CN114021655A - Object class evaluation method and device - Google Patents

Object class evaluation method and device Download PDF

Info

Publication number
CN114021655A
CN114021655A CN202111308616.1A CN202111308616A CN114021655A CN 114021655 A CN114021655 A CN 114021655A CN 202111308616 A CN202111308616 A CN 202111308616A CN 114021655 A CN114021655 A CN 114021655A
Authority
CN
China
Prior art keywords
node
vector
entities
centrality
feature vector
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
CN202111308616.1A
Other languages
Chinese (zh)
Inventor
郑志骏
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.)
Hengrui Chongqing Artificial Intelligence Technology Research Institute Co ltd
Original Assignee
Hengrui Chongqing Artificial Intelligence Technology Research Institute 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 Hengrui Chongqing Artificial Intelligence Technology Research Institute Co ltd filed Critical Hengrui Chongqing Artificial Intelligence Technology Research Institute Co ltd
Priority to CN202111308616.1A priority Critical patent/CN114021655A/en
Publication of CN114021655A publication Critical patent/CN114021655A/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
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
    • G06F16/367—Ontology
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/02—Knowledge representation; Symbolic representation
    • G06N5/027—Frames

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Computational Linguistics (AREA)
  • Software Systems (AREA)
  • Mathematical Physics (AREA)
  • Computing Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Animal Behavior & Ethology (AREA)
  • Databases & Information Systems (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The invention relates to the field of artificial intelligence, and particularly provides an object class evaluation method and device, aiming at solving the problem that the existing method is low in neural network accuracy and is obtained based on small-scale direct training of a regular sample. For this purpose, the object category assessment method of the invention comprises the steps of obtaining a training sample and a sample to be analyzed; generating a decision stream based on the training samples; training and generating an confrontation network based on the decision flow to obtain a trained and generated confrontation network; and evaluating the sample to be analyzed based on the trained generated confrontation network to obtain the object class.

Description

Object class evaluation method and device
Technical Field
The invention relates to the field of artificial intelligence, and particularly provides an object class evaluation method and device.
Background
At present, the application of artificial intelligence in the public security field is more and more extensive, and most notably, an analysis system based on a machine learning algorithm is used as a main part. The analysis system mainly based on the machine learning algorithm mainly trains the neural network by using small-scale positive samples, and then analyzes the text by using the trained neural network to obtain an analysis result. However, the neural network obtained by direct training based on small-scale positive samples has low accuracy, and the evaluation of object categories with abnormal activity rules, such as virus-involved people, fleeing people, stolen people, and the like, can only be realized by experts, so that the actual requirements are difficult to meet.
For this reason, there is a need in the art for a new solution to the above problems.
Disclosure of Invention
In order to overcome the above-mentioned drawbacks, the present invention is proposed to provide a solution or at least a partial solution to the technical problem of low accuracy of neural networks obtained by the existing methods based on direct training of small-scale positive examples. The invention provides an object category evaluation method and device.
In a first aspect, the present invention provides an object class evaluation method, including: acquiring a training sample and a sample to be analyzed; generating a decision stream based on the training samples; training and generating an confrontation network based on the decision flow to obtain a trained and generated confrontation network; and evaluating the sample to be analyzed based on the trained generated confrontation network to obtain the object class.
In some embodiments, generating a decision stream based on the training samples further comprises: extracting entities in the training sample and relations among the entities; establishing a knowledge base based on the entities and the relationship between the entities; generating a decision flow based on the established knowledge base.
In some embodiments, the knowledge base includes embedded feature vectors and decision flow sub-models; building a knowledge base based on the entities and relationships between entities further comprises: obtaining a subject database corresponding to the entity based on the entity; obtaining a knowledge graph based on the entities and relationships between the entities; extracting and storing embedded characteristic vectors based on the knowledge graph; and obtaining and storing a decision flow sub-model based on the embedded feature vector and a thematic library corresponding to the entity.
In some embodiments, extracting embedded feature vectors based on the knowledge-graph further comprises: acquiring a community structure in the knowledge graph based on a community discovery algorithm; evaluating each node in the community structure to obtain the feature vector centrality and the point centrality corresponding to each node; and obtaining an embedded feature vector based on the feature vector centrality and the point centrality corresponding to each node.
In some embodiments, obtaining the embedded feature vector based on the feature vector centrality and the point centrality corresponding to each node further comprises: acquiring the weight of each node and the weight of each edge in the community structure based on the feature vector centrality and the point centrality corresponding to each node; constructing an adjacency matrix of the community structure based on the weight of each node and the weight of each edge; decomposing the adjacency matrix to obtain the vector representation of each node in the community structure; carrying out vector expansion on the vector representation of each node to obtain an expanded vector; and carrying out attribute reduction on the expansion vector to obtain an embedded feature vector.
In a second aspect, the present invention provides an object category evaluation device including: an acquisition module configured to acquire a training sample and a sample to be analyzed; a generation module configured to generate a decision stream based on the training samples; a training module configured to train a generation countermeasure network based on the decision flow, resulting in a trained generation countermeasure network; and the evaluation module is configured to evaluate the sample to be analyzed based on the trained generated confrontation network to obtain the object class.
In some embodiments, the generating module further comprises: an entity and relationship extraction module configured to extract entities and relationships between entities in the training sample; a knowledge base establishing module configured to establish a knowledge base based on the entities and relationships between the entities; a decision stream generation module configured to generate a decision stream based on the established knowledge base.
In some embodiments, the knowledge base includes embedded feature vectors and decision flow sub-models; the knowledge base establishing module further comprises: the special subject library acquisition module is configured to acquire a special subject library corresponding to the entity based on the entity; a knowledge graph acquisition module configured to obtain a knowledge graph based on the entities and relationships between the entities; a vector extraction module configured to extract and save embedded feature vectors based on the knowledge-graph; and the decision flow sub-model acquisition module is configured to acquire and store a decision flow sub-model based on the embedded feature vector and the topic library corresponding to the entity.
In some embodiments, the vector extraction module further comprises: a community structure acquisition module configured to acquire a community structure in the knowledge graph based on a community discovery algorithm; the centrality obtaining module is configured to evaluate each node in the community structure to obtain a feature vector centrality and a centrality of degree corresponding to each node; and the embedded feature vector acquisition module is configured to acquire an embedded feature vector based on the feature vector centrality and the point centrality corresponding to each node.
In some embodiments, the embedded feature vector acquisition module further comprises: a weight obtaining module configured to obtain a weight of each node and a weight of each edge in the community structure based on the feature vector centrality and the point centrality corresponding to each node; a construction module configured to construct an adjacency matrix of the community structure based on the weight of each node and the weight of each edge; a decomposition module configured to decompose the adjacency matrix to obtain a vector representation of each node in the community structure; the vector expansion module is configured to perform vector expansion on the vector representation of each node to obtain an expanded vector; and the attribute reduction module is configured to perform attribute reduction on the extended vector to obtain an embedded feature vector.
In a third aspect, there is provided a control apparatus comprising a processor and a storage device adapted to store a plurality of program codes adapted to be loaded and run by the processor to perform the object class assessment method of any of the preceding claims.
In a fourth aspect, a computer readable storage medium is provided, having stored therein a plurality of program codes adapted to be loaded and run by a processor to perform the object class assessment method of any of the preceding claims.
One or more technical schemes of the invention at least have one or more of the following beneficial effects:
according to the method, the training samples are processed to generate the decision flow, the decision flow is used for training the neural network, the neural network with high accuracy is obtained, and the accuracy of the later-stage object category evaluation is improved.
The method comprises the steps of extracting entities in training samples and relations among the entities, further establishing a knowledge base, obtaining decision flow based on the entities in the knowledge base, the relations among the entities and a decision flow sub-model, further utilizing the decision flow to achieve training of the neural network, and compared with the existing method for directly utilizing small-scale regular samples to train the neural network, improving the accuracy of the neural network and facilitating later-stage object category evaluation.
The embedded feature vector is obtained through the feature vector centrality and the degree centrality corresponding to each node, the method is novel, and a foundation is provided for later-stage object category evaluation.
Drawings
The disclosure of the present invention will become more readily understood with reference to the accompanying drawings. As is readily understood by those skilled in the art: these drawings are for illustrative purposes only and are not intended to constitute a limitation on the scope of the present invention. Moreover, in the drawings, like numerals are used to indicate like parts, and in which:
FIG. 1 is a flow chart illustrating the main steps of an object class assessment method according to an embodiment of the present invention;
FIG. 2 is a diagram of a syntax tree according to one embodiment of the present invention;
fig. 3 is a main block diagram of an object category evaluation apparatus according to an embodiment of the present invention.
List of reference numerals：
11: an acquisition module; 12: a generation module; 13: a training module; 14: and an evaluation module.
Detailed Description
Some embodiments of the invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only for explaining the technical principle of the present invention, and are not intended to limit the scope of the present invention.
In the description of the present invention, a "module" or "processor" may include hardware, software, or a combination of both. A module may comprise hardware circuitry, various suitable sensors, communication ports, memory, may comprise software components such as program code, or may be a combination of software and hardware. The processor may be a central processing unit, microprocessor, image processor, digital signal processor, or any other suitable processor. The processor has data and/or signal processing functionality. The processor may be implemented in software, hardware, or a combination thereof. Non-transitory computer readable storage media include any suitable medium that can store program code, such as magnetic disks, hard disks, optical disks, flash memory, read-only memory, random-access memory, and the like. The term "a and/or B" denotes all possible combinations of a and B, such as a alone, B alone or a and B. The term "at least one A or B" or "at least one of A and B" means similar to "A and/or B" and may include only A, only B, or both A and B. The singular forms "a", "an" and "the" may include the plural forms as well.
Some terms to which the present invention relates are explained first.
And (3) generating a countermeasure network: the generation of the countermeasure network is a generation model, and the main structure includes a generator G (Generator) and a discriminator D (discriminator), which uses the sample distribution to fit the real sample distribution to achieve the goal of falseness and falseness.
Feature vector centrality: the value of a node is measured according to the importance of its neighbors.
Dot degree centrality: how many lines are connected to a certain node, namely the sum of the in-degree and the out-degree of the node.
At present, the traditional analysis system mainly based on a machine learning algorithm is mainly based on a small-scale positive example direct training to obtain a neural network with lower accuracy, and the evaluation on object categories with abnormal activity rules, such as virus-involved persons, fleeing persons, stolen persons and the like, can only be realized by experts, so that the actual requirements are difficult to meet. Therefore, the application provides an object class evaluation method and device, a decision flow is generated by processing a training sample, and then the decision flow is used for training a neural network, so that the neural network with higher accuracy is obtained, and the accuracy of later-stage object class evaluation is further improved.
Referring to fig. 1, fig. 1 is a flow chart illustrating the main steps of an object classification evaluation method according to an embodiment of the present invention. As shown in fig. 1, the object class evaluation method in the embodiment of the present invention mainly includes the following steps S101 to S104.
Step S101: the method comprises the steps of obtaining a training sample and a sample to be analyzed, wherein the training sample comprises a training text, and the sample to be analyzed comprises a text to be analyzed. Specifically, the training text may be a historical file or historical data containing object categories, such as those involved in virus, fleeing people, stolen people, and the like. The text to be analyzed may be a series of texts, audio or video, etc., which need to be analyzed or processed.
Step S102: a decision stream is generated based on the training samples. Specifically, this step can be realized by steps S1021 to S1023 described below, which is not described herein again. Compared with the existing method for training the neural network by directly utilizing small-scale positive samples, the method has the advantages that the accuracy of the neural network is improved, and the accuracy of later-stage object class evaluation is favorably improved.
Step S1021: and extracting entities in the training sample and relations among the entities. In particular, OCR techniques and natural language processing techniques may be utilized in the present application to extract entities and relationships between entities in training samples. Further, an entity refers to a combination of data according to the dimension of a control object, including an identification id and an entity attribute, where the entity attribute may be age, gender, and the like, and the control object includes but is not limited to a person, an event, an organization, and the like. Relationships between entities may be represented by relationship ids and relationship attributes.
Step S1022: and establishing a knowledge base based on the entities and the relation between the entities, wherein the knowledge base comprises embedded feature vectors and decision flow sub-models.
In some embodiments, the step of establishing a knowledge base based on the entities and the relationships between the entities specifically includes: firstly, a subject database corresponding to an entity is obtained based on the entity, wherein the subject database can comprise a personnel call record subject database, a personnel fund transaction subject database, a personnel travel record subject database, a personnel daily activity subject database, a personnel virtual identity subject database and the like. On the basis of obtaining the subject database corresponding to the entity, the knowledge graph is further obtained based on the entity and the relationship between the entities. On the basis of acquiring the knowledge graph, an embedded feature vector can be further extracted by using a graph embedding algorithm and stored. And finally, obtaining and storing a decision flow sub-model based on the embedded feature vectors and the topic library corresponding to the entity.
Specifically, the graph embedding algorithm used in the present application is a self-developed graph embedding algorithm, but the graph embedding algorithm may also be a metapath2vec, random walk, or the like, but is not limited thereto. The step of extracting the embedded feature vector by using the self-research image embedding algorithm in the application specifically comprises the following steps: the community structure in the knowledge graph is first obtained by using a community discovery algorithm in a social network analysis algorithm, wherein the "community discovery algorithm" here may be a K-L algorithm, a GN algorithm, or the like, but is not limited thereto. And secondly, evaluating each node in the community structure to obtain the feature vector centrality and the point centrality corresponding to each node. The embedded feature vector can be further obtained on the basis of obtaining the feature vector centrality and the point centrality corresponding to each node, and the method specifically comprises the following steps.
Firstly, the weight of each node and the weight of each edge in a community structure are obtained according to the feature vector centrality and the point degree centrality corresponding to each node. Here, the weight of each node and the weight of each edge can be both 1/(1-Xe)-αβ) And calculating to obtain alpha as the feature vector centrality, beta as the point centrality and X as the scaling parameter.
Next, an adjacency matrix of the community structure is constructed based on the weight of each node and the weight of each edge, and in this embodiment, the weight of an edge composed of two nodes is equal to the product of the weight of the edge constituting the shortest path between the two nodes and the weight of the node to obtain the adjacency matrix.
After the adjacency matrix is obtained, the adjacency matrix is decomposed by utilizing a factorization method of the graph to obtain the vector representation y of each node in the community structureiThe factorization method of the above-mentioned figures includes, but is not limited to, laplacian graph embedding, cauchy graph embedding, etc.
Vector representation for each node yiAnd carrying out vector expansion to obtain an expanded vector. Specifically, each node is extended through its vector representation, its own label and entity attributes to form a new vector YiThus, an extended vector W ═ Y is obtained1,Y2,…,Yn]。
And carrying out attribute reduction on the obtained expansion vector to obtain an embedded feature vector. The attribute reduction is mainly the attribute reduction of a rough set, specifically, the first row of attributes of the extended vector is removed, when the remaining attributes are queried, the decision attributes of the same conditional attribute do not conflict, the removed first row does not generate inconsistent data, the first row of attributes of the extended vector can be removed, and the process is repeated until each attribute in the extended vector is judged, and the embedded feature vector can be obtained. The embedded feature vector is obtained through the feature vector centrality and the degree centrality corresponding to each node, the method is novel, and a foundation is provided for later-stage object category evaluation.
The process of obtaining the decision flow sub-model based on the embedded feature vector and the topic library corresponding to the entity is mainly realized by a supervised model and an unsupervised model, wherein the supervised model can include but is not limited to DNN, Wide-Deep, logistic regression model and the like. Specifically, the decision flow submodel herein includes a human anomaly model and a human activity dispersion model. In the step, firstly, a supervised model is used for simulating the embedded feature vector and a subject library corresponding to the entity so as to obtain a personnel abnormality degree model, and then an unsupervised model is used for obtaining a personnel activity dispersion degree model from a personnel activity track in the subject library corresponding to the entity.
Step S1023: a decision stream is generated based on the established knowledge base. Specifically, after the knowledge base is established, the user can form a decision flow by means of dragging or man-machine interaction. Exemplarily, the following description will be made in a man-machine conversation manner. Specifically, for a section of provided audio for troubleshooting of virus-related personnel, firstly, the audio is analyzed into a text by using an NLP technology, and a grammar tree is generated based on the text. Specifically, the obtained "after audio analysis is firstly found out and output the people who are out of the day and night in the latest month from the thematic library; then, a person with abnormal behavior is found from the persons who go out at daytime and night, a text of a key object to be investigated is output, and a grammar tree as shown in fig. 2 can be generated by using the NLP technology. Among these NLP techniques include, but are not limited to LSTM, BERT, etc. And finally, analyzing the syntax tree by utilizing the subsequent traversal of the tree to obtain the decision flow. The post-order traversal (LRD) is a kind of binary tree traversal, and mainly includes a method of traversing a left sub-tree first, then traversing a right sub-tree, and finally accessing a root node to obtain a decision flow.
The training samples are processed to generate a decision flow, and the decision flow is used for training the neural network, so that the neural network with higher accuracy is obtained, and the accuracy of later-stage object class evaluation is improved.
Step S103: and training a generation countermeasure network based on the decision flow generated in the step to obtain the trained generation countermeasure network.
Step S104: and evaluating the sample to be analyzed based on the trained generated confrontation network to obtain the object class. Specifically, before the step of evaluating the to-be-analyzed sample by using the trained generation countermeasure network, the to-be-analyzed sample is processed to generate a decision flow, and then the decision flow is input to the trained generation countermeasure network, so as to obtain the object class, where the object class may be a person involved in a virus, a fleeing person, a stolen person, and the like, and the method of processing the to-be-analyzed sample to generate the decision flow may refer to the foregoing step S101 to step S102, which is not described herein again. In addition, the sample to be analyzed and the object category of the step can be saved in a knowledge base for data enhancement. The method has the advantages that the generated countermeasure network is utilized to evaluate the sample to be analyzed, convenience and conciseness are realized, and the efficiency of the personnel classification evaluation can be effectively improved.
Based on the steps S101 to S104, the entities in the training samples and the relationships between the entities are extracted, a knowledge base is further established, a decision flow is obtained through the established knowledge base, and the neural network is trained by using the decision flow.
It should be noted that, although the foregoing embodiments describe each step in a specific sequence, those skilled in the art will understand that, in order to achieve the effect of the present invention, different steps do not necessarily need to be executed in such a sequence, and they may be executed simultaneously (in parallel) or in other sequences, and these changes are all within the protection scope of the present invention.
Furthermore, the invention also provides an object class evaluation device.
Referring to fig. 3, fig. 3 is a main configuration block diagram of an object class evaluation apparatus according to an embodiment of the present invention. As shown in fig. 3, the object class evaluation apparatus in the embodiment of the present invention mainly includes an obtaining module 11, a generating module 12, a training module 13, and an evaluating module 14. In some embodiments, one or more of acquisition module 11, generation module 12, and training module 13 and evaluation module 14 may be combined together into one module. In some embodiments, the obtaining module 11 may be configured to obtain a training sample and a sample to be analyzed, the training sample including training text, and the sample to be analyzed including text to be analyzed. The generation module 12 may be configured to generate a decision stream based on the training samples. The training module 13 may be configured to train the generated countermeasure network based on the decision flow, resulting in a trained generated countermeasure network. The evaluation module 14 may be configured to evaluate the sample to be analyzed based on the trained generated confrontation network, resulting in the object class.
In one embodiment, the generation module further comprises an entity and relationship extraction module, a knowledge base establishment module and a decision flow generation module, wherein the entity and relationship extraction module is configured to extract entities in the training sample and relationships between the entities; the knowledge base establishing module is configured to establish a knowledge base based on the entities and the relationships between the entities; the decision stream generation module is configured to generate a decision stream based on the established knowledge base.
In one embodiment, the knowledge base establishing module further comprises a thematic base obtaining module, a knowledge map obtaining module, a vector extracting module and a decision flow sub-model obtaining module. The thematic library acquisition module is configured to acquire a thematic library corresponding to the entity based on the entity; the knowledge-graph acquisition module is configured to obtain a knowledge-graph based on the entities and relationships between the entities; the vector extraction module is configured to extract and save the embedded feature vectors based on the knowledge graph; and the decision flow sub-model acquisition module is configured to acquire and store a decision flow sub-model based on the embedded feature vectors and the topic library corresponding to the entity.
In one embodiment, the vector extraction module further comprises a community structure acquisition module, a centrality acquisition module and an embedded feature vector acquisition module. The community structure acquisition module is configured to acquire a community structure in the knowledge graph based on a community discovery algorithm; the centrality obtaining module is configured to evaluate each node in the community structure to obtain the centrality of the feature vector and the centrality of the centrality; the embedded feature vector obtaining module is configured to obtain an embedded feature vector based on the feature vector centrality and the point centrality corresponding to each node.
In one embodiment, the embedded feature vector acquisition module further comprises a weight acquisition module, a construction module, a decomposition module, a vector expansion module, and an attribute reduction module. The weight obtaining module is configured to obtain the weight of each node and the weight of each edge in the community structure based on the feature vector centrality and the point centrality corresponding to each node; the construction module is configured to construct an adjacency matrix of the community structure based on the weight of each node and the weight of each edge; the decomposition module is configured to decompose the adjacency matrix to obtain a vector representation of each node in the community structure; the vector expansion module is configured to perform vector expansion on the vector representation of each node to obtain an expanded vector; the attribute reduction module is configured to perform attribute reduction on the augmented vector to obtain an embedded feature vector.
The object class estimation apparatus is used for executing the embodiment of the object class estimation method shown in fig. 1, and the technical principles, the solved technical problems, and the generated technical effects of the two are similar, and it can be clearly understood by those skilled in the art that, for convenience and simplicity of description, the specific working process and related descriptions of the object class estimation apparatus may refer to the contents described in the embodiment of the object class estimation method, and are not repeated herein.
It will be understood by those skilled in the art that all or part of the flow of the method according to the above-described embodiment may be implemented by a computer program, which may be stored in a computer-readable storage medium and used to implement the steps of the above-described embodiments of the method when the computer program is executed by a processor. Wherein the computer program comprises computer program code, which may be in the form of source code, object code, an executable file or some intermediate form, etc. The computer-readable storage medium may include: any entity or device capable of carrying said computer program code, media, usb disk, removable hard disk, magnetic diskette, optical disk, computer memory, read-only memory, random access memory, electrical carrier wave signals, telecommunication signals, software distribution media, etc. It should be noted that the computer readable storage medium may contain content that is subject to appropriate increase or decrease as required by legislation and patent practice in jurisdictions, for example, in some jurisdictions, computer readable storage media that does not include electrical carrier signals and telecommunications signals in accordance with legislation and patent practice.
Furthermore, the invention also provides a control device. In an embodiment of the control device according to the invention, the control device comprises a processor and a storage device, the storage device may be configured to store a program for executing the object class assessment method of the above-described method embodiment, and the processor may be configured to execute the program in the storage device, the program including but not limited to the program for executing the object class assessment method of the above-described method embodiment. For convenience of explanation, only the parts related to the embodiments of the present invention are shown, and details of the specific techniques are not disclosed. The control device may be a control device apparatus formed including various electronic apparatuses.
Further, the invention also provides a computer readable storage medium. In one computer-readable storage medium embodiment according to the present invention, a computer-readable storage medium may be configured to store a program that executes the object class evaluation method of the above-described method embodiment, and the program may be loaded and executed by a processor to implement the above-described object class evaluation method. For convenience of explanation, only the parts related to the embodiments of the present invention are shown, and details of the specific techniques are not disclosed. The computer readable storage medium may be a storage device formed by including various electronic devices, and optionally, the computer readable storage medium is a non-transitory computer readable storage medium in the embodiment of the present invention.
Further, it should be understood that, since the configuration of each module is only for explaining the functional units of the apparatus of the present invention, the corresponding physical devices of the modules may be the processor itself, or a part of software, a part of hardware, or a part of a combination of software and hardware in the processor. Thus, the number of individual modules in the figures is merely illustrative.
Those skilled in the art will appreciate that the various modules in the apparatus may be adaptively split or combined. Such splitting or combining of specific modules does not cause the technical solutions to deviate from the principle of the present invention, and therefore, the technical solutions after splitting or combining will fall within the protection scope of the present invention.
So far, the technical solutions of the present invention have been described in connection with the preferred embodiments shown in the drawings, but it is easily understood by those skilled in the art that the scope of the present invention is obviously not limited to these specific embodiments. Equivalent changes or substitutions of related technical features can be made by those skilled in the art without departing from the principle of the invention, and the technical scheme after the changes or substitutions can fall into the protection scope of the invention.

Claims (12)

1. An object class evaluation method, comprising the steps of:
acquiring a training sample and a sample to be analyzed;
generating a decision stream based on the training samples;
training and generating an confrontation network based on the decision flow to obtain a trained and generated confrontation network;
and evaluating the sample to be analyzed based on the trained generated confrontation network to obtain the object class.
2. The object class assessment method of claim 1, wherein generating a decision stream based on the training samples comprises:
extracting entities in the training sample and relations among the entities;
establishing a knowledge base based on the entities and the relationship between the entities;
generating a decision flow based on the established knowledge base.
3. The object class assessment method according to claim 2, wherein said knowledge base comprises embedded feature vectors and decision flow sub-models;
building a knowledge base based on the entities and relationships between entities further comprises:
obtaining a subject database corresponding to the entity based on the entity;
obtaining a knowledge graph based on the entities and relationships between the entities;
extracting and storing embedded characteristic vectors based on the knowledge graph;
and obtaining and storing a decision flow sub-model based on the embedded feature vector and a thematic library corresponding to the entity.
4. The object class assessment method according to claim 3, wherein extracting embedded feature vectors based on the knowledge-graph further comprises:
acquiring a community structure in the knowledge graph based on a community discovery algorithm;
evaluating each node in the community structure to obtain the feature vector centrality and the point centrality corresponding to each node;
and obtaining an embedded feature vector based on the feature vector centrality and the point centrality corresponding to each node.
5. The object class evaluation method according to claim 4, wherein obtaining the embedded feature vector based on the feature vector centrality and the point centrality corresponding to each node further comprises:
acquiring the weight of each node and the weight of each edge in the community structure based on the feature vector centrality and the point centrality corresponding to each node;
constructing an adjacency matrix of the community structure based on the weight of each node and the weight of each edge;
decomposing the adjacency matrix to obtain the vector representation of each node in the community structure;
carrying out vector expansion on the vector representation of each node to obtain an expanded vector;
and carrying out attribute reduction on the expansion vector to obtain an embedded feature vector.
6. An object category evaluation apparatus, comprising:
an acquisition module configured to acquire a training sample and a sample to be analyzed;
a generation module configured to generate a decision stream based on the training samples;
a training module configured to train a generation countermeasure network based on the decision flow, resulting in a trained generation countermeasure network;
and the evaluation module is configured to evaluate the sample to be analyzed based on the trained generated confrontation network to obtain the object class.
7. The object class assessment apparatus of claim 6, wherein said generation module further comprises:
an entity and relationship extraction module configured to extract entities and relationships between entities in the training sample;
a knowledge base establishing module configured to establish a knowledge base based on the entities and relationships between the entities;
a decision stream generation module configured to generate a decision stream based on the established knowledge base.
8. The object class assessment apparatus according to claim 7, wherein said knowledge base comprises embedded feature vectors and decision flow sub-models; the knowledge base establishing module further comprises:
a topic library acquisition module configured to acquire a topic library corresponding to an entity based on the entity,
a knowledge graph acquisition module configured to obtain a knowledge graph based on the entities and relationships between the entities;
a vector extraction module configured to extract and save embedded feature vectors based on the knowledge-graph;
and the decision flow sub-model acquisition module is configured to acquire and store a decision flow sub-model based on the embedded feature vector and the topic library corresponding to the entity.
9. The object class assessment apparatus according to claim 8, wherein said vector extraction module further comprises:
a community structure acquisition module configured to acquire a community structure in the knowledge graph based on a community discovery algorithm;
the centrality obtaining module is configured to evaluate each node in the community structure to obtain a feature vector centrality and a centrality of degree corresponding to each node;
and the embedded feature vector acquisition module is configured to acquire an embedded feature vector based on the feature vector centrality and the point centrality corresponding to each node.
10. The object class evaluation device according to claim 9, wherein the embedded feature vector acquisition module further comprises:
a weight obtaining module configured to obtain a weight of each node and a weight of each edge in the community structure based on the feature vector centrality and the point centrality corresponding to each node;
a construction module configured to construct an adjacency matrix of the community structure based on the weight of each node and the weight of each edge;
a decomposition module configured to decompose the adjacency matrix to obtain a vector representation of each node in the community structure;
the vector expansion module is configured to perform vector expansion on the vector representation of each node to obtain an expanded vector;
and the attribute reduction module is configured to perform attribute reduction on the extended vector to obtain an embedded feature vector.
11. A control device comprising a processor and a storage device adapted to store a plurality of program codes, characterized in that said program codes are adapted to be loaded and run by said processor to perform the object class assessment method according to any one of claims 1 to 5.
12. A computer-readable storage medium, in which a plurality of program codes are stored, characterized in that the program codes are adapted to be loaded and executed by a processor to perform the object class assessment method according to any one of claims 1 to 5.
CN202111308616.1A 2021-11-05 2021-11-05 Object class evaluation method and device Pending CN114021655A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111308616.1A CN114021655A (en) 2021-11-05 2021-11-05 Object class evaluation method and device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111308616.1A CN114021655A (en) 2021-11-05 2021-11-05 Object class evaluation method and device

Publications (1)

Publication Number Publication Date
CN114021655A true CN114021655A (en) 2022-02-08

Family

ID=80061630

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111308616.1A Pending CN114021655A (en) 2021-11-05 2021-11-05 Object class evaluation method and device

Country Status (1)

Country Link
CN (1) CN114021655A (en)

Similar Documents

Publication Publication Date Title
CN108763319B (en) Social robot detection method and system fusing user behaviors and text information
CN110956254B (en) Case reasoning method based on dynamic knowledge representation learning
CN112468347B (en) Security management method and device for cloud platform, electronic equipment and storage medium
CN111259154B (en) Data processing method and device, computer equipment and storage medium
CN109325232A (en) A kind of user behavior exception analysis method, system and storage medium based on LDA
CN117421595B (en) System log anomaly detection method and system based on deep learning technology
CN113011889A (en) Account abnormity identification method, system, device, equipment and medium
CN113362852A (en) User attribute identification method and device
CN113378160A (en) Graph neural network model defense method and device based on generative confrontation network
CN112100398A (en) Patent blank prediction method and system
CN115565038A (en) Content audit, content audit model training method and related device
CN113626826A (en) Intelligent contract security detection method, system, equipment, terminal and application
CN118018260A (en) Network attack detection method, system, equipment and medium
CN112435034A (en) Marketing arbitrage black product identification method based on multi-network graph aggregation
CN117078259A (en) Cross-chain abnormal transaction detection method and system based on graph random neural network
CN115567305B (en) Sequential network attack prediction analysis method based on deep learning
CN114021655A (en) Object class evaluation method and device
CN113869049B (en) Fact extraction method and device with legal attribute based on legal consultation problem
CN115293872A (en) Method for establishing risk identification model and corresponding device
CN115587318A (en) Source code classification method based on neural network
Bhat et al. A density-based approach for mining overlapping communities from social network interactions
CN114329184A (en) Expert recommendation method and system in question and answer website
CN113806517A (en) Outbound method, device, equipment and medium based on machine learning model
De Maria et al. A model-checking approach to reduce spiking neural networks
Lv et al. CEP rule extraction framework based on evolutionary algorithm

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