CN113408665A - Object identification method, device, equipment and medium - Google Patents

Object identification method, device, equipment and medium Download PDF

Info

Publication number
CN113408665A
CN113408665A CN202110829878.6A CN202110829878A CN113408665A CN 113408665 A CN113408665 A CN 113408665A CN 202110829878 A CN202110829878 A CN 202110829878A CN 113408665 A CN113408665 A CN 113408665A
Authority
CN
China
Prior art keywords
clustering
feature
features
objects
feature selection
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202110829878.6A
Other languages
Chinese (zh)
Inventor
祖辰
杨立军
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Jingdong Century Trading Co Ltd
Beijing Wodong Tianjun Information Technology Co Ltd
Original Assignee
Beijing Jingdong Century Trading Co Ltd
Beijing Wodong Tianjun Information Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing Jingdong Century Trading Co Ltd, Beijing Wodong Tianjun Information Technology Co Ltd filed Critical Beijing Jingdong Century Trading Co Ltd
Publication of CN113408665A publication Critical patent/CN113408665A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G06F18/253—Fusion techniques of extracted features

Landscapes

  • Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Image Analysis (AREA)

Abstract

The present disclosure provides an object recognition method, including: obtaining a plurality of objects to be identified, each object comprising a plurality of features; and selecting a plurality of characteristics of each object by adopting the characteristic selection matrix to obtain a plurality of objects comprising at least one characteristic. And clustering a plurality of objects comprising at least one characteristic to obtain a clustering result. Under the condition that the clustering result meets the preset condition, identifying information included by the objects according to the clustering result; and under the condition that the clustering result is determined not to meet the preset condition, updating the feature selection matrix according to the clustering result, and returning to the operation of respectively selecting the plurality of features of each object by adopting the feature selection matrix. The present disclosure also provides an object recognition apparatus, an electronic device and a computer-readable storage medium.

Description

Object identification method, device, equipment and medium
Technical Field
The present disclosure relates to the field of computer technologies, and in particular, to a method, an apparatus, a device, and a medium for object recognition.
Background
With the rapid growth of high-dimensional data, high-dimensional data clustering plays an increasingly important role in many fields, such as text mining, image searching, computer vision, and bioinformatics. For example, in the text mining process, the information included in the text is clustered, and the information included in the text is identified based on the clustering result, so that the unknown, understandable and finally available knowledge and information in the text can be accurately mined. For another example, in image searching, images are clustered, key information included in the images is identified based on a clustering result, and then the images to be searched are efficiently and accurately searched based on the key information. Therefore, how to accurately cluster the texts or the images directly affects the accuracy of text or image information identification, and further affects the efficiency and the accuracy of text mining or image searching.
Disclosure of Invention
In view of the above, the present disclosure provides an object recognition method, apparatus, device and medium.
One aspect of the present disclosure provides an object recognition method, including: acquiring a plurality of objects to be identified, each of the objects comprising a plurality of features; selecting the plurality of characteristics of each object by adopting a characteristic selection matrix to obtain a plurality of objects comprising at least one characteristic; clustering the plurality of objects comprising the at least one characteristic to obtain a clustering result; under the condition that the clustering result meets the preset condition, identifying information included by the objects according to the clustering result; and under the condition that the clustering result is determined not to meet the preset condition, updating the feature selection matrix according to the clustering result, and returning to the operation of respectively selecting the plurality of features of each object by adopting the feature selection matrix.
According to an embodiment of the present disclosure, the selecting the plurality of features of each object by using a feature selection matrix to obtain a plurality of objects including at least one feature includes: respectively extracting the characteristics of different dimensions of each object; forming the feature of each object with different dimensions into a feature vector of each object; and respectively selecting the feature vector of each object by adopting a feature selection matrix to obtain the feature vector of each object comprising at least one feature.
According to an embodiment of the present disclosure, the selecting the plurality of features of each object by using a feature selection matrix to obtain a plurality of objects including at least one feature includes: constructing a row sparse matrix; and respectively selecting the plurality of characteristics of each object through the row sparse matrix to obtain a plurality of objects comprising at least one characteristic.
According to an embodiment of the present disclosure, the updating the feature selection matrix according to the clustering result when it is determined that the clustering result does not satisfy the preset condition includes: under the condition that the clustering accuracy included in the clustering result is determined to belong to a first preset range, updating the feature selection matrix to select different kinds of features; and updating the feature selection matrix to select different numbers of features under the condition that the clustering accuracy included in the clustering result is determined to belong to a second preset range.
According to an embodiment of the present disclosure, the clustering the plurality of objects including the at least one feature to obtain a clustering result includes: fixing the numerical values of any two variables in the clustering center, the feature selection matrix and the confidence coefficient, and calculating the numerical value of the other variable to obtain a group of numerical values comprising the three variables; and calculating a plurality of groups of numerical values comprising the three variables in an iterative mode to obtain a clustering center, the feature selection matrix and a convergence value of the confidence coefficient.
According to an embodiment of the present disclosure, the object includes at least one of text, image, voice, and video.
According to an embodiment of the present disclosure, the image set comprises a high dimensional image dataset.
According to an embodiment of the present disclosure, the high dimensional image dataset comprises a COIL20 dataset or a COIL100 dataset or an ORL face dataset or a YALE face dataset.
According to an embodiment of the present disclosure, the object is an image; the object comprises an image; the respectively extracting the features of different dimensions of each object comprises the following steps: extracting at least one of the following features of the image: gray value; a two-dimensional histogram; a scale invariant feature transform value; histogram of directional gradients.
Another aspect of the present disclosure provides an object recognition apparatus, including: an acquisition module for acquiring a plurality of objects to be identified, each of the objects comprising a plurality of features; a selection module, configured to select the multiple features of each object by using a feature selection matrix, respectively, to obtain multiple objects including at least one feature; the clustering module is used for clustering the plurality of objects comprising the at least one characteristic to obtain a clustering result; the first determining module is used for identifying information included by the objects according to the clustering result under the condition that the clustering result is determined to meet the preset condition; and a second determining module, configured to update the feature selection matrix according to the clustering result and return to the operation of selecting the plurality of features of each object by using the feature selection matrix, respectively, when it is determined that the clustering result does not satisfy the preset condition.
According to an embodiment of the present disclosure, the selection module includes: the extraction unit is used for respectively extracting the characteristics of different dimensions of each object; and the composition unit is used for composing the characteristics of each object with different dimensions into a characteristic vector of each object. The first selection unit is used for selecting the feature vector of each object by adopting the feature selection matrix to obtain the feature vector of each object including at least one feature.
According to an embodiment of the present disclosure, the selection module further comprises: the construction unit is used for constructing a row sparse matrix; and the second selection unit is used for respectively selecting the plurality of characteristics of each object through the row sparse matrix to obtain a plurality of objects comprising at least one characteristic.
According to an embodiment of the present disclosure, the second determining module includes: a first determining unit, configured to update the feature selection matrix to select different kinds of features when it is determined that the clustering accuracy included in the clustering result belongs to a first preset range; and the second determining unit is used for updating the feature selection matrix to select different numbers of features under the condition that the clustering accuracy included in the clustering result is determined to belong to a second preset range.
According to an embodiment of the present disclosure, the clustering module includes: the first calculation unit is used for fixing numerical values of any two variables in the clustering center, the feature selection matrix and the confidence coefficient, and calculating the numerical value of the other variable to obtain a group of numerical values comprising the three variables; and the second calculation unit is used for calculating a plurality of groups of numerical values comprising the three variables in an iterative mode to obtain a clustering center, the feature selection matrix and a convergence value of the confidence coefficient.
Another aspect of the present disclosure provides an electronic device including: one or more processors; memory for storing one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the method as described above.
Another aspect of the present disclosure provides a computer-readable storage medium storing computer-executable instructions for implementing the method as described above when executed.
Another aspect of the disclosure provides a computer program comprising computer executable instructions for implementing the method as described above when executed.
Drawings
The above and other objects, features and advantages of the present disclosure will become more apparent from the following description of embodiments of the present disclosure with reference to the accompanying drawings, in which:
FIG. 1 schematically illustrates an exemplary system architecture 100 in which an object recognition method may be implemented, according to an embodiment of the present disclosure;
FIG. 2 schematically illustrates a flow chart of an object recognition method according to an embodiment of the present disclosure;
FIG. 3 schematically illustrates a flow diagram of selecting features according to an embodiment of the disclosure;
FIG. 4 schematically illustrates a select features flow diagram according to another embodiment of the present disclosure;
FIG. 5 schematically shows a flow chart of a clustering method according to an embodiment of the present disclosure;
FIG. 6 is a flow chart schematically illustrating a weighted iterative solution method according to an embodiment of the present disclosure
Fig. 7 schematically shows a block diagram of an object recognition arrangement according to an embodiment of the present disclosure;
FIG. 8 schematically shows a block diagram of a selection module according to an embodiment of the disclosure;
FIG. 9 schematically shows a block diagram of a selection module according to another embodiment of the present disclosure;
fig. 10 schematically shows a block diagram of an electronic device adapted to implement the above described method according to an embodiment of the present disclosure.
Detailed Description
Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. It should be understood that the description is illustrative only and is not intended to limit the scope of the present disclosure. In the following detailed description, for purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the embodiments of the disclosure. It may be evident, however, that one or more embodiments may be practiced without these specific details. Moreover, in the following description, descriptions of well-known structures and techniques are omitted so as to not unnecessarily obscure the concepts of the present disclosure.
The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. The terms "comprises," "comprising," and the like, as used herein, specify the presence of stated features, steps, operations, and/or components, but do not preclude the presence or addition of one or more other features, steps, operations, or components.
All terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art unless otherwise defined. It is noted that the terms used herein should be interpreted as having a meaning that is consistent with the context of this specification and should not be interpreted in an idealized or overly formal sense.
Where a convention analogous to "at least one of A, B and C, etc." is used, in general such a construction is intended in the sense one having skill in the art would understand the convention (e.g., "a system having at least one of A, B and C" would include but not be limited to systems that have a alone, B alone, C alone, a and B together, a and C together, B and C together, and/or A, B, C together, etc.). Where a convention analogous to "A, B or at least one of C, etc." is used, in general such a construction is intended in the sense one having skill in the art would understand the convention (e.g., "a system having at least one of A, B or C" would include but not be limited to systems that have a alone, B alone, C alone, a and B together, a and C together, B and C together, and/or A, B, C together, etc.).
The clustering mode of the high-dimensional data comprises the following steps: firstly, feature selection is carried out on high-dimensional data to realize dimension reduction, and then clustering is carried out on the data subjected to dimension reduction by using a clustering algorithm. However, the feature selection task in the method is unrelated to the subsequent clustering task, the feature selection is likely to screen out the features useful for the clustering task, the subsequent clustering performance is directly influenced, and the clustering accuracy is low.
The embodiment of the disclosure provides an object identification method and a device capable of applying the method. The method includes obtaining a plurality of objects to be identified, each object including a plurality of features. And selecting a plurality of characteristics of each object by adopting the characteristic selection matrix to obtain a plurality of objects comprising at least one characteristic. And clustering a plurality of objects comprising at least one characteristic to obtain a clustering result. And judging whether the clustering result meets a preset condition, if not, updating the feature selection matrix according to the clustering result, returning to the previous operation, and if so, executing the next operation. And identifying information included by the plurality of objects according to the clustering result.
Fig. 1 schematically illustrates an exemplary system architecture 100 in which an object recognition method may be implemented, according to an embodiment of the disclosure. It should be noted that fig. 1 is only an example of a system architecture to which the embodiments of the present disclosure may be applied to help those skilled in the art understand the technical content of the present disclosure, and does not mean that the embodiments of the present disclosure may not be applied to other devices, systems, environments or scenarios.
As shown in fig. 1, a system architecture 100 according to this embodiment may include a database 101, a network 102, and a server 103. Network 102 is the medium used to provide communication links between database 100 and server 103. Network 102 may include various connection types, such as wired and/or wireless communication links, and so forth.
The database 100 may store objects to be recognized, which may include, for example, text, images, voice, and video, the network 102 may input the objects to be recognized into the server 103, and the server 103 may be a server providing various services, such as feature selection and clustering of the objects to be recognized, and information included in the objects may be recognized according to the clustering result.
It should be noted that the object identification method provided by the embodiment of the present disclosure may be generally executed by the server 103. Accordingly, the object recognition apparatus provided by the embodiments of the present disclosure may be generally disposed in the server 103. The object recognition method provided by the embodiments of the present disclosure may also be performed by a server or a server cluster that is different from the server 103 and is capable of communicating with the database 101, and/or the server 103. Accordingly, the object recognition device provided by the embodiment of the present disclosure may also be disposed in a server or a server cluster different from the server 103 and capable of communicating with the database 101 and/or the server 103.
For example, in the object recognition, the process of clustering objects and the recognition of the information included in the objects according to the clustering result are not directly executed by the server 103, but executed by a server or a server cluster capable of communicating with the database 101 and the server 103, and after the object recognition is completed, the recognized information is transmitted to the server 103.
It should be understood that the number of databases, networks, and servers in fig. 1 are merely illustrative. There may be any number of databases, networks, and servers, as desired for implementation.
Fig. 2 schematically shows a flow chart of an object recognition method according to an embodiment of the present disclosure.
As shown in fig. 2, the object recognition method includes operations S201 to S205.
In operation S201, a plurality of objects to be recognized, each object including a plurality of features, is acquired.
For example, the obtained object to be recognized may be at least one of text, image, voice, and video with multidimensional characteristics, or may be multidimensional data expressed in the form of numerical values, vectors, or matrices.
In operation S202, a plurality of features of each object are respectively selected by using a feature selection matrix, so as to obtain a plurality of objects including at least one feature.
High-dimensional data has a number of characteristic attributes, i.e., its dimensions can reach hundreds of dimensions, or even higher. In the data identification process, partial features of high-dimensional data are screened out in a targeted manner to form new data, so that the calculation amount of data identification is reduced and the operation cost is reduced.
In operation S203, a plurality of objects including at least one feature are clustered, resulting in a clustering result.
In operation S204, in case that it is determined that the clustering result satisfies the preset condition, information included in the plurality of objects is identified according to the clustering result.
In operation S205, in a case that it is determined that the clustering result does not satisfy the preset condition, the feature selection matrix is updated according to the clustering result, and an operation of selecting the plurality of features of each object respectively using the feature selection matrix is returned.
According to the embodiment of the disclosure, the feature selection and clustering processes are performed synchronously, the feature selection and clustering processes are interdependent, and the result of the feature selection and the result of the clustering affect each other. The number and the types of the features actually participating in clustering are dynamically controlled in the clustering process, so that the phenomenon that features useful for the clustering process are discarded due to feature selection performed firstly according to prior experience is avoided.
In order to facilitate understanding of the object recognition method provided by the embodiments of the present disclosure, the following example is provided for description. It should be understood that this example is not intended to limit the present disclosure.
For example, the object identification method provided by the embodiment of the disclosure is applied to commodity recommendation. After the user purchases the goods through the shopping platform, the corresponding purchase information may be stored in the form of text, for example, the user ID, the purchase address, the type of the purchased goods, and the like, that is, the text content includes the information. In order to recommend interested commodities to corresponding users to obtain the relevance between the users, known information with values needs to be mined from texts as a reference for commodity recommendation, and the valuable information can include, for example, the period or frequency of purchasing a certain commodity by the user, and the period or frequency can accurately reflect the commodities required by the user at different times, so that the commodities required by the user can be recommended to the users with similar purchase frequency in batches at the correct time, and further the commodity purchase success rate is improved. At this time, text mining needs to be employed to extract the period or frequency of a user's purchase of a certain product from features including the type of purchased product, the number of times of purchasing the type of product, and the time interval of purchasing the type of product based on text data. By adopting the object identification method provided by the embodiment of the disclosure, massive texts comprising user IDs, purchase addresses and types of purchased commodities are taken as objects, and the massive texts are clustered, so that key information included in the texts can be efficiently and accurately identified according to clustering results, and valuable information recommended to the commodities is further mined.
For another example, the object recognition method provided by the embodiment of the present disclosure is applied to image search. When the method is used for quickly and accurately searching the needed images from massive data, the massive images to be searched are taken as objects, the massive images are clustered to obtain the clustering results of the images, information included in different types of images can be quickly identified based on the clustering results, and the images needing to be matched with the current images are searched based on the identified information, so that the searching efficiency is higher, and the matching is more accurate.
For another example, the object identification method provided by the embodiment of the present disclosure is applied to another commodity recommendation. In order to recommend similar commodities to each user, the obtained objects to be identified are massive original data including user purchase information, and the original data includes various characteristics, such as user age, purchase address, purchase time period, purchased commodity type and the like. And performing cluster analysis on the original data, namely classifying users corresponding to the original data to obtain a user set with certain feature similarity so as to recommend different commodities to users of different user sets. For the same original data, clustering can be performed based on different characteristics to obtain different clustering results.
For example, clustering is carried out based on the age of the user and the type characteristics of the purchased goods, and the clustering result is a user composition set for purchasing clothing goods by 20-30 years old and a user composition set for purchasing daily goods by 30-40 years old; for another example, clustering is performed based on the purchase address and the value characteristics of the purchased commodities, the clustering result is a user composition set of commodities with the purchase value of 500 yuan or less in the university region and the purchase address is a user composition set of commodities with the purchase value of 500 yuan or more in the business center; for another example, based on the gender, the time period of purchase and the type of goods purchased by the user, the clustering result is a female user group set for purchasing clothes at 20-24 points, a male user group set for purchasing electronic products at 22-24 points, a female user group set for purchasing fresh products at 7-9 points, and a male user group set for purchasing instant food at 11-13 points.
For the plurality of sets, each set has a different clustering quality. Understandably, the object identification method provided by the present disclosure needs to identify the maximum similarity of the objects, and needs to obtain the user set with the optimal clustering quality. Therefore, in the clustering process, the optimal solution can be obtained by determining which kind of features and the number of features are based on each original data, and accurate judgment cannot be performed according to prior. If feature selection is performed on high-dimensional raw data only according to historical feature selection experience, key features may be screened out, and an optimal clustering result cannot be obtained.
According to the object identification method provided by the embodiment of the disclosure, in the clustering process, the feature selection quantity and the type are continuously changed by dynamically and continuously updating the feature selection matrix. Compared with the traditional technical scheme that the feature selection task is unrelated to the subsequent clustering task, the method can avoid discarding features which have key effects on the clustering process in the feature selection, and further can accurately identify the key information included by the objects and the relevance among the objects according to the clustering result. And because clustering is carried out first and then different classes are respectively identified (the same class has commonality), information identification is not directly carried out from mass objects, and the efficiency of object identification is improved.
In order to more clearly illustrate the present invention, the object recognition method shown in fig. 2 is further described below with reference to specific embodiments.
FIG. 3 schematically illustrates a flow diagram of selecting features according to an embodiment of the disclosure.
As shown in fig. 3, operation S202 may further include operations S301 to S303.
In operation S301, features of different dimensions of each object are extracted, respectively.
According to the embodiment of the present disclosure, the object recognition method may be used for recognizing text, image, voice, video, and the like, and correspondingly processing high-dimensional data, which may include, for example, a COIL20 data set or a COIL100 data set or an ORL face data set or a YALE face data set, taking the high-dimensional image data as an example. When the object to be identified is an image data set, different features of the image data are extracted for each data, and the extracted features include at least one of gray-scale values, two-dimensional histograms, scale invariant feature transform values (SIFT), Histogram of Oriented Gradients (HOG), and the like.
In operation S302, the features of each object in different dimensions are combined into a feature vector of each object.
According to the embodiment of the disclosure, for an object such as text, image, voice, video and the like with multidimensional features, a plurality of features extracted from each object are combined respectively to form a feature vector of the object, namely, text, image, voice, video data are converted into feature vector data.
Feature extraction is performed in the same manner for each object, and feature combinations are performed in the same manner to form feature vectors expressed in the same form.
For example, when the object to be identified is an image data set, the gray value, the two-dimensional histogram, the SIFT value and the HOG corresponding to one image data are extracted, and the gray value, the two-dimensional histogram, the SIFT value and the HOG value are serially spliced to form a feature vector representing the image. In the feature vector, each feature can be converted into at least one numerical value, and the feature vector can be a column vector. The same feature of each object is converted in the same manner, so that the final column vector corresponding to each object has the same number of rows, and the same number of rows of each column vector represents the same feature.
In operation S303, the feature vector of each object is selected by using the feature selection matrix, so as to obtain a feature vector of each object including at least one feature.
The method for selecting the features provided by the embodiment of the disclosure converts data in the forms of text, image, voice, video and the like into the feature vectors, and the obtained feature vectors can more accurately obtain the features representing the object, so that the clustering and recognition of the object can be better performed in the future, and the accuracy of information recognition is improved.
FIG. 4 schematically illustrates a flow diagram of selecting features according to another embodiment of the disclosure.
As shown in fig. 4, operation S202 may further include operations S401 to S402.
In operation S401, a row sparse matrix, which is a feature selection matrix, is constructed.
In operation S402, a plurality of features of each object are respectively selected through a row sparse matrix, resulting in a plurality of objects including at least one feature.
According to the embodiment of the disclosure, the row sparse matrix is multiplied by the eigenvector representing each object characteristic, so that the sparsification of the eigenvector can be realized, and the number of the characteristics with discriminability actually participating in the clustering can be controlled. The motivation for feature selection is to select at least one feature from a plurality of original features of high dimensional data. On one hand, the dimension reduction is carried out on the original high-dimensional data, and the operation amount and the operation cost are reduced. On the other hand, the features that are usually selected are discriminative, and the relevance between the objects can be found by the discriminative features.
Fig. 5 schematically shows a flow chart of a clustering method according to an embodiment of the present disclosure.
As shown in fig. 5, in the case where it is determined that the clustering result does not satisfy the preset condition in operation S205, updating the feature selection matrix according to the clustering result may include operations S501 to S502.
In operation S501, in the case where it is determined that the clustering accuracy included in the clustering result belongs to the first preset range, the feature selection matrix is updated to select a different kind of features.
In operation S502, in the case where it is determined that the clustering accuracy included in the clustering result belongs to the second preset range, the feature selection matrix is updated to select a different number of features.
According to an embodiment of the present disclosure, in particular, clustering may be performed by the following target formula:
Figure RE-GDA0003221335780000111
Figure RE-GDA0003221335780000112
wherein n is the number of objects in the object to be identified, c is the number of clustering centers, i is the number of the objects in the object to be identified, j is the number of the clustering centers, and xiTo show the feature vector corresponding to the ith object, mjIs the jth clusterVector of class center, yijFor the confidence that the ith object belongs to the jth cluster, W is the feature selection matrix, i.e., the row sparse matrix, WTSelecting an orthogonal matrix of matrices for the feature, | W |)2,1Is l of the row sparse matrix W2，1Norm, lambda is a regularization parameter, lambda is more than or equal to 0 and less than or equal to 1, I is an identity matrix, and s.t. represents constraint.
It should be noted that the capital bold letters are used in this disclosure to denote the matrix, such as W, I; vectors are represented using lower case bold italics, such as x, and scalars are represented using lower case italics, such as y.
Looking at the above target formula, the first term, as a whole
Figure RE-GDA0003221335780000113
Representing the similarity of the feature vector corresponding to the projected data and the center vector of different clusters for loss terms, wherein WTxiThe vector after feature selection is carried out on the feature vector is shown, the orthogonality of the feature selection matrix is increased, and the situation that the vector falls into trivial solution m in the subsequent solving process can be avoidedjRepresenting the cluster centers of the feature-selected vectors. The second term λ | W |2，1And the number of the discriminative features is used for dynamically controlling the actual participation in clustering, lambda is used for controlling the sparsity of W, and the matrix W is sparser when the lambda value is larger.
According to an embodiment of the present disclosure, the preset condition may be an iteration stop condition. And when the clustering result meets the iteration stop condition, obtaining the clustering result, and executing the operation of identifying the information included by the objects according to the clustering result. And when the obtained clustering result does not meet the iteration stop condition, updating the feature selection matrix, reusing the updated feature selection matrix to perform feature selection on the object to obtain a new feature vector after feature selection, clustering the new feature vector, and repeating the operation until the iteration stop condition is met. In general, the feature vector after the newly obtained feature selection is different from the feature vector after the feature selection is obtained last time. Those skilled in the art will understand that the iteration stop condition may include the number of iterations, and when the number of iterations reaches a preset number, the iteration is stopped to obtain the clustering result. The iteration stop condition may also be a numerical range of satisfaction of the clustering accuracy. Of course, the disclosed embodiments are not so limited.
For example, assuming that a plurality of objects are acquired having 5 features A, B, C, D, E, the 5 features are selected using a feature selection matrix. Specifically, in operation S502, if the clustering accuracy does not meet the preset condition, the feature selection matrix is updated, and the operation returns to the previous step.
For example, if the value of the clustering accuracy obtained by clustering based on the features a and B is 0.3 or less (the first preset range), which indicates that the accuracy of clustering based on the features a and B is low, and therefore the selected feature needs to be changed, the feature selection matrix is updated. The updated feature selection feature may replace one or all of the original features, for example, the updated feature selection matrix may select two features a and C, or select two features C and D, perform selection on a plurality of features of the object by using the updated feature selection matrix, obtain a plurality of feature vectors including two features a and C, or include two feature vectors C and D, and perform re-clustering.
If the value of the clustering accuracy obtained by clustering the two eigenvectors including C and D is greater than 0.3 and less than or equal to 0.5 (second preset range), which indicates that the clustering result generated by clustering based on the eigenvectors C and D has a certain accuracy, and the similarity between the eigenvectors can be further mined, the eigenvector selection matrix is updated, a new eigenvector can be added to the updated eigenvector selection matrix, for example, the updated eigenvector selection matrix can select A, C and D, the selection of the plurality of characteristics of the object by using the updated eigenvector selection matrix is performed, a plurality of eigenvectors including A, C and D are obtained, and clustering is performed again.
If the value of the clustering accuracy obtained by clustering the three features including A, C and D is greater than 0.5 (the third preset range satisfying the preset condition), which indicates that the generated clustering result obtained by clustering based on the three features A, C and D satisfies the requirement, an operation of identifying information included in the plurality of objects from the clustering result may be performed.
It should be noted that the number of features, the types of features, the preset conditions, the range for dividing the clustering accuracy, and the manner of selecting the features, which are listed in the above example, are all exemplarily described, and do not limit the application scenario of the technical solution provided by the present disclosure, nor limit the specific embodiment of the technical solution provided by the present disclosure.
By the embodiment of the disclosure, the integration of the clustering process and the feature selection process is realized by continuously and dynamically adjusting the number and the types of the features in the clustering process until the clustering result meets the iteration stop condition. The orthogonal matrix of the sparse matrix is adopted for feature selection, so that a trivial solution can be prevented from being trapped in a clustering solving process, the clustering accuracy is improved, further, the object information can be identified according to a more accurate result, and the information identification accuracy is improved. In addition, the number of distinguishing features actually participating in clustering is dynamically selected through the row sparse matrix, clustering and feature selection can be better fused, flexible clustering is realized, the accuracy and efficiency of clustering are improved, and the accuracy and efficiency of object information identification are further improved.
In order to further improve the accuracy of the object identification method provided by the embodiment of the present disclosure, the embodiment of the present disclosure further provides a weighted iteration solving method of the above target formula. The method comprises the steps of clustering a plurality of objects comprising at least one feature to obtain a clustering result, wherein the clustering result comprises numerical values of any two variables of a fixed clustering center, a feature selection matrix and confidence, and calculating the numerical value of another variable to obtain a group of numerical values comprising the three variables. And calculating a plurality of groups of numerical values comprising the three variables in an iterative mode to obtain a clustering center, a feature selection matrix and a convergence value of confidence coefficient.
Fig. 6 schematically shows a flowchart of a weighted iterative solution method according to an embodiment of the present disclosure.
As shown in fig. 6, the weighted iterative solution method may include operations S601 to S603, for example.
In operation S601, W and m are fixedjAnd updating Y to solve the target formula. The matrix Y is a confidence matrix and is setThe confidence matrix Y is the confidence coefficient Y of the ith object belonging to different clustersijA matrix is formed.
Based on operation S601, W and m are fixedjWithout change, the above target formula can be simplified as:
Figure RE-GDA0003221335780000141
Figure RE-GDA0003221335780000142
through the limitation, the problem of solving the target formula can be converted into a linear optimization problem with constraint, and the existing optimization tool box can be directly used for solving, for example, the minimum value is solved by using the Lagrange algorithm.
In operation S602, W and Y are fixed, and m is updatedjAnd solving the target formula.
Based on operation S602, fixing W and Y unchanged, the above target formula can be simplified as:
Figure RE-GDA0003221335780000143
s.t.WTW＝I
for variable m in simplified formulajTaking the derivative and making the derivative zero to obtain mjIs solved into
Figure RE-GDA0003221335780000144
In operation S603, m is fixedjAnd Y, updating W to solve the target formula.
Based on operation S603, m is fixedjAnd Y is unchanged, the target formula can be simplified as follows:
Figure RE-GDA0003221335780000145
s.t.WTW＝I
order to
Figure RE-GDA0003221335780000146
Wherein the content of the first and second substances,
Figure RE-GDA0003221335780000147
mean vector, S, representing the original data of the jth clusterwRepresenting compactness of the cluster for the divergence matrix in the cluster, the method can be further simplified as follows:
Figure RE-GDA0003221335780000151
s.t.WTW＝I
where D is a diagonal matrix with the value of the ith element on the diagonal of the matrix being
Figure RE-GDA0003221335780000152
Epsilon is an arbitrarily small constant. Based on the final simplified formula, Y can be derived directly.
The iterative solution is performed through the above operations S601 to S603 until W, mjAnd Y is converged approximately to obtain a final clustering result. It should be noted that the execution of the operations S601 to S603 does not necessarily follow the above sequence, and the logic sequence of the execution may be selected according to the actual situation.
It should be noted that, before object clustering, the objects to be clustered are projected as data matrices, and the feature selection matrix can be constructed as
Figure RE-GDA0003221335780000153
And representing, wherein d represents the dimension of the data matrix, namely the number of the objects, and e is the number of the features included in the feature vector corresponding to the object. II WTxi-mj‖2Is WTxi-mjL of2Norm, | W |2，1Is l of the row sparse matrix W2.1And (4) norm.
The following is to the present disclosureExamples of2Norm sum l2.1The norm is explained in detail.
For matrix
Figure RE-GDA0003221335780000154
qab，qaRespectively representing the vector formed by the elements at the positions of the matrixes (a, b) and the elements in the a-th column. For vectors
Figure RE-GDA0003221335780000155
It lpNorm is defined as
Figure RE-GDA0003221335780000156
Figure RE-GDA0003221335780000157
E.g. of vectors0Norm is expressed as
Figure RE-GDA0003221335780000158
It l1Norm is expressed as
Figure RE-GDA0003221335780000159
Where | represents an absolute value. l2Norm is expressed as
Figure RE-GDA00032213357800001510
Figure RE-GDA00032213357800001511
Extending from the vector norm to the matrix norm, the F norm of matrix Q is defined as:
Figure RE-GDA00032213357800001512
l of the matrix Q2.0The norm is defined as:
Figure RE-GDA00032213357800001513
l of the matrix Q2.1The norm is defined as:
Figure RE-GDA00032213357800001514
according to the above definition, the above objective formula can be solved.
Compared with the clustering operation in which the feature selection and the clustering are unrelated in the prior art, the clustering target formula provided by the embodiment of the invention can well fuse the clustering and the feature selection, increase the orthogonality of the feature selection matrix to avoid the situation that the feature selection matrix falls into trivial solution in the subsequent solving process, and can better fuse the clustering and the feature selection by dynamically selecting the number of distinguishing features actually participating in the clustering through the row sparse matrix and the regularization parameters, thereby realizing flexible clustering, improving the accuracy and the efficiency of the clustering and further improving the accuracy and the efficiency of object information identification.
According to the weighted iterative solution method provided by the embodiment of the disclosure, the clustered target formula is solved in a weighted iterative solution mode until W, mjAnd Y approaches convergence, which can further improve the accuracy of the object identification method.
In order to more fully highlight the effectiveness and advantages of the object identification method provided by the above embodiment of the present disclosure, the data clustering method provided by the embodiment of the present disclosure is adopted to perform identification experiments on 4 standard data sets, i.e., COIL20, COIL100, face data sets ORL and YALE, where information of an object is shown in table 1:
TABLE 1
Object Number of samples Number of features Categories
COIL20 1440 1024 20
COIL100 7200 1024 100
ORL 400 1024 40
YALE 165 1024 15
And selecting three methods in the prior art to perform clustering comparison with the clustering method provided by the embodiment of the disclosure. The method comprises the following steps: (Baseline): performing k-means clustering using all the features; the second method (Max-var) is that when the variance in a certain direction in the data is larger, the expression capability of the feature is stronger, so that the corresponding feature is selected according to the variance of the data to perform k-means clustering; method three (Laplacian Score): and (3) performing feature selection by using a classical Laplacian Score feature selection algorithm, and inputting the selected features into k-menas for clustering. The clustering accuracy was used as the evaluation result of this experiment. The clustering accuracy is obtained by performing accuracy evaluation on the final clustering center, the feature selection matrix and the confidence coefficient value solved by the clustering algorithm. The clustering results are shown in table 2 (clustering result of each method with a feature number of 50) and table 3 (clustering result of each method with a feature number of 100):
TABLE 2
Data set Baseline Max-var Laplacian Score The method of the disclosure
COIL20 0.52 0.43 0.49 0.54
COIL100 0.47 0.03 0.29 0.49
ORL 0.46 0.37 0.38 0.53
YALE 0.37 0.34 0.38 0.45
TABLE 3
Data set Baseline Max-var Laplacian Score The method of the disclosure
COIL20 0.52 0.49 0.51 0.55
COIL100 0.47 0.03 0.31 0.49
ORL 0.46 0.40 0.44 0.54
YALE 0.37 0.35 0.38 0.46
As is apparent from the result data provided in tables 2 and 3, the data clustering method provided in the embodiment of the present disclosure has the highest clustering accuracy compared with the three methods in the prior art, and further improves the accuracy of object information identification.
Fig. 7 schematically shows a block diagram of an object recognition apparatus according to an embodiment of the present disclosure.
As shown in fig. 7, the object recognition apparatus 700 may include, for example, an obtaining module 710, a selecting module 720, a clustering module 730, a first determining module 740, and a second determining module 750.
The obtaining module 710 is configured to obtain a plurality of objects to be identified, where each object includes a plurality of features.
A selecting module 720, configured to select, by using the feature selection matrix, a plurality of features of each object respectively, so as to obtain a plurality of objects including at least one feature.
The clustering module 730 is configured to cluster a plurality of objects including at least one feature to obtain a clustering result.
The first determining module 740 is configured to, when it is determined that the clustering result satisfies the preset condition, identify information included in the plurality of objects according to the clustering result; and
and a second determining module 750, configured to update the feature selection matrix according to the clustering result when it is determined that the clustering result does not meet the preset condition, and return to an operation of selecting the plurality of features of each object respectively by using the feature selection matrix.
The object recognition apparatus 700 shown in fig. 7 will be further described with reference to the accompanying drawings.
FIG. 8 schematically shows a block diagram of a selection module according to one embodiment of the present disclosure.
According to an embodiment of the present disclosure, as shown in fig. 8, the selecting module 720 includes:
an extracting unit 721 is used for extracting features of different dimensions of each object respectively.
And a composing unit 722, configured to compose features of different dimensions of each object into a feature vector of each object.
The first selecting unit 723 is configured to select the feature vector of each object by using the feature selection matrix, so as to obtain a feature vector of each object including at least one feature.
FIG. 9 schematically shows a block diagram of a selection module according to yet another embodiment of the present disclosure.
According to an embodiment of the present disclosure, as shown in fig. 9, the selecting module 720 further includes:
a constructing unit 724 configured to construct a row sparse matrix;
the second selecting unit 725 is configured to select a plurality of features of each object through the row sparse matrix, so as to obtain a plurality of objects including at least one feature.
Any number of modules, sub-modules, units, sub-units, or at least part of the functionality of any number thereof according to embodiments of the present disclosure may be implemented in one module. Any one or more of the modules, sub-modules, units, and sub-units according to the embodiments of the present disclosure may be implemented by being split into a plurality of modules. Any one or more of the modules, sub-modules, units, sub-units according to embodiments of the present disclosure may be implemented at least in part as a hardware circuit, such as a Field Programmable Gate Array (FPGA), a Programmable Logic Array (PLA), a system on a chip, a system on a substrate, a system on a package, an Application Specific Integrated Circuit (ASIC), or may be implemented in any other reasonable manner of hardware or firmware by integrating or packaging a circuit, or in any one of or a suitable combination of software, hardware, and firmware implementations. Alternatively, one or more of the modules, sub-modules, units, sub-units according to embodiments of the disclosure may be at least partially implemented as a computer program module, which when executed may perform the corresponding functions.
For example, any plurality of the obtaining module 710, the selecting module 720, the clustering module 730, the first determining module 740, and the second determining module 750 may be combined in one module/unit/sub-unit to be implemented, or any one of the modules/units/sub-units may be split into a plurality of modules/units/sub-units. Alternatively, at least part of the functionality of one or more of these modules/units/sub-units may be combined with at least part of the functionality of other modules/units/sub-units and implemented in one module/unit/sub-unit. According to an embodiment of the present disclosure, at least one of the obtaining module 710, the selecting module 720, the clustering module 730, the first determining module 740, and the second determining module 750 may be implemented at least partially as a hardware circuit, such as a Field Programmable Gate Array (FPGA), a Programmable Logic Array (PLA), a system on a chip, a system on a substrate, a system on a package, an Application Specific Integrated Circuit (ASIC), or may be implemented by hardware or firmware in any other reasonable manner of integrating or packaging a circuit, or implemented by any one of three implementations of software, hardware, and firmware, or an appropriate combination of any several of them. Alternatively, at least one of the obtaining module 710, the selecting module 720, the clustering module 730, the first determining module 740, and the second determining module 750 may be at least partially implemented as a computer program module, which when executed, may perform a corresponding function.
It should be noted that the object recognition device portion in the embodiment of the present disclosure corresponds to the object recognition method portion in the embodiment of the present disclosure, and the specific implementation details and the technical effects thereof are also the same, and are not described herein again.
Fig. 10 schematically shows a block diagram of an electronic device adapted to implement the above described method according to an embodiment of the present disclosure. The electronic device shown in fig. 10 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiments of the present disclosure.
As shown in fig. 10, an electronic device 800 according to an embodiment of the present disclosure includes a processor 801 that can perform various appropriate actions and processes according to a program stored in a Read Only Memory (ROM)802 or a program loaded from a storage section 808 into a Random Access Memory (RAM) 803. The processor 801 may include, for example, a general purpose microprocessor (e.g., a CPU), an instruction set processor and/or associated chipset, and/or a special purpose microprocessor (e.g., an Application Specific Integrated Circuit (ASIC)), among others. The processor 801 may also include onboard memory for caching purposes. The processor 801 may include a single processing unit or multiple processing units for performing different actions of the method flows according to embodiments of the present disclosure.
In the RAM803, various programs and data necessary for the operation of the electronic apparatus 800 are stored. The processor 801, the ROM 802, and the RAM803 are connected to each other by a bus 804. The processor 801 performs various operations of the method flows according to the embodiments of the present disclosure by executing programs in the ROM 802 and/or RAM 803. Note that the programs may also be stored in one or more memories other than the ROM 802 and RAM 803. The processor 801 may also perform various operations of method flows according to embodiments of the present disclosure by executing programs stored in the one or more memories.
Electronic device 800 may also include input/output (I/O) interface 805, input/output (I/O) interface 805 also connected to bus 804, according to an embodiment of the present disclosure. Electronic device 800 may also include one or more of the following components connected to I/O interface 805: an input portion 806 including a keyboard, a mouse, and the like; an output section 807 including a signal such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and the like, and a speaker; a storage portion 808 including a hard disk and the like; and a communication section 809 including a network interface card such as a LAN card, a modem, or the like. The communication section 809 performs communication processing via a network such as the internet. A drive 810 is also connected to the I/O interface 805 as necessary. A removable medium 811 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is mounted on the drive 810 as necessary, so that a computer program read out therefrom is mounted on the storage section 808 as necessary.
According to embodiments of the present disclosure, method flows according to embodiments of the present disclosure may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program carried on a computer readable storage medium, the computer program comprising program code for performing the method illustrated by the flow chart. In such an embodiment, the computer program can be downloaded and installed from a network through the communication section 809 and/or installed from the removable medium 811. The computer program, when executed by the processor 801, performs the above-described functions defined in the system of the embodiments of the present disclosure. The systems, devices, apparatuses, modules, units, etc. described above may be implemented by computer program modules according to embodiments of the present disclosure.
The present disclosure also provides a computer-readable storage medium, which may be included in the apparatus/device/system described in the above embodiments; or may exist separately and not be assembled into the device/apparatus/system. The computer-readable storage medium carries one or more programs which, when executed, implement the method according to an embodiment of the disclosure.
According to an embodiment of the present disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium. Examples may include, but are not limited to: a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the present disclosure, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.
For example, according to embodiments of the present disclosure, a computer-readable storage medium may include the ROM 802 and/or RAM803 described above and/or one or more memories other than the ROM 802 and RAM 803.
The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams or flowchart illustration, and combinations of blocks in the block diagrams or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
Those skilled in the art will appreciate that various combinations and/or combinations of features recited in the various embodiments and/or claims of the present disclosure can be made, even if such combinations or combinations are not expressly recited in the present disclosure. In particular, various combinations and/or combinations of the features recited in the various embodiments and/or claims of the present disclosure may be made without departing from the spirit or teaching of the present disclosure. All such combinations and/or associations are within the scope of the present disclosure.
The embodiments of the present disclosure have been described above. However, these examples are for illustrative purposes only and are not intended to limit the scope of the present disclosure. Although the embodiments are described separately above, this does not mean that the measures in the embodiments cannot be used in advantageous combination. The scope of the disclosure is defined by the appended claims and equivalents thereof. Various alternatives and modifications can be devised by those skilled in the art without departing from the scope of the present disclosure, and such alternatives and modifications are intended to be within the scope of the present disclosure.

Claims (17)

1. An object recognition method, comprising:
acquiring a plurality of objects to be identified, each of the objects comprising a plurality of features;
selecting the plurality of characteristics of each object by adopting a characteristic selection matrix to obtain a plurality of objects comprising at least one characteristic;
clustering the plurality of objects comprising the at least one characteristic to obtain a clustering result;
under the condition that the clustering result meets the preset condition, identifying information included by the objects according to the clustering result; and
and under the condition that the clustering result is determined not to meet the preset condition, updating the feature selection matrix according to the clustering result, and returning to the operation of respectively selecting the plurality of features of each object by adopting the feature selection matrix.
2. The method of claim 1, wherein the selecting the plurality of features of each object using a feature selection matrix to obtain a plurality of objects including at least one feature comprises:
respectively extracting the characteristics of different dimensions of each object;
forming the feature of each object with different dimensions into a feature vector of each object;
and respectively selecting the feature vector of each object by adopting a feature selection matrix to obtain the feature vector of each object comprising at least one feature.
3. The method according to claim 1 or 2, wherein the selecting the plurality of features of each object using a feature selection matrix to obtain a plurality of objects including at least one feature comprises:
constructing a row sparse matrix;
and respectively selecting the plurality of characteristics of each object through the row sparse matrix to obtain a plurality of objects comprising at least one characteristic.
4. The method according to claim 1, wherein the updating the feature selection matrix according to the clustering result in the case that the clustering result is determined not to satisfy the preset condition comprises:
under the condition that the clustering accuracy included in the clustering result is determined to belong to a first preset range, updating the feature selection matrix to select different kinds of features;
and updating the feature selection matrix to select different numbers of features under the condition that the clustering accuracy included in the clustering result is determined to belong to a second preset range.
5. The method of claim 1, wherein the clustering the plurality of objects including the at least one feature to obtain a clustering result comprises:
fixing the numerical values of any two variables in the clustering center, the feature selection matrix and the confidence coefficient, and calculating the numerical value of the other variable to obtain a group of numerical values comprising the three variables;
and calculating a plurality of groups of numerical values comprising the three variables in an iterative mode to obtain a clustering center, the feature selection matrix and a convergence value of the confidence coefficient.
6. The method of claim 2, wherein the object comprises at least one of text, image, voice, and video.
7. The method of claim 6, wherein the image comprises a high-dimensional image dataset.
8. The method of claim 7, wherein the high-dimensional image dataset comprises a COIL20 dataset or a COIL100 dataset or an ORL face dataset or a yal face dataset.
9. The method of claim 6, wherein the object comprises an image; the respectively extracting the features of different dimensions of each object comprises the following steps:
extracting at least one of the following features of the image:
gray value;
a two-dimensional histogram;
a scale invariant feature transform value;
histogram of directional gradients.
10. An object recognition apparatus comprising:
an acquisition module for acquiring a plurality of objects to be identified, each of the objects comprising a plurality of features;
a selection module, configured to select the multiple features of each object by using a feature selection matrix, respectively, to obtain multiple objects including at least one feature;
the clustering module is used for clustering the plurality of objects comprising the at least one characteristic to obtain a clustering result;
the first determining module is used for identifying information included by the objects according to the clustering result under the condition that the clustering result is determined to meet the preset condition; and
and the second determining module is used for updating the feature selection matrix according to the clustering result and returning the operation of respectively selecting the plurality of features of each object by adopting the feature selection matrix under the condition that the clustering result is determined not to meet the preset condition.
11. The apparatus of claim 10, wherein the selection module comprises:
the extraction unit is used for respectively extracting the characteristics of different dimensions of each object;
the composition unit is used for composing the characteristics of each object with different dimensions into a characteristic vector of each object;
the first selection unit is used for selecting the feature vector of each object by adopting the feature selection matrix to obtain the feature vector of each object including at least one feature.
12. The apparatus of claim 10 or 11, wherein the selection module further comprises:
the construction unit is used for constructing a row sparse matrix;
and the second selection unit is used for respectively selecting the plurality of characteristics of each object through the row sparse matrix to obtain a plurality of objects comprising at least one characteristic.
13. The apparatus of claim 10, wherein the second determining means comprises:
a first determining unit, configured to update the feature selection matrix to select different kinds of features when it is determined that the clustering accuracy included in the clustering result belongs to a first preset range;
and the second determining unit is used for updating the feature selection matrix to select different numbers of features under the condition that the clustering accuracy included in the clustering result is determined to belong to a second preset range.
14. The apparatus of claim 10, wherein the clustering module comprises:
the first calculation unit is used for fixing numerical values of any two variables in the clustering center, the feature selection matrix and the confidence coefficient, and calculating the numerical value of the other variable to obtain a group of numerical values comprising the three variables;
and the second calculation unit is used for calculating a plurality of groups of numerical values comprising the three variables in an iterative mode to obtain a clustering center, the feature selection matrix and a convergence value of the confidence coefficient.
15. An electronic device, comprising:
one or more processors;
a memory for storing one or more programs,
wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the method of any of claims 1-9.
16. A computer readable storage medium having stored thereon executable instructions which, when executed by a processor, cause the processor to carry out the method of any one of claims 1 to 9.
17. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1 to 9.
CN202110829878.6A 2021-02-04 2021-07-22 Object identification method, device, equipment and medium Pending CN113408665A (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202110157595 2021-02-04
CN2021101575951 2021-02-04

Publications (1)

Publication Number Publication Date
CN113408665A true CN113408665A (en) 2021-09-17

Family

ID=77687250

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110829878.6A Pending CN113408665A (en) 2021-02-04 2021-07-22 Object identification method, device, equipment and medium

Country Status (1)

Country Link
CN (1) CN113408665A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117076962A (en) * 2023-10-13 2023-11-17 腾讯科技（深圳）有限公司 Data analysis method, device and equipment applied to artificial intelligence field

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117076962A (en) * 2023-10-13 2023-11-17 腾讯科技（深圳）有限公司 Data analysis method, device and equipment applied to artificial intelligence field
CN117076962B (en) * 2023-10-13 2024-01-26 腾讯科技（深圳）有限公司 Data analysis method, device and equipment applied to artificial intelligence field

Similar Documents

Publication Publication Date Title
US11526799B2 (en) Identification and application of hyperparameters for machine learning
CN112559784B (en) Image classification method and system based on incremental learning
Erisoglu et al. A new algorithm for initial cluster centers in k-means algorithm
Chernozhukov et al. hdm: High-dimensional metrics
Ocepek et al. Improving matrix factorization recommendations for examples in cold start
WO2022063151A1 (en) Method and system for relation learning by multi-hop attention graph neural network
US9536201B2 (en) Identifying associations in data and performing data analysis using a normalized highest mutual information score
CN109522435B (en) Image retrieval method and device
Doersch et al. Context as supervisory signal: Discovering objects with predictable context
WO2020253357A1 (en) Data product recommendation method and apparatus, computer device and storage medium
CN108764726B (en) Method and device for making decision on request according to rules
US20190279295A1 (en) Image analysis and identification using machine learning with output personalization
CN113377964B (en) Knowledge graph link prediction method, device, equipment and storage medium
CN113657087B (en) Information matching method and device
Yu et al. A content-based goods image recommendation system
CN112529665A (en) Product recommendation method and device based on combined model and computer equipment
CN114492601A (en) Resource classification model training method and device, electronic equipment and storage medium
CN113408665A (en) Object identification method, device, equipment and medium
CN114692889A (en) Meta-feature training model for machine learning algorithm
CN112818068A (en) Big data and multidimensional feature-based data tracing method and system
Aaron et al. Multivariate and functional robust fusion methods for structured Big Data
Dessureault et al. Explainable global error weighted on feature importance: The xGEWFI metric to evaluate the error of data imputation and data augmentation
Rad et al. A multi-view-group non-negative matrix factorization approach for automatic image annotation
Mall et al. Agglomerative hierarchical kernel spectral data clustering
US20230297651A1 (en) Cost equalization spectral clustering

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination