establish connections with API interfaces of multiple data sources to collect multi-dimensional data of users in real time; The data stream management storage unit is used to introduce stream processing technology to process data streams, and at the same time store the collected data in a data warehouse through a distributed database.

3. The real-time credit scoring system based on deep learning according to claim 2, characterized in that: The multi-source heterogeneous data real-time collection unit includes a financial data source subunit, an e-commerce platform data subunit, a social network behavior data subunit, a geographic location information subunit, and a device usage data subunit, and the financial data source subunit, the e-commerce platform data subunit, the social network behavior data subunit, the geographic location information subunit, and the device usage data subunit are respectively connected to the data flow management and storage unit; The financial data source subunit is used to establish a real-time connection with financial institutions to collect users' bank transaction records, credit card usage, loan application records, and repayment history; The e-commerce platform data sub-unit is used to obtain users' consumption records on the e-commerce platform in real time through the API interface, including shopping habits, purchase frequency, payment methods and consumption amounts; The social network behavior data subunit is used to collect user behavior on social media platforms; The geographic location information subunit is used to obtain the user's location information through the geographic location service of the user equipment; The device usage data subunit is used to collect the usage of the user's device, including usage frequency, device type, operating system version and network connection mode.

4. The real-time credit scoring system based on deep learning according to claim 2, characterized in that: The data preprocessing module includes a data cleaning and normalization unit and an advanced feature engineering unit, the data cleaning and normalization unit is connected to the data flow management and storage unit, and the advanced feature engineering unit is connected to the data cleaning and normalization unit; The data cleaning and normalization unit is used to clean the collected raw data, delete invalid data, correct erroneous records, and format and normalize the cleaned data; The advanced feature engineering unit is used to adopt automatic feature generation technology to automatically generate higher-level features based on the normalized data, and further perform dimensionality reduction on the features before using them as input to the deep learning model.

5. The real-time credit scoring system based on deep learning according to claim 4, characterized in that: The deep learning model module includes a hybrid deep neural network architecture and an adaptive learning dynamic update unit, the hybrid deep neural network architecture is connected to the advanced feature engineering unit, and the adaptive learning dynamic update unit is connected to the hybrid deep neural network architecture; The hybrid deep neural network architecture is used to integrate deep learning technologies such as convolutional neural networks, long short-term memory networks, graph neural networks, and transformer models based on self-attention mechanisms to form a model system. In particular, a hierarchical attention mechanism is used to integrate different networks in the following ways: Convolutional neural networks are used to extract local pattern features of user behavior data; Long short-term memory networks are used to capture the time series characteristics of user behavior; Graph neural networks are used to model user social relationship networks; Transformer model for global feature interaction and attention allocation; Through the multimodal feature fusion layer, the outputs of different networks are adaptively weighted and fused; The adaptive learning dynamic update unit is used to perform online learning, incremental training and reinforcement learning on the model to ensure that the model maintains high accuracy and stability in a constantly changing data environment.

6. The real-time credit scoring system based on deep learning according to claim 5, characterized in that: The real-time scoring module includes an efficient real-time scoring engine and a risk warning feedback unit, wherein the efficient real-time scoring engine is connected to the hybrid deep neural network architecture, and the risk warning feedback unit is connected to the efficient real-time scoring engine; The efficient real-time scoring engine is used to monitor changes in user behavior data in real time, process the changing data stream, and generate credit scores in real time. It uses a stream processing architecture based on in-memory computing, combined with a lightweight neural network inference engine, to achieve millisecond-level response, with an average response time of less than 50ms and 99.9% of requests completed within 100ms. The risk warning feedback unit automatically triggers a risk warning mechanism when a user's credit score falls below a set threshold. It uses a dynamic threshold adjustment algorithm to automatically adjust the risk threshold based on real-time market data and user behavior patterns, increasing warning accuracy by 30% and reducing false alarm rates by 45%.

7. The real-time credit scoring system based on deep learning according to claim 5, characterized in that: The real-time credit scoring system based on deep learning also includes a model optimization monitoring module, which is connected to the hybrid deep neural network architecture; The model optimization monitoring module is used to automatically adjust the hyperparameters and structure of the deep learning model, and to monitor the scoring accuracy and computational efficiency of the model in real time.

8. The real-time credit scoring system based on deep learning according to claim 7, characterized in that: The model optimization and monitoring module includes an automated model optimization unit and a model performance monitoring anomaly detection unit, wherein the automated model optimization unit and the model performance monitoring anomaly detection unit are respectively connected to the hybrid deep neural network architecture; The automated model optimization unit is used to integrate reinforcement learning algorithms and automatically adjust the hyperparameters and structure of deep learning models. The automated hyperparameter search algorithm based on Bayesian optimization, combined with the online adjustment strategy of reinforcement learning, improves model optimization efficiency by 60% and model performance by 15%. The model performance monitoring anomaly detection unit is used to monitor the scoring accuracy and computational efficiency of the model in real time. If abnormal fluctuations in the model score are detected, anomaly detection and analysis are automatically performed, and corresponding corrective measures are taken. Among them, an anomaly detection algorithm based on time series analysis is used, combined with an adaptive threshold adjustment mechanism, and the anomaly detection accuracy rate reaches 95%, with an average detection time of less than 1 second.

Images