A training method, device, medium and equipment for feature mining model

Abstract

The present invention discloses a method for training a feature mining model, including: determining a first business label and a second business label; obtaining a first training set corresponding to the first business label and a second training set corresponding to the second business label in a target business scenario, wherein the first training set is used to train a first machine learning model; using the trained first machine learning model to perform feature extraction on the second training set to obtain a first feature set; performing feature extraction on the second training set by manual extraction to obtain a second feature set; merging the first feature set and the second feature set to obtain a merged feature set; and training a second machine learning model using the merged feature set. The present invention performs feature mining through a machine learning model, so that the feature mining process is fully automated, and there is no need to manually define features, thus saving manpower and material resources. At the same time, by performing feature mining through a machine learning model, high-dimensional features can be quickly generated, and effective information in the data can be fully mined.

Description

Technical Field

[0001] The present invention relates to the field of feature mining technology, and in particular to a training method, device, medium and equipment for a feature mining model. Background Art

[0002] In the application scenarios of machine learning models, the problem of processing time series features is often encountered. For example, in financial risk control, the original data is the user's credit card consumption flow or the company's invoice flow information. Traditional feature processing methods rely entirely on modeling experts to manually define features (such as "the total amount of consumption flow in the past three months", "the average monthly consumption times in the past year", etc.) to train machine learning models.

[0003] The above-mentioned method of manually defining features relies heavily on the modeling experts' understanding of business features. When encountering an unfamiliar business domain, which features are more effective can only be determined by manually constructing a large number of features and then verifying and screening them, which consumes a lot of manpower and material resources.

[0004] On the other hand, manually defined features have their limitations and cannot fully utilize the value of the data. They are likely to miss some important features, which in turn affects the final modeling effect. Summary of the Invention

[0005] In view of the above-mentioned shortcomings of the prior art, an object of the present invention is to provide a method, apparatus, medium and equipment for training a feature mining model to solve at least one defect in the prior art.

[0006] To achieve the above-mentioned and other related purposes, the present invention provides a method for training a feature mining model, comprising:

[0007] Determine a first service tag and a second service tag;

[0008] Obtaining a first training set corresponding to the first business label and a second training set corresponding to the second business label in a target business scenario, where the first training set is used to train a first machine learning model;

[0009] Performing feature extraction on the second training set using the trained first machine learning model to obtain a first feature set;

[0010] Performing feature extraction on the second training set by manual extraction to obtain a second feature set;

[0011] Merging the first feature set and the second feature set to obtain a merged feature set;

[0012] A second machine learning model is trained using the combined feature set.

[0013] Optionally, the first machine learning model includes: a time series model.

[0014] Optionally, the method further includes screening the merged feature set to obtain a target feature set; the target feature set is used to train a second machine learning model.

[0015] Optionally, the first service tag is different from the second service tag.

[0016] To achieve the above-mentioned and other related purposes, the present invention provides an automatic feature mining device based on a time series model, comprising:

[0017] A label determination module, configured to determine a first service label and a second service label;

[0018] A training set acquisition module, configured to acquire a first training set corresponding to the first service label and a second training set corresponding to the second service label in a target service scenario, wherein the first training set is used to train a first machine learning model;

[0019] A first feature extraction module is used to extract features from the second training set using the trained first machine learning model to obtain a first feature set;

[0020] A second feature extraction module is used to extract features from the second training set by manual extraction to obtain a second feature set;

[0021] a feature merging module, configured to merge the first feature set and the second feature set to obtain a merged feature set;

[0022] A model training module is used to train a second machine learning model using the combined feature set.

[0023] Optionally, the first machine learning model includes: a time series model.

[0024] Optionally, the device further includes: a feature screening module, used to screen the merged feature set to obtain a target feature set; the target feature set is used to train a second machine learning model.

[0025] Optionally, the first service tag is different from the second service tag.

[0026] To achieve the above-mentioned and other related purposes, the present invention provides a training device for a feature mining model, comprising:

[0027] one or more processors; and

[0028] One or more machine-readable media having instructions stored thereon, which, when executed by the one or more processors, cause the apparatus to perform one or more of the described methods.

[0029] To achieve the above and other related objectives, the present invention provides one or more machine-readable media having instructions stored thereon, which, when executed by one or more processors, cause a device to perform one or more of the methods described above.

[0030] As described above, the method, apparatus, medium, and device for training a feature mining model of the present invention have the following beneficial effects:

[0031] A method for training a feature mining model of the present invention includes: determining a first business label and a second business label; obtaining a first training set corresponding to the first business label and a second training set corresponding to the second business label in a target business scenario, wherein the first training set is used to train a first machine learning model; performing feature extraction on the second training set using the trained first machine learning model to obtain a first feature set; performing feature extraction on the second training set by manual extraction to obtain a second feature set; merging the first feature set and the second feature set to obtain a merged feature set; and training a second machine learning model using the merged feature set. The present invention performs feature mining through a machine learning model, so that the feature mining process is fully automated, and there is no need to manually define features, thus saving manpower and material resources. At the same time, by performing feature mining through a machine learning model, high-dimensional features can be quickly generated to fully mine the effective information in the data.

[0032] Illustrations

[0033] Figure 1 This is a flow chart of a method for training a feature mining model according to an embodiment of the present invention;

[0034] Figure 2 A schematic diagram of a timing model according to an embodiment of the present invention;

[0035] Figure 3 This is a schematic diagram of the hardware structure of a training device for a feature mining model according to an embodiment of the present invention;

[0036] Figure 4 A schematic diagram of the hardware structure of a terminal device in one embodiment of the present invention;

[0037] Figure 5 FIG. 1 is a schematic diagram of the hardware structure of a terminal device in one embodiment of the present invention. DETAILED DESCRIPTION

[0038] The following describes the embodiments of the present invention through specific examples. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments. The details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other unless they conflict.

[0039] It should be noted that the illustrations provided in the following embodiments are merely schematic illustrations of the basic concept of the present invention. Therefore, the illustrations only show components related to the present invention and are not drawn according to the number, shape, and size of components in actual implementation. In actual implementation, the type, quantity, and proportion of each component may be changed arbitrarily, and the component layout may also be more complex.

[0040] like Figure 1 As shown, the embodiment of the present application provides a method for training a feature mining model, comprising the following steps:

[0041] S100, determining a first service tag and a second service tag;

[0042] S101: Obtain a first training set corresponding to the first service label and a second training set corresponding to the second service label in a target service scenario, where the first training set is used to train a first machine learning model;

[0043] S102, using the trained first machine learning model to perform feature extraction on the second training set to obtain a first feature set;

[0044] S103, performing feature extraction on the second training set by manual extraction to obtain a second feature set;

[0045] S104, merging the first feature set and the second feature set to obtain a merged feature set;

[0046] S105: Use the merged feature set to train a second machine learning model.

[0047] This invention uses machine learning models to perform feature mining, making the feature mining process fully automated, eliminating the need for manual feature definition and saving manpower and resources. It also allows for rapid generation of high-dimensional features, fully mining the useful information in the data.

[0048] The following is a detailed description of each step

[0049] In step S101, the target business scenarios may include financial transaction business scenarios (such as securities trading systems and currency payment systems), Internet of Things business scenarios (such as wearable devices and smart home systems), environmental monitoring business scenarios (such as humidity monitoring systems and air particulate matter monitoring systems), medical intelligence business scenarios (such as key health indicator tracking systems and health status monitoring systems), and event application scenarios (such as user information management systems and membership management systems). In this embodiment, the target business scenario is a financial risk control scenario.

[0050] In step S100, the business tag refers to the purpose of the model. For example, in a financial risk control scenario, the business tag might be "overdue," meaning the model is trained to predict whether a customer will default. In this embodiment, the first business tag represents the purpose of training the first machine learning model, and the second business tag represents the purpose of training the second machine learning model.

[0051] For example, the first business tag is "Overdue?", which predicts whether a customer will default based on their loan overdue assessment information. Depending on the source, this information may include directly accessible information such as age, gender, education level, industry, and loan product. It may also include information that requires data processing, such as the proportion of loan age to loan term, the amount of funds inflow within the current period, and the difference between the current inflow and the repayment amount. The amount of information included in the customer's loan overdue assessment information is determined based on actual needs and is not limited in this embodiment of the present invention.

[0052] In one embodiment, the first service label is different from the second service label. That is, the sample labels used to train the first machine learning model can be different from those used to train the second machine learning model. This allows for training an unsupervised machine learning model (without sample labels) while also allowing the label information used to train the first machine learning model to be transferred to the final second machine learning model.

[0053] The first training set is a set of samples used to train a first machine learning model, which is used to extract features from data. The first training set includes historical data from a target business scenario. The first training set includes at least one sample data item, each of which consists of a historical data item, and each sample data item includes at least one feature. This embodiment does not limit this. In actual applications, the number of samples and features in the first training set can be determined based on the processing capabilities of the corresponding system and the actual business scenario.

[0054] The second training set is a sample used to train the second machine learning model, wherein the second machine learning model is used to perform feature mining on the data. The second training set includes historical data under the target business scenario. Among them, the second training set includes at least one sample data, each sample data is composed of a historical data, and each sample includes at least one feature. This embodiment is not limited to this. In actual applications, the number of samples and the number of features in the second training set can be determined in combination with the processing capabilities of the corresponding system and the actual business scenario. It should be noted that before training the feature mining model, the data of the second training set needs to be processed accordingly, that is, the features of the second training set are extracted, and then the extracted features are used to train the feature mining model.

[0055] In one embodiment, the first machine learning model is a time series model. Since the first machine learning model is a time series model, the sample data of the first training set is time series data, which refers to data collected at different times in the target business scenario and is used to describe how phenomena occurring in the target business scenario change over time.

[0056] The schematic diagram of the timing model is as follows Figure 2 As shown, X1-Xn are original statistical information at each moment, and the time granularity (for example, day / week / month as a time unit) can be set according to actual conditions.

[0057] Z1~Zk are the outputs corresponding to each category label. For example, in the financial risk control model, Z1~Zk are the confidence levels of each category.

[0058] A represents a time series model, such as RNN (Recurrent Neural Network, RNN), LSTM (Long Short-Term Memory Network, LSTM, Long Short-Term Memory), etc.

[0059] H0~Hn are the model outputs at each moment. They are high-dimensional vectors. Ht can remember the data information at moment t and before. The last Hn remembers all the information of the pipeline, so Hn can be used as the final output feature.

[0060] In step S102, the trained first machine learning model is used to perform feature extraction on the second training set to obtain a first feature set.

[0061] Specifically, the second training set is input into the first machine learning model to obtain a first feature set. The first feature set is presented in the form of a two-dimensional table. In the two-dimensional table, each column represents a feature, and each row represents a sample data.

[0062] In step S103, feature extraction is performed on the second training set by manual extraction to obtain a second feature set;

[0063] Specifically, artificial features are calculated from the second training set using the computational logic of artificially defined features to form the second feature set. The first feature set is presented in the form of a two-dimensional table, where each column represents a feature and each row represents a sample data point.

[0064] In step S104, the first feature set and the second feature set are merged to obtain a merged feature set.

[0065] Since the first feature set is presented in the form of a two-dimensional table and the second feature set is presented in the form of a two-dimensional table, when the first feature set and the second feature set are merged, the first two-dimensional table of the first feature set and the second two-dimensional table of the second feature set are merged.

[0066] When merging the first two-dimensional table with the second two-dimensional table, if the first two-dimensional table and the second two-dimensional table do not have the same features, that is, if the first two-dimensional table and the second two-dimensional table do not have the same columns, the first two-dimensional table and the second two-dimensional table columns are directly merged. The first two-dimensional table and the second two-dimensional table are merged, that is, the first feature set and the second feature set are added together in the sample dimension to form a feature table with a larger dimension. In this feature table, each row represents a sample data, and each column represents a one-dimensional feature.

[0067] For example, if there are 20-dimensional features in the first two-dimensional table and 100-dimensional features in the second two-dimensional table, after merging the first two-dimensional table with the second two-dimensional table, the number of sample data items does not change, but the dimension of the features is 120.

[0068] When merging the first two-dimensional table with the second two-dimensional table, if the first two-dimensional table and the second two-dimensional table have the same features, that is, the same columns, the columns with the same names in the first two-dimensional table and the second two-dimensional table are merged to obtain a merged column.

[0069] Specifically, a computer intelligently identifies and compares the names of the columns in the first and second two-dimensional tables, selects columns with the same name, and merges the columns with the same name in the first and second two-dimensional tables to obtain a merged column. Then, the columns in the first and second two-dimensional tables that have different names from the columns to be merged are concatenated with the merged column to obtain a merged feature set.

[0070] Specifically, after the merged columns are determined, the remaining columns in the first and second two-dimensional tables can be concatenated with the merged columns. This results in a new merged two-dimensional table, or merged feature set. In this feature table, each row represents a sample data point, and each column represents a one-dimensional feature.

[0071] In step S105, the combined feature set is used to train a second machine learning model. The second machine learning model can be a classification model, such as LR (Logistic Regression), XGBoost (Extreme Gradient Boosting), GBDT (Decision Tree), DNN (Deep Neural Network), clustering model (such as Kmeans), regression model (such as linear regression, DNN regression), etc.

[0072] In one embodiment, the method further includes screening the merged feature set to obtain a target feature set; the target feature set is used to train the second machine learning model. In this embodiment, by screening the merged feature set, a feature set with higher discrimination, i.e., the target feature set, is obtained. Here, higher discrimination means that certain features are utilized during the model training process so that the indicators of the final trained model can better meet the set requirements. For example, in the field of financial risk control, the set requirement is that the model can identify whether the customer is overdue, then income, whether there is a history of overdue payments, etc. are features with higher discrimination, i.e., target features; while gender is a feature with lower discrimination, i.e., non-target features.

[0073] This paper addresses the challenges of traditional feature definition methods by providing a method for automated time series feature mining based on time series models (e.g., LSTM, RNN, etc.). This method utilizes sample labels to guide feature mining, generating tens of thousands of features. These features can then be combined with other manually defined features to train machine learning models, thereby improving the overall model's discriminability.

[0074] like Figure 3 As shown, the embodiment of the present application provides an automatic feature mining device based on a time series model, comprising:

[0075] A label determination module 300 is used to determine a first service label and a second service label;

[0076] A training set acquisition module 301 is configured to acquire a first training set corresponding to the first service label and a second training set corresponding to the second service label in a target service scenario, wherein the first training set is used to train a first machine learning model;

[0077] A first feature extraction module 302 is configured to extract features from the second training set using the trained first machine learning model to obtain a first feature set;

[0078] A second feature extraction module 303 is configured to extract features from the second training set manually to obtain a second feature set;

[0079] A feature merging module 304 is configured to merge the first feature set and the second feature set to obtain a merged feature set;

[0080] The model training module 305 is used to train a second machine learning model using the combined feature set.

[0081] This invention uses machine learning models to perform feature mining, making the feature mining process fully automated, eliminating the need for manual feature definition and saving manpower and resources. It also allows for rapid generation of high-dimensional features, fully mining the useful information in the data.

[0082] The target business scenarios may include financial transaction business scenarios (such as securities trading systems and currency payment systems), Internet of Things business scenarios (such as wearable devices and smart home systems), environmental monitoring business scenarios (such as humidity monitoring systems and air particulate matter monitoring systems), medical intelligence business scenarios (such as key health indicator tracking systems and health status monitoring systems), and event application application scenarios (such as user information management systems and membership management systems). In this embodiment, the target business scenario is a financial risk control scenario.

[0083] A business tag refers to the purpose of the model. For example, in a financial risk control scenario, a business tag might be "overdue," meaning the model is trained to predict whether a customer will default. In this embodiment, the first business tag represents the purpose of training the first machine learning model, and the second business tag represents the purpose of training the second machine learning model.

[0084] For example, the first business tag is "Overdue?", which predicts whether a customer will default based on their loan overdue assessment information. Depending on the source, this information may include directly accessible information such as age, gender, education level, industry, and loan product. It may also include information that requires data processing, such as the proportion of loan age to loan term, the amount of funds inflow within the current period, and the difference between the current inflow and the repayment amount. The amount of information included in the customer's loan overdue assessment information is determined based on actual needs and is not limited in this embodiment of the present invention.

[0085] In one embodiment, the first service label is different from the second service label. That is, the sample labels used to train the first machine learning model can be different from those used to train the second machine learning model. This allows for training an unsupervised machine learning model (without sample labels) while also allowing the label information used to train the first machine learning model to be transferred to the final second machine learning model.

[0086] The first training set is a set of samples used to train a first machine learning model, which is used to extract features from data. The first training set includes historical data from a target business scenario. The first training set includes at least one sample data item, each of which consists of a historical data item, and each sample data item includes at least one feature. This embodiment does not limit this. In actual applications, the number of samples and features in the first training set can be determined based on the processing capabilities of the corresponding system and the actual business scenario.

[0087] The second training set is a sample used to train the second machine learning model, wherein the second machine learning model is used to perform feature mining on the data. The second training set includes historical data under the target business scenario. Among them, the second training set includes at least one sample data, each sample data is composed of a historical data, and each sample includes at least one feature. This embodiment is not limited to this. In actual applications, the number of samples and the number of features in the second training set can be determined in combination with the processing capabilities of the corresponding system and the actual business scenario. It should be noted that before training the feature mining model, the data of the second training set needs to be processed accordingly, that is, the features of the second training set are extracted, and then the extracted features are used to train the feature mining model.

[0088] In one embodiment, the first machine learning model is a time series model. Since the first machine learning model is a time series model, the sample data of the first training set is time series data, which refers to data collected at different times in the target business scenario and is used to describe how phenomena occurring in the target business scenario change over time.

[0089] The schematic diagram of the timing model is as follows Figure 2 As shown, X1-Xn are original statistical information at each moment, and the time granularity (for example, day / week / month as a time unit) can be set according to actual conditions.

[0090] Z1~Zk are the outputs corresponding to each category label. For example, in the financial risk control model, Z1~Zk are the confidence levels of each category.

[0091] A represents a time series model, such as RNN (Recurrent Neural Network, RNN), LSTM (Long Short-Term Memory Network, LSTM, Long Short-Term Memory), etc.

[0092] H0~Hn are the model outputs at each moment. They are high-dimensional vectors. Ht can remember the data information at moment t and before. The last Hn remembers all the information of the pipeline, so Hn can be used as the final output feature.

[0093] In this embodiment, the first feature extraction module 302 uses the trained first machine learning model to perform feature extraction on the second training set to obtain a first feature set.

[0094] Specifically, the second training set is input into the first machine learning model to obtain a first feature set. The first feature set is presented in the form of a two-dimensional table. In the two-dimensional table, each column represents a feature, and each row represents a sample data.

[0095] In this embodiment, the second feature extraction module 303 extracts features from the second training set by manual extraction to obtain a second feature set;

[0096] Specifically, artificial features are calculated from the second training set using the computational logic of artificially defined features to form the second feature set. The first feature set is presented in the form of a two-dimensional table, where each column represents a feature and each row represents a sample data point.

[0097] In this embodiment, the feature merging module 304 merges the first feature set and the second feature set to obtain a merged feature set.

[0098] Since the first feature set is presented in the form of a two-dimensional table and the second feature set is presented in the form of a two-dimensional table, when the first feature set and the second feature set are merged, the first two-dimensional table of the first feature set and the second two-dimensional table of the second feature set are merged.

[0099] When merging the first two-dimensional table with the second two-dimensional table, if the first two-dimensional table and the second two-dimensional table do not have the same features, that is, if the first two-dimensional table and the second two-dimensional table do not have the same columns, the first two-dimensional table and the second two-dimensional table columns are directly merged. The first two-dimensional table and the second two-dimensional table are merged, that is, the first feature set and the second feature set are added together in the sample dimension to form a feature table with a larger dimension. In this feature table, each row represents a sample data, and each column represents a one-dimensional feature.

[0100] For example, if there are 20-dimensional features in the first two-dimensional table and 100-dimensional features in the second two-dimensional table, after merging the first two-dimensional table with the second two-dimensional table, the number of sample data items does not change, but the dimension of the features is 120.

[0101] When merging the first two-dimensional table with the second two-dimensional table, if the first two-dimensional table and the second two-dimensional table have the same features, that is, the same columns, the columns with the same names in the first two-dimensional table and the second two-dimensional table are merged to obtain a merged column.

[0102] Specifically, a computer intelligently identifies and compares the names of the columns in the first and second two-dimensional tables, selects columns with the same name, and merges the columns with the same name in the first and second two-dimensional tables to obtain a merged column. Then, the columns in the first and second two-dimensional tables that have different names from the columns to be merged are concatenated with the merged column to obtain a merged feature set.

[0103] Specifically, after the merged columns are determined, the remaining columns in the first and second two-dimensional tables can be concatenated with the merged columns. This results in a new merged two-dimensional table, or merged feature set. In this feature table, each row represents a sample data point, and each column represents a one-dimensional feature.

[0104] In this embodiment, the model training module 305 uses the merged feature set to train a second machine learning model. The second machine learning model can be a classification model, such as LR (Logistic Regression, LR), XGBoost (Extreme Gradient Boosting, XGBoost), GBDT (GradientBoosting Decision Tree, GBDT), DNN (Deep Neural Network, DNN), clustering model (such as Kmeans model), regression model (such as linear regression, DNN regression model, etc.).

[0105] In one embodiment, the device further includes: a feature screening module for screening the merged feature set to obtain a target feature set; the target feature set is used to train the second machine learning model. In this embodiment, a feature set with higher discrimination, i.e., a target feature set, is obtained by screening the merged feature set. Here, higher discrimination means that certain features are utilized during the model training process so that the indicators of the final trained model can better meet the set requirements. For example, in the field of financial risk control, the set requirement is that the model can identify whether the customer is overdue, then income, whether there is a history of overdue payments, etc. are features with higher discrimination, i.e., target features; while gender is a feature with lower discrimination, i.e., a non-target feature.

[0106] This paper addresses the challenges of traditional feature definition methods by providing a device for automated time series feature mining based on time series models (e.g., LSTM, RNN, etc.). This device uses sample labels to guide feature mining, generating tens of thousands of features. These features can then be combined with other manually defined features to train machine learning models, thereby improving the overall model's discriminability.

[0107] Since the above-mentioned device embodiment corresponds to the method embodiment, the functional implementation of each module in the device embodiment can refer to the implementation method of the method embodiment, and will not be repeated here.

[0108] The present application also provides a device, which may include: one or more processors; and one or more machine-readable media having instructions stored thereon, which, when executed by the one or more processors, enable the device to execute Figure 1 In practical applications, the device can be used as a terminal device or as a server. Examples of terminal devices may include: smartphones, tablet computers, e-book readers, MP3 (Moving Picture Experts Group Audio Layer III) players, MP4 (Moving Picture Experts Group Audio Layer IV) players, laptop computers, car computers, desktop computers, set-top boxes, smart TVs, wearable devices, etc. The embodiments of this application do not limit the specific devices.

[0109] The present application also provides a non-volatile readable storage medium, which stores one or more modules (programs). When the one or more modules are applied to a device, the device can execute the embodiment of the present application. Figure 1 Instructions for the steps in a method.

[0110] Figure 4 A schematic diagram of the hardware structure of a terminal device provided in one embodiment of the present application. As shown in the figure, the terminal device may include: an input device 1100, a first processor 1101, an output device 1102, a first memory 1103, and at least one communication bus 1104. Communication bus 1104 is used to achieve communication connections between components. First memory 1103 may include high-speed RAM memory, and may also include non-volatile storage NVM, such as at least one disk storage device. Various programs can be stored in first memory 1103 to perform various processing functions and implement the method steps of this embodiment.

[0111] Optionally, the first processor 1101 may be implemented as, for example, a central processing unit (CPU), an application-specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD), a field programmable gate array (FPGA), a controller, a microcontroller, a microprocessor or other electronic components, and the first processor 1101 is coupled to the input device 1100 and the output device 1102 via a wired or wireless connection.

[0112] Optionally, the input device 1100 may include multiple input devices, such as at least one of a user interface for a user, a device interface for a device, a software programmable interface, a camera, and a sensor. Optionally, the device interface for a device may be a wired interface for data transmission between devices, or a hardware plug-in interface for data transmission between devices (such as a USB interface, a serial port, etc.); optionally, the user interface for a user may be, for example, a user-oriented control button, a voice input device for receiving voice input, and a touch sensing device for receiving user touch input (such as a touch screen or touchpad with touch sensing function); optionally, the software programmable interface may be, for example, an entry for a user to edit or modify a program, such as an input pin interface or input interface of a chip; the output device 1102 may include output devices such as a display and a speaker.

[0113] In this embodiment, the processor of the terminal device includes a processor for executing the functions of each module in each device. The specific functions and technical effects can be referred to the above embodiments and will not be repeated here.

[0114] Figure 5 A schematic diagram of the hardware structure of a terminal device provided for one embodiment of the present application. Figure 5This is a specific embodiment of the implementation of Figure 4. As shown in the figure, the terminal device of this embodiment may include a second processor 1201 and a second memory 1202.

[0115] The second processor 1201 executes the computer program code stored in the second memory 1202 to implement the above embodiment. Figure 1 The method.

[0116] The second memory 1202 is configured to store various types of data to support operations on the terminal device. Examples of such data include instructions for any application or method operating on the terminal device, such as messages, images, videos, etc. The second memory 1202 may include random access memory (RAM) and may also include non-volatile memory (non-volatile memory), such as at least one disk storage device.

[0117] Optionally, the second processor 1201 is provided in the processing component 1200. The terminal device may further include: a communication component 1203, a power component 1204, a multimedia component 1205, a voice component 1206, an input / output interface 1207, and / or a sensor component 1208. The specific components included in the terminal device are set according to actual needs and are not limited in this embodiment.

[0118] The processing component 1200 generally controls the overall operation of the terminal device. The processing component 1200 may include one or more second processors 1201 to execute instructions to complete all or part of the steps in the above-described data processing method. In addition, the processing component 1200 may include one or more modules to facilitate interaction between the processing component 1200 and other components. For example, the processing component 1200 may include a multimedia module to facilitate interaction between the multimedia component 1205 and the processing component 1200.

[0119] The power supply component 1204 provides power to various components of the terminal device. The power supply component 1204 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the terminal device.

[0120] The multimedia component 1205 includes a display screen that provides an output interface between the terminal device and the user. In some embodiments, the display screen may include a liquid crystal display (LCD) and a touch panel (TP). If the display screen includes a touch panel, the display screen may be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor can not only sense the boundaries of a touch or slide action, but also detect the duration and pressure associated with the touch or slide operation.

[0121] The voice component 1206 is configured to output and / or input voice signals. For example, the voice component 1206 includes a microphone (MIC), which is configured to receive external voice signals when the terminal device is in an operating mode, such as a voice recognition mode. The received voice signals can be further stored in the second memory 1202 or transmitted via the communication component 1203. In some embodiments, the voice component 1206 also includes a speaker for outputting voice signals.

[0122] The input / output interface 1207 provides an interface between the processing component 1200 and peripheral interface modules, which may be click wheels, buttons, etc. These buttons may include but are not limited to: volume buttons, start buttons, and lock buttons.

[0123] Sensor component 1208 includes one or more sensors for providing various status assessments for the terminal device. For example, sensor component 1208 can detect the open / closed state of the terminal device, the relative positioning of components, and the presence or absence of user contact with the terminal device. Sensor component 1208 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact, including detecting the distance between the user and the terminal device. In some embodiments, sensor component 1208 may also include a camera, etc.

[0124] The communication component 1203 is configured to facilitate wired or wireless communication between the terminal device and other devices. The terminal device can access a wireless network based on a communication standard, such as WiFi, 2G or 3G, or a combination thereof. In one embodiment, the terminal device may include a SIM card slot for inserting a SIM card, allowing the terminal device to log into a GPRS network and establish communication with a server via the Internet.

[0125] From the above, we can see that Figure 5 The communication component 1203, the voice component 1206, the input / output interface 1207, and the sensor component 1208 involved in the embodiment can all be used as Figure 4 Implementation of the input device in the embodiment.

[0126] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the present invention. Anyone skilled in the art may modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by one of ordinary skill in the art without departing from the spirit and technical principles disclosed herein are intended to be covered by the claims of the present invention.

Claims

1. A method for training a feature mining model, characterized in that: include: Determine a first service tag and a second service tag; Obtaining a first training set corresponding to the first business label and a second training set corresponding to the second business label in a target business scenario, where the first training set is used to train a first machine learning model; Performing feature extraction on the second training set using the trained first machine learning model to obtain a first feature set; Performing feature extraction on the second training set by manual extraction to obtain a second feature set; Merging the first feature set and the second feature set to obtain a merged feature set; training a second machine learning model using the combined feature set; The first service tag is different from the second service tag.

2. The method for training a feature mining model according to claim 1, wherein: The first machine learning model includes: a time series model.

3. The method for training a feature mining model according to claim 1, wherein: The method also includes screening the merged feature set to obtain a target feature set; the target feature set is used to train a second machine learning model.

4. A training device for a feature mining model, characterized in that: include: A label determination module, configured to determine a first service label and a second service label; A training set acquisition module, configured to acquire a first training set corresponding to the first service label and a second training set corresponding to the second service label in a target service scenario, wherein the first training set is used to train a first machine learning model; A first feature extraction module is used to extract features from the second training set using the trained first machine learning model to obtain a first feature set; A second feature extraction module is used to extract features from the second training set by manual extraction to obtain a second feature set; a feature merging module, configured to merge the first feature set and the second feature set to obtain a merged feature set; a model training module, configured to train a second machine learning model using the combined feature set; The first service tag is different from the second service tag.

5. The feature mining model training device according to claim 4, characterized in that: The first machine learning model includes: a time series model.

6. The feature mining model training device according to claim 4, characterized in that: The device also includes: a feature screening module, used to screen the merged feature set to obtain a target feature set; the target feature set is used to train a second machine learning model.

7. A training device for a feature mining model, characterized in that: include: one or more processors; and One or more machine-readable media having instructions stored thereon, which, when executed by the one or more processors, cause the apparatus to perform the method of one or more of claims 1-3.

8. One or more machine-readable media, characterized in that Instructions are stored thereon, which, when executed by one or more processors, cause the apparatus to perform the method as claimed in one or more of claims 1-3.

Images