Feature selection method and device, model training method and device, equipment and medium

Abstract

The present disclosure provides a computer-implemented feature selection method and device, a model training method and device, equipment and a medium, relating to the field of artificial intelligence, and in particular to the field of big data technology. The implementation scheme is as follows: determining a weight coefficient and a correlation coefficient of each feature in a plurality of features of a to-be-processed object, wherein the weight coefficient of each feature is determined according to the association degree of the feature and a to-be-processed task, and the correlation coefficient of each feature is determined according to the correlation between the feature and other features; and determining a selected feature for performing the to-be-processed task from the plurality of features based on the weight coefficient and the correlation coefficient of each feature in the plurality of features.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the field of artificial intelligence, in particular to the field of big data, and more particularly to a feature selection method, a feature selection model training method, an apparatus, an electronic device, a computer readable storage medium and a computer program product. BACKGROUND

[0002] Artificial intelligence is a discipline that studies enabling computers to simulate some thinking processes and intelligent behaviors (such as learning, reasoning, thinking, planning, etc.) of human beings, which includes both hardware technology and software technology. Artificial intelligence hardware technology generally includes technologies such as sensors, special artificial intelligence chips, cloud computing, distributed storage, big data processing, etc. Artificial intelligence software technology mainly includes computer vision technology, speech recognition technology, natural language processing technology, and machine learning / deep learning, big data processing technology, knowledge graph technology, etc.

[0003] The methods described in this section can not have been previously conceived or made. Unless otherwise indicated herein, the mere fact that any method described in this section is set forth in this section should not be construed as an indication that such method was known in the art prior to its disclosure. Similarly, the mere fact that a problem is described in this section should not be construed as an indication that the problem was recognized in the art prior to its disclosure. SUMMARY

[0004] The present disclosure provides a feature selection method, a feature selection model training method, an apparatus, an electronic device, a computer readable storage medium and a computer program product.

[0005] According to an aspect of the present disclosure, there is provided a computer-implemented feature selection method, comprising: determining a weight coefficient and a correlation coefficient of each of a plurality of features of a to-be-processed object, wherein the weight coefficient of each feature is determined according to an association degree of the feature with a to-be-processed task, and the correlation coefficient of each feature is determined according to a correlation between the feature and other features; and determining, based on the weight coefficient and the correlation coefficient of each of the plurality of features, selected features from the plurality of features for performing the to-be-processed task.

[0006] According to an aspect of the present disclosure, a training method of a feature selection model is provided, wherein the feature selection model comprises a feature extraction module, a weight module and a correlation module, the feature selection model is connected with a task processing model, and the training method comprises: inputting a plurality of features of a sample object into the feature extraction module to obtain representation information of each of the plurality of features; obtaining a weight coefficient of each of the plurality of features based on the representation information of each of the plurality of features by using the weight module; obtaining a correlation coefficient of each of the plurality of features based on the representation information of each of the plurality of features by using the correlation module; determining a prediction selected feature in the plurality of features based on the weight coefficient and the correlation coefficient of each of the plurality of features; and adjusting a parameter of the feature selection model based on the weight coefficient and the correlation coefficient of the prediction selected feature and a feedback result of the prediction selected feature, wherein the feedback result of the prediction selected feature is determined according to a prediction result of the task processing model on the prediction selected feature.

[0007] According to an aspect of the present disclosure, a feature selection device is provided, comprising: a first determination unit configured to determine a weight coefficient and a correlation coefficient of each of a plurality of features of a to-be-processed object, wherein the weight coefficient of each feature is determined according to the relevance of the feature to a to-be-processed task, and the correlation coefficient of each feature is determined according to the correlation between the feature and other features; and a second determination unit configured to determine a selected feature in the plurality of features based on the weight coefficient and the correlation coefficient of each of the plurality of features.

[0008] According to an aspect of the present disclosure, a training device of a feature selection model is provided, wherein the feature selection model comprises a feature extraction module, a weight module and a correlation module, the feature selection model is connected with a task processing model, and the training device comprises: a first obtaining unit configured to input a plurality of features of a sample object into the feature extraction module to obtain representation information of each of the plurality of features; a second obtaining unit configured to obtain a weight coefficient of each of the plurality of features based on the representation information of each of the plurality of features by using the weight module; a third obtaining unit configured to obtain a correlation coefficient of each of the plurality of features based on the representation information of each of the plurality of features by using the correlation module; a third determination unit configured to determine a prediction selected feature in the plurality of features based on the weight coefficient and the correlation coefficient of each of the plurality of features; and an adjusting unit configured to adjust a parameter of the feature selection model based on the weight coefficient and the correlation coefficient of the prediction selected feature and a feedback result of the prediction selected feature, wherein the feedback result of the prediction selected feature is determined according to a prediction result of the task processing model on the prediction selected feature.

[0009] According to another aspect of the present disclosure, an electronic device is provided, comprising: at least one processor; and a memory connected with the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method described above.

[0010] According to another aspect of the present disclosure, a non-transitory computer readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable a computer to perform the method described above.

[0011] According to another aspect of the present disclosure, a computer program product is provided, comprising a computer program, wherein the computer program, when executed by a processor, implements the method described above.

[0012] According to one or more embodiments of the present disclosure, the execution efficiency of a computer device on a to-be-processed task can be improved on the basis of guaranteeing the execution effect of the to-be-processed task.

[0013] It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of the present disclosure, nor is it used to limit the scope of the present disclosure. Other features of the present disclosure will become apparent through the following description. BRIEF DESCRIPTION OF DRAWINGS

[0014] The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification, illustrate embodiments and together with the description serve to explain exemplary implementations of the application. The illustrated embodiments are exemplary only and not limiting of the scope of the claims. In all the drawings, like reference numerals refer to like parts throughout the several views.

[0015] Figure 1 A schematic diagram of an exemplary system in which the various methods described herein can be implemented according to embodiments of the present disclosure is shown;

[0016] Figure 2 A flowchart of a computer-implemented feature selection method according to embodiments of the present disclosure is shown;

[0017] Figure 3 A schematic diagram of a feature selection method according to embodiments of the present disclosure is shown;

[0018] Figure 4 A flowchart of a training method of a feature selection model according to embodiments of the present disclosure is shown;

[0019] Figure 5 A structural block diagram of a feature selection device according to embodiments of the present disclosure is shown;

[0020] Figure 6A structural block diagram of a training apparatus of a feature selection model is shown according to an embodiment of the present disclosure; and

[0021] Figure 7 A structural block diagram of an exemplary electronic device that can be used to implement embodiments of the present disclosure is shown. DETAILED DESCRIPTION

[0022] Exemplary embodiments of the present disclosure are described below with reference to the accompanying drawings, in which various details of embodiments of the present disclosure are set forth to assist in understanding the present disclosure. It will be apparent, however, that various changes and modifications can be made to the embodiments described herein without departing from the scope of the present disclosure. Also, the description is made in the following order for clarity and conciseness.

[0023] In the present disclosure, the terms "first", "second", and the like are used to describe various elements only and do not intend to limit the positional relationship, the chronological relationship, or the importance of the elements, and such terms are used only to distinguish one element from another. In some examples, the first element and the second element can refer to the same instance of the element, and in some cases, based on the context of the description, they can also refer to different instances.

[0024] The terms used in the description of various examples in the present disclosure are only for the purpose of describing the specific examples and are not intended to be limiting. Unless the number of elements is specifically limited, the element can be one or more than one, unless the context clearly indicates otherwise. In addition, the term "and / or" used in the present disclosure encompasses any one of the listed items and all possible combinations thereof.

[0025] In the field of artificial intelligence technology, data processing for a to-be-processed object is often performed in reliance on features extracted from the to-be-processed object. For a to-be-processed object, a plurality of features of the to-be-processed object can be obtained from a plurality of data sources, which describe the attributes of the to-be-processed object from different aspects and distinguish the to-be-processed object from other objects.

[0026] In the related art, regardless of the to-be-processed task to be performed by the to-be-processed object, data processing is performed on all features of the to-be-processed object by a computing device. With the development of big data technology, the number of features of the to-be-processed object that can be obtained has greatly expanded. Undoubtedly, the processing method in the related art will cause a great burden on the processing resources of the computing device and result in low processing efficiency of the computing device.

[0027] Based on this, the present disclosure proposes a computer-implemented feature selection method, which determines the degree of dependence of the execution of a to-be-processed task on each feature in a plurality of features based on both a weight coefficient and a correlation coefficient of the feature, and then filters out selected features matching the to-be-processed task from the plurality of features of the to-be-processed object to execute the to-be-processed task, thereby effectively reducing the data processing amount on the basis of ensuring the execution effect of the to-be-processed task, and further reducing the resource consumption of the computing device in the execution process of the to-be-processed task, reducing the configuration requirements of the computing device, and effectively improving the execution efficiency of the to-be-processed task.

[0028] Embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings.

[0029] Figure 1 A schematic diagram of an example system 100 in which various methods and apparatus described herein can be implemented according to embodiments of the present disclosure is shown. Referring to Figure 1 , the system 100 includes one or more client devices 101, 102, 103, 104, 105, and 106, a server 120, and one or more communication networks 110 coupling the one or more client devices to the server 120. The client devices 101, 102, 103, 104, 105, and 106 can be configured to execute one or more application programs.

[0030] In embodiments of the present disclosure, the server 120 can run one or more services or software applications that enable execution of a feature selection method or a training method of a feature selection model.

[0031] In certain embodiments, the server 120 can also provide other services or software applications that can include non-virtual environments and virtual environments. In certain embodiments, these services can be provided as web-based services or cloud services, for example, to users of the client devices 101, 102, 103, 104, 105, and / or 106 under a software as a service (SaaS) model.

[0032] In Figure 1 In the configuration shown, the server 120 can include one or more components that implement the functionality performed by the server 120. These components can include software components executable by one or more processors, hardware components, or a combination thereof. Users operating the client devices 101, 102, 103, 104, 105, and / or 106 can in turn utilize one or more client application programs to interact with the server 120 to utilize the services provided by these components. It should be understood that various different system configurations are possible, which can differ from the system 100. Thus, Figure 1 is one example of a system for implementing the various methods described herein and is not intended to be limiting.

[0033] A user can use client devices 101, 102, 103, 104, 105, and / or 106 to obtain objects to be processed. The client devices can provide an interface that enables a user of the client device to interact with the client device. The client devices can also output information to the user via the interface. Although Figure 1 Only six client devices are depicted, but those skilled in the art will appreciate that the present disclosure can support any number of client devices.

[0034] Client devices 101, 102, 103, 104, 105, and / or 106 can include various types of computer devices, such as portable handheld devices, general purpose computers (such as personal computers and laptop computers), workstation computers, wearable devices, smart screen devices, self-service kiosk devices, service robots, gaming systems, thin clients, various messaging devices, sensors or other sensing devices, and the like. These computer devices can run various types and versions of software applications and operating systems, such as MICROSOFT Windows, APPLE iOS, UNIX-like operating systems, Linux or Linux-like operating systems (such as GOOGLE Chrome OS); or include various mobile operating systems, such as MICROSOFT Windows Mobile OS, iOS, Windows Phone, Android. Portable handheld devices can include cellular telephones, smartphones, tablet computers, personal digital assistants (PDAs), and the like. Wearable devices can include head-mounted displays (such as smart glasses) and other devices. Gaming systems can include various handheld gaming devices, Internet-enabled gaming devices, and the like. The client devices are capable of executing various different applications, such as various Internet-related applications, communication applications (such as email applications), short message service (SMS) applications, and can use various communication protocols.

[0035] Networks 110 can be any type of networks known to those skilled in the art that can support data communications using any of a number of available protocols, including without limitation TCP / IP, SNA, IPX, etc. As examples only, one or more of networks 110 can be a local area network (LAN), an Ethernet-based network, a token ring, a wide area network (WAN), the Internet, a virtual network, a virtual private network (VPN), an intranet, an extranet, a public switched telephone network (PSTN), an infrared network, a wireless network (e.g., Bluetooth, WIFI), and / or any combination of these and / or other networks.

[0036] The server 120 can include one or more general purpose computers, special purpose server computers (e.g., PC (personal computer) servers, UNIX servers, mid-range servers), blade servers, mainframe computers, server clusters, or any other appropriate arrangement and / or combination. The server 120 can include one or more virtual machines running virtual operating systems, or other computing architectures involving virtualization (e.g., one or more flexible pools of logical storage devices that can be virtualized to maintain virtual storage devices for the servers). In various embodiments, the server 120 can run one or more services or software applications that provide the functionality described below.

[0037] The computing units in the server 120 can run one or more operating systems including any of the operating systems described above, as well as any commercially available server operating systems. The server 120 can also run any of a variety of additional server applications and / or mid-tier applications, including HTTP servers, FTP servers, CGI servers, JAVA servers, database servers, etc.

[0038] In some embodiments, the server 120 can include one or more applications to analyze and consolidate data feeds and / or event updates from users of the client devices 101, 102, 103, 104, 105, and / or 106. The server 120 can also include one or more applications to display the data feeds and / or real-time events via one or more display devices of the client devices 101, 102, 103, 104, 105, and / or 106.

[0039] In some embodiments, the server 120 can be a server of a distributed system, or a server combined with a blockchain. The server 120 can also be a cloud server, or an intelligent cloud computing server or intelligent cloud host with artificial intelligence technology. The cloud server is a host product in the cloud computing service system, to solve the defects of large management difficulty and weak business scalability in traditional physical host and virtual private server (VPS, Virtual Private Server) services.

[0040] The system 100 can also include one or more databases 130. In certain embodiments, these databases can be used to store data and other information. For example, one or more of the databases 130 can be used to store information such as audio files and video files. The databases 130 can reside in various locations. For example, databases used by the server 120 can be local to the server 120, or can be remote from the server 120 and can communicate with the server 120 via a network- based or dedicated connection. The databases 130 can be of different types. In certain embodiments, databases used by the server 120 can be, for example, relational databases. One or more of these databases can store, update, and retrieve data to and from the databases in response to commands.

[0041] In certain embodiments, one or more of the databases 130 can also be used by applications to store application data. Databases used by applications can be databases of different types, such as key-value stores, object stores, or regular stores supported by file systems.

[0042] Figure 1 The system 100 can be configured and operated in various ways to enable the application of various methods and apparatuses described in accordance with the present disclosure.

[0043] In the technical solutions of the present disclosure, the collection, storage, use, processing, transmission, provision, and disclosure of user personal information involved in the technical solutions comply with relevant laws and regulations and do not violate public order and good customs.

[0044] Figure 2 A flowchart of a computer-implemented feature selection method according to an embodiment of the present disclosure is shown in FIG. 2. As shown in FIG. 2, the method includes the following steps. Figure 2 S201: determining a weight coefficient and a correlation coefficient of each feature of a plurality of features of a to-be-processed object, wherein the weight coefficient of each feature is determined according to the relevance of the feature to a to-be-processed task, and the correlation coefficient of each feature is determined according to the correlation between the feature and other features; and S202: determining a selected feature for performing the to-be-processed task from the plurality of features based on the weight coefficient and the correlation coefficient of each feature of the plurality of features.

[0045] In this way, the data processing amount can be effectively reduced on the basis of ensuring the execution effect of the to-be-processed task, thereby reducing the resource consumption of the computing device in the execution process of the to-be-processed task, reducing the configuration requirements for the computing device, and effectively improving the execution efficiency of the to-be-processed task.

[0046] According to some embodiments, the to-be-processed task can include any one of: an image processing task; a speech processing task; a natural language processing task; and a data clustering task. Wherein, when the to-be-processed task is an image processing task, the to-be-processed object is image data; when the to-be-processed task is a speech processing task, the to-be-processed object is audio data; when the to-be-processed task is a natural language processing task, the to-be-processed object is text data.

[0047] For step S201, the weight coefficient of each feature is determined according to the degree of association between the feature and the to-be-processed task, in other words, the higher the dependence of the execution of the to-be-processed task on the feature, the greater the weight coefficient of the feature. For different to-be-processed tasks, the weight coefficient of the same feature in the to-be-processed object is different.

[0048] Wherein, the correlation coefficient of each feature is determined according to the correlation between the feature and other features. According to some embodiments, the correlation coefficient of each feature can be determined according to the correlation between the feature and other features in the process of executing the to-be-processed task. In other words, the correlation between the same feature in the to-be-processed object and other features changes with the change of the to-be-processed task. Determining the correlation between features based on the to-be-processed task can make the obtained correlation coefficient more in line with the application requirements of the to-be-processed task.

[0049] According to some embodiments, the correlation coefficient of each feature is determined based on a Gaussian Copula function. Thereby, the correlation relationship between the correlation coefficients of different features is conveniently established.

[0050] Wherein, the Gaussian Copula function can link the marginal distribution function and the joint distribution function, and the variables in the marginal distribution function satisfy the uniform distribution in the range of [0, 1].

[0051] For step S202, according to some embodiments, determining the selected feature for executing the to-be-processed task from the plurality of features based on the weight coefficient and the correlation coefficient of each feature in the plurality of features can include: for each feature in the plurality of features, determining an indication value of the feature based on the weight coefficient and the correlation coefficient of the feature; and in response to the indication value of the feature satisfying a preset condition, determining the feature as the selected feature. Thereby, it can be conveniently judged whether the feature is the selected feature based on the indication value determined by the weight coefficient and the correlation coefficient.

[0052] In an implementation, the indication function about the weight coefficient and the correlation coefficient can be constructed by the following way:

[0053]

[0054] Wherein, is the indication function of feature l, and αl U represents the weight coefficient of feature l. l Let g be the correlation coefficient of feature l. l For u l Intermediate variables.

[0055] t is an adjustment parameter; when the value of t approaches 0, satisfy:

[0056]

[0057] Therefore, when If the value of t approaches 0 and equals 1, meaning the above indicator value meets the preset condition, then this feature can be identified as the selected feature. If the value of t approaches 0 and equals 0, meaning the above indicator value does not meet the preset conditions, then the feature will not be selected.

[0058] Through the above As can be seen from the expression, The value of u l and α l Differentiable. Thus, α can be derived using a neural network. l or u l The possible values ​​of are provided; in other words, it can be determined by targeting . The predicted and feedback values ​​are used to perform backpropagation gradient calculation, which is then used to train a algorithm to derive u. l or α l Neural networks.

[0059] Furthermore, the correlation coefficient u mentioned above can be calculated by constructing a Gaussian Copula function. l , specifically,

[0060] Φ R (x1,…,x d ) = C gaussian (L(x1),…,L(x d );R);

[0061] Φ R (x1,…,x d )=P(g1≤x1,…,g d ≤x d );

[0062]

[0063] Among them, C gaussiandenotes a Gaussian Copula function, R is a correlation matrix of the Gaussian Copula function, the correlation matrix R is a positive definite matrix, and the correlation among the correlation coefficients of the plurality of features can be uniquely determined by the Gaussian Copula function constructed based on the R correlation matrix.

[0064] In order to calculate the positive definite correlation matrix R, the following expression can be used:

[0065] ∑=LL T +σ 2 I

[0066] R=Norm(LL T +σ 2 I)

[0067] wherein Norm(·) is used to map the covariance matrix to the correlation matrix R, and satisfies (Norm(∑)) ij =∑ ij / (∑ ii / ∑ jj ) 1 / 2 , i and j represent the horizontal and vertical coordinates of the matrix, respectively. The matrix L can be calculated by a corresponding neural network, and the matrix L can implicitly reflect the correlation among the plurality of features, σ 2 denotes a noise level, and I is a unit matrix.

[0068] The Cholesky decomposition of the matrix ∑ can further obtain the Cholesky factor V, and then the Gaussian vector q can be calculated, wherein q=(q1, q2, …, q d d represents the number of the plurality of features of the object to be processed. Specifically,

[0069] q=Vζ

[0070] wherein ζ is a Gaussian noise vector obtained by a standard normal distribution.

[0071] By applying the Gaussian Copula function constructed based on the R correlation matrix, the correlation coefficient u l of the feature l can be calculated, that is,

[0072] u l =Φ R (q l )

[0073] According to the embodiment, the number of the selected features to be selected is a non-fixed value, and the number of the selected features can be different for different objects to be processed.

[0074] In another embodiment, the indicator function about the weight coefficient and the correlation coefficient can be constructed in the following manner:

[0075]

[0076] wherein, is an indicator function of feature l, k is a preset number of selected features to be selected, may be further represented as:

[0077]

[0078] wherein, d represents a number of features of the object to be processed, t is an adjustment parameter, when satisfies that it equals to 1 when the value of t tends to 0, i.e., the above indicator value satisfies the preset condition, the feature can be determined as a selected feature, when satisfies that it equals to 0 when the value of t tends to 0, i.e., the above indicator value does not satisfy the preset condition, the feature is not determined as a selected feature.

[0079] wherein, may be further recursively represented as:

[0080] when s e {2, 3, …, k}

[0081] when s = 1

[0082] wherein, α l is a weight coefficient of feature l, u l is a correlation coefficient of feature l. It can be seen that, the value of is differentiable with respect to u l and α l . Thus, it provides a possibility to derive the value of α l or u l through a neural network, in other words, the neural network for deriving u l or α l may be trained by performing a backward gradient calculation with respect to the predicted value and the feedback value of .

[0083] According to the embodiment, the number of selected features to be selected is a fixed value k, and for any object to be processed, the k most matching features among the multiple features of the object to be processed can be selected as the selected features.

[0084] Figure 3 shows a schematic diagram of a feature selection method according to an embodiment of the present disclosure, as shown in Figure 3 , the task to be processed has d different features, which can be represented as {x1, x2, x3, …, x d}.

[0085] The weight parameter calculation module includes a plurality of full connection layers, and a weight matrix a composed of weight coefficients of each feature can be obtained based on the d different features.

[0086] The weight matrix a is input into the feature screening module, and an indication matrix z composed of an indication value of each feature can be constructed based on the weight matrix a and the correlation matrix u determined by the Gaussian copula submodule. In the indication matrix z, if the indication value corresponding to each feature is 1, it means that the feature is determined to be a selected feature, and if the indication value is 0, it means that the feature is not determined to be a selected feature. Thus, multiplying the d different features with their corresponding indication values can automatically screen out the selected features from the feature set {x1, x2, x3, …, xd} for input into the task processing model to perform the to-be-processed task. d} for input into the task processing model to perform the to-be-processed task.

[0087] The selected features are input into the task processing model, and the final processing result y of the task processing model on the to-be-processed object can be obtained.

[0088] Figure 4 A flowchart of a training method of a feature selection model according to an embodiment of the present disclosure is shown, wherein the feature selection model includes a feature extraction module, a weight module, and a correlation module, the feature selection model is connected with a task processing model, as shown in Figure 4 The training method includes: step S401, inputting a plurality of features of a sample object into the feature extraction module to obtain representation information of each feature in the plurality of features; step S402, based on the representation information of each feature in the plurality of features, using the weight module to obtain a weight coefficient of each feature in the plurality of features; step S403, based on the representation information of each feature in the plurality of features, using the correlation module to obtain a correlation coefficient of each feature in the plurality of features; step S404, based on the weight coefficient and the correlation coefficient of each feature in the plurality of features, determining a predicted selected feature for input into the task processing model in the plurality of features; and step S405, based on the weight coefficient and the correlation coefficient of the predicted selected feature, and a feedback result of the predicted selected feature, adjusting parameters of the feature selection model, wherein the feedback result of the predicted selected feature is determined according to a prediction result of the task processing model on the predicted selected feature.

[0089] Thus, the feature selection model for different task processing models can be trained, so that the trained feature selection model can automatically screen out features matched with the task processing model.

[0090] For step S403, according to some embodiments, obtaining, by the correlation module, the correlation coefficient of each of the plurality of features based on the representation information of each of the plurality of features can include: inputting the representation information of each of the plurality of features into the correlation module to obtain a correlation matrix; and determining the correlation coefficient of each of the plurality of features by using a Gaussian Copula function constructed based on the correlation matrix.

[0091] wherein the correlation matrix is the correlation matrix R described above. Since the correlation matrix R used for constructing the Gaussian Copula function is a positive definite matrix, in order to ensure that a positive definite matrix is obtained, the matrix L and the noise level σ 2 On this basis, the positive definite correlation matrix R is obtained according to R = Norm(LL T + σ 2 I).

[0092] For step S404, according to some embodiments, determining, from the plurality of features, the prediction-selected features for inputting into the task processing model based on the weight coefficient and the correlation coefficient of each of the plurality of features can include: determining, based on the weight coefficient and the correlation coefficient of each of the plurality of features, a prediction indication value of the feature; and for each of the plurality of features, in response to the prediction indication value of the feature satisfying a preset condition, determining the feature as a prediction-selected feature.

[0093] Specifically, the prediction indication value can be expressed in the form of or Since and are differentiable with respect to the weight coefficient and the correlation coefficient, the backpropagation can be performed during the training process of the feature selection model.

[0094] According to some embodiments, the feedback result is a feedback indication value of each of the prediction-selected features, and wherein adjusting the parameters of the feature selection model based on the weight coefficient and the correlation coefficient of the prediction-selected features and the feedback result of the prediction-selected features can include: for each of the prediction-selected features, adjusting the parameters of the feature selection model based on the prediction indication value of the feature and the feedback indication value of the feature.

[0095] Thus, the parameter adjustment of the feature extraction module, the weight module and the correlation module used for generating the weight coefficient and the correlation coefficient can be realized based on the difference between the prediction indication value and the feedback indication value of the selected feature.

[0096] According to some embodiments, the sample object has a label, and the feedback indication value is determined based on the label of the sample object and the prediction result of the prediction-selected features by the task processing model.

[0097] It can be understood that the training of the task processing model and the feature selection model can be performed simultaneously, and the parameters of both the task processing model and the feature selection model are adjusted reversely based on the difference between the label of the sample object and the prediction result of the prediction selected feature by the task processing model, so that the trained feature selection model can be matched with the task processing model, and the matched selected feature is screened out for the to-be-processed task performed in the task processing model.

[0098] Figure 5 A structural block diagram of a feature selection device according to an embodiment of the present disclosure is shown, as shown in the figure, the device 500 includes: a first determination unit 501 configured to determine a weight coefficient and a correlation coefficient of each feature of a plurality of features of a to-be-processed object, wherein the weight coefficient of each feature is determined according to the relevance of the feature to a to-be-processed task, and the correlation coefficient of each feature is determined according to the correlation between the feature and other features; and a second determination unit 502 configured to determine a selected feature for performing the to-be-processed task from the plurality of features based on the weight coefficient and the correlation coefficient of each feature of the plurality of features. Figure 5

[0099] According to some embodiments, the correlation coefficient of each feature is determined according to the correlation between the feature and other features in the process of performing the to-be-processed task.

[0100] According to some embodiments, the correlation coefficient of each feature is determined based on a Gaussian Copula function.

[0101] According to some embodiments, the second determination unit includes: a subunit for determining, for each feature of the plurality of features, an indication value of the feature based on the weight coefficient and the correlation coefficient of the feature; and a subunit for determining the feature as the selected feature in response to the indication value of the feature satisfying a preset condition.

[0102] According to some embodiments, the to-be-processed task includes any one of: an image processing task; a speech processing task; a natural language processing task; and a data clustering task.

[0103] Figure 6 A structural block diagram of a training device of a feature selection model according to an embodiment of the present disclosure is shown, wherein the feature selection model includes a feature extraction module, a weight module and a correlation module, and the feature selection model is connected with a task processing model, as shown in the figure, the device 500 includes: a first determination unit 501 configured to determine a weight coefficient and a correlation coefficient of each feature of a plurality of features of a to-be-processed object, wherein the weight coefficient of each feature is determined according to the relevance of the feature to a to-be-processed task, and the correlation coefficient of each feature is determined according to the correlation between the feature and other features; and a second determination unit 502 configured to determine a selected feature for performing the to-be-processed task from the plurality of features based on the weight coefficient and the correlation coefficient of each feature of the plurality of features. Figure 6 ​As shown, the apparatus 600 includes: a first obtaining unit 601 configured to input a plurality of features of a sample object into a feature extraction module to obtain representation information of each feature of the plurality of features; a second obtaining unit 602 configured to obtain a weight coefficient of each feature of the plurality of features by using a weight module based on the representation information of each feature of the plurality of features; a third obtaining unit 603 configured to obtain a correlation coefficient of each feature of the plurality of features by using a correlation module based on the representation information of each feature of the plurality of features; a third determining unit 604 configured to determine a prediction-selected feature for inputting into a task processing model among the plurality of features based on the weight coefficient and the correlation coefficient of each feature of the plurality of features; and an adjusting unit 605 configured to adjust a parameter of a feature selection model based on the weight coefficient and the correlation coefficient of the prediction-selected feature and a feedback result of the prediction-selected feature, wherein the feedback result of the prediction-selected feature is determined according to a prediction result of the prediction-selected feature by the task processing model.

[0104] According to some embodiments, the third determining unit includes: a sub-unit configured to determine a prediction indication value of each feature of the plurality of features based on the weight coefficient and the correlation coefficient of the feature; and a sub-unit configured to determine, for each feature of the plurality of features, the feature as the prediction-selected feature in response to the prediction indication value of the feature satisfying a preset condition.

[0105] According to some embodiments, the feedback result is a feedback indication value of each feature of the prediction-selected features, and the adjusting unit includes: a sub-unit configured to adjust, for each feature of the prediction-selected features, the parameter of the feature selection model based on the prediction indication value of the feature and the feedback indication value of the feature.

[0106] According to some embodiments, the sample object has a label, and the feedback indication value is determined based on the label of the sample object and the prediction result of the prediction-selected feature by the task processing model.

[0107] According to some embodiments, the third obtaining unit includes: a sub-unit configured to input the representation information of each feature of the plurality of features into the correlation module to obtain a correlation matrix; and a sub-unit configured to determine the correlation coefficient of each feature of the plurality of features by using a Gaussian Copula function constructed based on the correlation matrix.

[0108] According to an embodiment of the present disclosure, an electronic device is also provided, which includes: at least one processor; and a memory communicatively connected with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform any of the above methods.

[0109] According to an embodiment of the present disclosure, a non-transitory computer readable storage medium storing computer instructions is also provided, wherein the computer instructions are used to cause a computer to perform any of the above methods.

[0110] According to an embodiment of the present disclosure, a computer program product is also provided, comprising a computer program, wherein the computer program implements any of the above methods when executed by a processor.

[0111] Reference Figure 7 A block diagram of an electronic device 700, which can be used as the server or the client of the present disclosure, will now be described, which is an example of a hardware device that can be applied to aspects of the present disclosure. The electronic device is intended to represent a wide variety of digital electronic computer devices such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computing devices. The electronic device can also represent a wide variety of mobile devices such as personal digital assistants, cellular telephones, smart phones, wearable devices, and other similar computing devices. The components shown here, their connections and relationships, and their functions, are meant to be examples only, and are not meant to limit implementations of the present disclosure described and / or claimed in this document.

[0112] As shown in FIG. 7, Figure 7 The electronic device 700 includes a computing unit 701 that can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 702 or a computer program loaded into a random access memory (RAM) 703 from a storage unit 708. Various programs and data required for the operation of the electronic device 700 can also be stored in the RAM 703. The computing unit 701, the ROM 702, and the RAM 703 are connected to each other through a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0113] A plurality of components in the electronic device 700 are connected to the I / O interface 705, including an input unit 706, an output unit 707, a storage unit 708, and a communication unit 709. The input unit 706 can be any type of device that can input information to the electronic device 700, and can receive inputted digital or character information, and generate key signal inputs related to user settings and / or function controls of the electronic device, and can include, but is not limited to, a mouse, a keyboard, a touch screen, a track pad, a track ball, a joystick, a microphone, and / or a remote controller. The output unit 707 can be any type of device that can present information, and can include, but is not limited to, a display, a speaker, a video / audio output terminal, a vibrator, and / or a printer. The storage unit 708 can include, but is not limited to, a magnetic disk, an optical disk. The communication unit 709 allows the electronic device 700 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks, and can include, but is not limited to, a modem, a network card, an infrared communication device, a wireless communication transceiver, and / or a chipset, such as a Bluetooth™ device, an 802.11 device, a WiFi device, a WiMax device, a cellular communication device, and / or the like.

[0114] The computing unit 701 can be various general and / or special purpose processing components having processing and computing capabilities. Some examples of the computing unit 701 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The computing unit 701 performs various methods and processes described above, such as a feature selection method or a training method of a feature selection model. For example, in some embodiments, the feature selection method or the training method of the feature selection model can be implemented as a computer software program tangibly embodied in a machine-readable medium, such as the storage unit 708. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 700 via the ROM 702 and / or the communication unit 709. When the computer program is loaded to the RAM 703 and executed by the computing unit 701, one or more steps of the feature selection method or the training method of the feature selection model described above can be performed. Alternatively, in other embodiments, the computing unit 701 can be configured to perform the feature selection method or the training method of the feature selection model by any other appropriate means, such as by means of firmware.

[0115] The various embodiments of the systems and techniques described above can be implemented in digital electronic circuitry, integrated circuitry, a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), a system on a chip (SOC), a complex programmable logic device (CPLD), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.

[0116] Program code for carrying out methods of the present disclosure can be written in any combination of one or more programming languages. The program code can be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the program code, when executed by the processor or controller, produces the functions / operations specified in the flowcharts and / or the block diagrams. The program code can be executed entirely on a machine, partially on a machine, partially on a machine as a stand-alone software package, or entirely on a remote machine or server.

[0117] In the context of the present disclosure, a machine-readable medium can be a tangible medium that contains or stores a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include but is not limited to an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of the machine-readable storage medium will include one or more lines of electrical conductors, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or Flash memory), optical fibers, portable compact disc read-only memories (CD-ROMs), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0118] To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the computer. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.

[0119] The systems and techniques described here can be implemented in a computing system that includes a back end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front end component (e.g., a user computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0120] The computer system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0121] It should be understood that the various forms of flow shown above can be re-ordered, added to, or deleted from without departing from the spirit of the present disclosure. For example, the steps recited in the present disclosure can be performed in parallel, in series, or in a different order, as long as the desired results of the technology disclosed in the present disclosure are achieved, which is not limited herein.

[0122] While embodiments or examples of this disclosure have been described with reference to the figures, it will be understood that the methods, systems, and devices described above are merely exemplary embodiments or examples, and the scope of the application is not limited to these embodiments or examples. Various elements of the embodiments or examples can be omitted or substituted by equivalents thereof. Furthermore, the steps can be performed in a different order than described in the disclosure. Further, various elements of the embodiments or examples can be combined in various ways. It is important that as technology evolves, many of the elements described herein can be substituted by equivalents which serve the same function.

Claims

1. A computer-implemented feature selection method, comprising: determining a weight coefficient and a correlation coefficient of each of a plurality of features of an object to be processed, wherein the weight coefficient of each feature is determined according to an association degree of the feature to a task to be processed, and the correlation coefficient of each feature is determined according to a correlation between the feature and other features, wherein the task to be processed comprises any one of an image processing task, a speech processing task, and a natural language processing task, and wherein the object to be processed is image data when the task to be processed is the image processing task, is audio data when the task to be processed is the speech processing task, and is text data when the task to be processed is the natural language processing task; and determining selected features for performing the task to be processed from the plurality of features based on the weight coefficient and the correlation coefficient of each of the plurality of features, wherein the determining comprises: For each of the plurality of features, an indicative value of the feature is determined based on the weight coefficient and the correlation coefficient of the feature, wherein the indicative value is determined based on any one of the following indicative functions: wherein, or wherein, k is a predetermined number of selected features to be selected, wherein, d represents a number of features of the object to be processed, may be further recursively represented as: When Time, When Time, = wherein wherein is a weight coefficient, l is a correlation coefficient, is a weight coefficient, l is a correlation coefficient, t is an adjustment parameter; and determining the feature as the selected feature in response to the indication value of the feature satisfying a preset condition, wherein the indication value of the feature satisfies the preset condition when the indication value of the feature equals 1 when the value of the indication value of the feature approaches 0, and the indication value of the feature satisfies the preset condition. determining the feature as the selected feature in response to the indication value of the feature satisfying a preset condition, wherein the indication value of the feature satisfies the preset condition when the indication value of the feature equals 1 when the value of the indication value of the feature approaches 0, and the indication value of the feature satisfies the preset condition.

2. The method of claim 1, wherein, the correlation coefficient of each feature is determined according to a correlation between the feature and other features in a process of performing the task to be processed.

3. The method of claim 1 or 2, wherein, the correlation coefficient of each feature is determined based on a Gaussian Copula function.

4. The method according to any one of claims 1 to 3, wherein, the task to be processed further comprises a data clustering task.

5. A method of training a feature selection model, wherein, the feature selection model comprises a feature extraction module, a weight module, and a correlation module, and the feature selection model is connected to a task processing model, wherein the task processing model is used to process any one of an image processing task, a speech processing task, and a natural language processing task, and wherein the training method comprises: inputting a plurality of features of a sample object to the feature extraction module to obtain representation information of each of the plurality of features, wherein the sample object is image data when processing an image processing task, is audio data when processing a speech processing task, and is text data when processing a natural language processing task; obtaining a weight coefficient of each of the plurality of features based on the representation information of each of the plurality of features by using the weight module; obtaining a correlation coefficient of each of the plurality of features based on the representation information of each of the plurality of features by using the correlation module; determining predicted selected features for inputting to the task processing model from the plurality of features based on the weight coefficient and the correlation coefficient of each of the plurality of features, wherein the determining comprises: Based on the weight coefficient and correlation coefficient of each of the plurality of features, a predicted indicator value for that feature is determined, wherein the predicted indicator value... Determined based on any of the following indicator functions: wherein, or wherein, k is a predetermined number of selected features to be selected, wherein, d represents a number of features of the object to be processed, may be further recursively represented as: When time, When Time, = ​ in, ,in, Features l The weighting coefficients, Features l The correlation coefficient, where t is the adjustment parameter; and For each of the plurality of features, in response to the predicted indication value of the feature satisfying a preset condition, the feature is determined as the predicted selected feature, wherein the predicted indication value of the feature satisfies the preset condition when the predicted indication value of the feature equals 1 when the value of the predicted indication value of the feature approaches 0, the predicted indication value of the feature satisfies the preset condition, and the feature is determined as the selected feature. ​ adjusting parameters of the feature selection model based on the weight coefficient and the correlation coefficient of the predicted selected features and a feedback result of the predicted selected features, wherein the feedback result of the predicted selected features is determined according to a prediction result of the task processing model on the predicted selected features.

6. The method of claim 5, wherein, the feedback result is a feedback indication value of each of the predicted selected features, and wherein the adjusting the parameters of the feature selection model based on the weight coefficient and the correlation coefficient of the predicted selected features and the feedback result of the predicted selected features comprises: adjusting, for each of the predicted selected features, a parameter of the feature selection model based on the predicted indication value of the feature and a feedback indication value of the feature.

7. The method of claim 6, wherein, The sample object has a label, and the feedback indication value is determined based on the label of the sample object and a prediction result of the predicted selected features by the task processing model.

8. The method of claim 5, wherein, The obtaining, by the correlation module, of the correlation coefficient of each of the plurality of features based on the representation information of each of the plurality of features comprises: inputting the representation information of each of the plurality of features into the correlation module to obtain a correlation matrix; and determining the correlation coefficient of each of the plurality of features by using a Gaussian Copula function constructed based on the correlation matrix.

9. A feature selection apparatus, comprising: a first determining unit configured to determine a weight coefficient and a correlation coefficient of each of a plurality of features of a to-be-processed object, wherein the weight coefficient of each feature is determined according to an association degree of the feature to a to-be-processed task, and the correlation coefficient of each feature is determined according to a correlation between the feature and other features, wherein the to-be-processed task comprises any one of an image processing task, a speech processing task, and a natural language processing task, and wherein, when the to-be-processed task is the image processing task, the to-be-processed object is image data, when the to-be-processed task is the speech processing task, the to-be-processed object is audio data, and when the to-be-processed task is the natural language processing task, the to-be-processed object is text data; and a second determining unit configured to determine, based on the weight coefficient and the correlation coefficient of each of the plurality of features, selected features for performing the to-be-processed task from the plurality of features, wherein the weight coefficient of each feature is determined according to an association degree of the feature to the to-be-processed task, and the correlation coefficient of each feature is determined according to a correlation between the feature and other features in a process of performing the to-be-processed task. For each of the plurality of features, an indicative value of the feature is determined based on the weight coefficient and the correlation coefficient of the feature, wherein the indicative value is determined based on any one of the following indicative functions: wherein, or wherein, k is a predetermined number of selected features to be selected, wherein, d denotes the number of features of the object to be processed, may be further recursively represented as: When time, When Time, = ​ wherein wherein is a weight coefficient, l is a correlation coefficient, is a correlation coefficient, l t is an adjustment parameter; and determining the feature as the selected feature in response to the indication value of the feature satisfying a preset condition, wherein the indication value of the feature satisfies the preset condition when the indication value of the feature equals 1 when the value of the indication value of the feature approaches 0, and the indication value of the feature satisfies the preset condition. determining the feature as the selected feature in response to the indication value of the feature satisfying a preset condition, wherein the indication value of the feature satisfies the preset condition when the indication value of the feature equals 1 when the value of the indication value of the feature approaches 0, and the indication value of the feature satisfies the preset condition.

10. The apparatus of claim 9, wherein, The correlation coefficient of each feature is determined based on a Gaussian Copula function.

11. The apparatus of claim 9 or 10, wherein, The to-be-processed task further comprises a data clustering task.

12. The apparatus of any of claims 9 to 11, wherein, The feature selection model comprises a feature extraction module, a weight module, and a correlation module, and the feature selection model is connected to a task processing model, wherein the task processing model is used to process any one of an image processing task, a speech processing task, and a natural language processing task, and wherein the training apparatus comprises:

13. An apparatus for training a feature selection model, wherein, a first obtaining unit configured to input a plurality of features of a sample object into the feature extraction module to obtain representation information of each of the plurality of features, wherein, when an image processing task is processed, the sample object is image data, when a speech processing task is processed, the sample object is audio data, and when a natural language processing task is processed, the sample object is text data; a second obtaining unit configured to obtain, based on the representation information of each of the plurality of features, a weight coefficient of each of the plurality of features by using the weight module; and a third obtaining unit configured to obtain, by using the correlation module, a correlation coefficient of each of the plurality of features based on the representation information of each of the plurality of features. a third obtaining unit, configured to obtain, by using the correlation module, a correlation coefficient of each of the plurality of features based on representation information of each of the plurality of features; a third determining unit, configured to determine, among the plurality of features, a prediction selected feature for inputting into the task processing model based on the weight coefficient and the correlation coefficient of each of the plurality of features, wherein the prediction selected feature comprises: determining a prediction indication value of each of the plurality of features based on a weight coefficient and a correlation coefficient of the feature, wherein the prediction indication value of the feature is determined based on any one of the following indication functions: determining based on any one of the following indication functions: wherein, or wherein, k is a preset number of selected features to be selected, wherein, d represents a number of features of the object to be processed, may be further recursively represented as: When Time, When time, = wherein wherein is a weight coefficient, l is a correlation coefficient, is a weight coefficient, l is a correlation coefficient, t is an adjustment parameter; and For each of the plurality of features, in response to the predicted indication value of the feature satisfying a preset condition, the feature is determined as the predicted selected feature, wherein the predicted indication value of the feature satisfies the preset condition when the predicted indication value of the feature equals 1 when a value of the predicted indication value of the feature approaches 0, the predicted indication value of the feature satisfies the preset condition, the feature is determined as the selected feature; and the predicted indication value of the feature satisfies the preset condition, the feature is determined as the selected feature; and an adjusting unit, configured to adjust a parameter of the feature selection model based on the weight coefficient and the correlation coefficient of the prediction selected feature and a feedback result of the prediction selected feature, wherein the feedback result of the prediction selected feature is determined according to a prediction result of the task processing model on the prediction selected feature.

14. The apparatus of claim 13, wherein, The feedback result is a feedback indication value of each of the prediction selected features, and the adjusting unit comprises: a subunit for adjusting, for each of the prediction selected features, the parameter of the feature selection model based on the prediction indication value of the feature and the feedback indication value of the feature.

15. The apparatus of claim 14, wherein, The sample object has a label, and the feedback indication value is determined based on the label of the sample object and the prediction result of the task processing model on the prediction selected feature.

16. The apparatus of claim 13, wherein, The third obtaining unit comprises: a subunit for inputting the representation information of each of the plurality of features into the correlation module to obtain a correlation matrix; and a subunit for determining the correlation coefficient of each of the plurality of features by using a Gaussian Copula function constructed based on the correlation matrix.

17. An electronic device, comprising: at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-8.

18. A non-transitory computer readable storage medium having stored thereon computer instructions, wherein, The computer instructions are used to enable the computer to perform the method of any one of claims 1-8.

19. A computer program product comprising a computer program, wherein, The computer program, when executed by a processor, implements the method of any one of claims 1-8. The computer program, when executed by a processor, implements the method of any one of claims 1-8.

Images