CN114218610A - Multi-dense block detection and extraction method based on Possion distribution - Google Patents

Multi-dense block detection and extraction method based on Possion distribution Download PDF

Info

Publication number
CN114218610A
CN114218610A CN202111406839.1A CN202111406839A CN114218610A CN 114218610 A CN114218610 A CN 114218610A CN 202111406839 A CN202111406839 A CN 202111406839A CN 114218610 A CN114218610 A CN 114218610A
Authority
CN
China
Prior art keywords
data
dense
block
tensor data
blocks
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.)
Granted
Application number
CN202111406839.1A
Other languages
Chinese (zh)
Other versions
CN114218610B (en
Inventor
王俊松
边荟凇
虞振峰
陈诚
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.)
Nanjing College of Information Technology
Original Assignee
Nanjing College of Information Technology
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 Nanjing College of Information Technology filed Critical Nanjing College of Information Technology
Priority to CN202111406839.1A priority Critical patent/CN114218610B/en
Publication of CN114218610A publication Critical patent/CN114218610A/en
Application granted granted Critical
Publication of CN114218610B publication Critical patent/CN114218610B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
    • G06F21/60—Protecting data
    • G06F21/64—Protecting data integrity, e.g. using checksums, certificates or signatures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/21—Design, administration or maintenance of databases
    • G06F16/215—Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Computer Security & Cryptography (AREA)
  • Databases & Information Systems (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioethics (AREA)
  • Software Systems (AREA)
  • Computer Hardware Design (AREA)
  • General Health & Medical Sciences (AREA)
  • Quality & Reliability (AREA)
  • Health & Medical Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Complex Calculations (AREA)

Abstract

The invention discloses a multi-dense block detection and extraction method based on Possion distribution, which comprises the following steps: carrying out suspicious degree measurement on the multi-dimensional tensor data by using a dense block suspicious degree measurement method to obtain a snapnotes list comprising a plurality of suspicious data snapshots; extracting a single dense block from the multi-dimensional tensor data according to a snapshotts list; removing the extracted single dense block from the multi-dimensional tensor data to obtain updated multi-dimensional tensor data; generating a new snapshotts list according to the updated multidimensional tensor data, and extracting new dense blocks until m dense blocks are extracted; the method for measuring the suspicious degree of the dense block is obtained by deducing Possion distribution including counting and density factors. The invention can effectively improve the accuracy and recall rate of the dense blocks while ensuring the detection efficiency.

Description

Multi-dense block detection and extraction method based on Possion distribution
Technical Field
The invention relates to a multi-dense block detection and extraction method based on Possion distribution, and belongs to the technical field of abnormal data detection.
Background
With the advent of the big data age, data anomaly detection becomes more and more important. Abnormal data not only brings data fraud and affects normal data analysis work, but also can cause loss, leakage and the like of normal data, so that the rapid and accurate detection and extraction of abnormal data from mass data is a research focus in the field of data detection.
The method comprises the steps that dense abnormal data exist in abnormal data, the dense abnormal data often have consistency, dense block structures are shown in tensor data, and a plurality of dense block detection and extraction methods appear on the market aiming at the abnormal data, but the existing detection method has some defects, for example, a CrossSpot detection method based on Suspiousness can only detect a single dense block, if the tensor contains a plurality of dense blocks with the same scale, a plurality of dense blocks are combined into a large dense block, and the detection result accuracy and recall rate are not high; the suspicious block detection method based on the binary tree firstly decomposes the dense block every time, and then judges whether the termination condition is met, so that the detection efficiency is very low, excessive decomposition is easy to occur, and the detection accuracy is insufficient.
Disclosure of Invention
Aiming at the problems of low detection accuracy and low recall rate of the existing dense block detection method, the invention provides a multi-dense block detection and extraction method based on Possion distribution, which not only considers the counting of dense blocks but also considers the density of the dense blocks on the basis of the Possion distribution, provides a novel dense block suspicious degree measurement method, and can effectively improve the efficiency, the accuracy and the recall rate of the dense block detection.
In order to solve the technical problems, the invention adopts the following technical means:
the invention provides a multi-dense block detection and extraction method based on Possion distribution, which comprises the following steps:
acquiring multi-dimensional tensor data, the number m of dense blocks to be extracted and the size range of the dense blocks;
carrying out suspicious degree measurement on the multi-dimensional tensor data by using a dense block suspicious degree measurement method to obtain a snapnotes list comprising a plurality of suspicious data snapshots;
extracting a single dense block from the multi-dimensional tensor data according to a snapshotts list;
removing the extracted single dense block from the multi-dimensional tensor data to obtain updated multi-dimensional tensor data;
generating a new snapshotts list according to the updated multidimensional tensor data, and extracting new dense blocks until m dense blocks are extracted;
the method for measuring the suspicious degree of the dense block is obtained by deducing Possion distribution including counting and density factors.
Further, the method for measuring the suspicious degree of the multidimensional tensor data by using the dense block suspicious degree measuring method by setting the dimension of the multidimensional tensor data D as K comprises the following steps:
taking K-dimensional tensor data D as input data;
deleting the column with the least count under each dimension of the input data to obtain the residual block b of each dimensioniWherein i represents dimension, i ═ 1,2, …, K;
calculating residual blocks b of each dimension by using dense block suspicious degree measurement methodiAnd obtaining K DGCS values according to the suspicious degree DGCS values.
Further, the expression of the mission distribution including the two factors of the count and the density is as follows:
Figure BDA0003372591460000031
where P (Y ═ qc) denotes the probability of occurrence of dense blocks in the original multidimensional tensor data under the poisson distribution, Q denotes the density of the original multidimensional tensor data, Q denotes the density of the dense blocks, and c denotes the total count of the dense blocks.
Further, the expression of the DGCS value of the suspicious degree is as follows:
Figure BDA0003372591460000032
wherein the content of the first and second substances,
Figure BDA0003372591460000033
residual block b representing the ith dimensioniThe DGCS value of the degree of suspicion of,
Figure BDA0003372591460000034
representing the remaining block biThe product of the sizes of (a) and (b),
Figure BDA0003372591460000035
representing the remaining block biN denotes a product of sizes of the multi-dimensional tensor data D, C denotes a total count of the multi-dimensional tensor data D,
Figure BDA0003372591460000036
representing the presence of residual blocks b in the multi-dimensional tensor data DiThe probability of (a) of (b) being,
Figure BDA0003372591460000037
representing the remaining block biQ denotes the density of the multi-dimensional tensor data D;
according to the calculation formula of the density, the expression of the DGCS value of the suspicious degree satisfies the following equation:
Q＝C/N (3)
Figure BDA0003372591460000038
Figure BDA0003372591460000039
further, the method for obtaining the snapnotes list containing the plurality of suspicious data snapshots comprises the following steps:
comparing K DGCS values, and recording the residual block with the highest DGCS value as bmax；
Judging the remaining block bmaxWhether the dense block size range is satisfied, if so, the remaining blocks bmaxAs data snapshot B;
the data snapshot B is associated with the DGCS value and then stored in a snapshotts list;
will remain block bmaxAnd as new input data, measuring the suspicious degree again, extracting a new data snapshot and storing the new data snapshot into a snapshotts list until the input data is empty, and obtaining the snapshotts list containing a plurality of suspicious data snapshots.
Further, the data to be detected is converted into multi-dimensional tensor data through data integration, data desensitization, data cleaning and data modeling.
The following advantages can be obtained by adopting the technical means:
the invention provides a multisection block detection and extraction method based on Possion distribution, which is characterized in that data to be detected is converted into multi-dimensional tensor data for subsequent detection, in the dense block detection process, a suspicious degree DGCS value of a data block in the multi-dimensional tensor data is calculated by a designed dense block suspicious degree measurement method to obtain data snapshots meeting requirements, all suspicious data snapshots are uniformly stored in a snapshots list, dense blocks are selected one by one according to the list, and finally a plurality of most suspicious dense blocks are obtained. The method of the invention uses the table to store the data snapshot, avoids splicing or decomposing the dense blocks, ensures the independence and the integrity of the dense blocks in each detection process, and in addition, the method for measuring the suspicious degree of the dense blocks can simultaneously consider the technology and the density factor of the data blocks, ensures that the dense blocks with high density obtain higher DGCS value, is easier to detect, solves the problems of the existing dense block detection method, and improves the efficiency, the accuracy and the recall rate of the dense block detection.
Drawings
FIG. 1 is a flow chart of a multi-dense block detection and extraction method based on Possion distribution according to the present invention;
FIG. 2 is a flow chart of single dense block detection and extraction in the present invention.
Detailed Description
The technical scheme of the invention is further explained by combining the accompanying drawings as follows:
the invention provides a multi-dense block detection and extraction method based on Possion distribution, which specifically comprises the following steps as shown in figures 1 and 2:
step A, acquiring multi-dimensional tensor data, the number m of dense blocks to be extracted and the size range of the dense blocks.
The method for acquiring the multi-dimensional tensor data comprises the following steps: 1. integrating data, namely integrating the data to be detected into a specified data center through an ETL technology; 2. desensitizing data, namely desensitizing sensitive information (such as an identification number) in a production environment in the data to be detected through a desensitizing technology; 3. data cleaning, namely cleaning the desensitized data to ensure the accuracy and consistency of the data; 4. and data preprocessing, namely performing data modeling on the cleaned data, and converting the data to be detected into multi-dimensional tensor data.
The dimension of the multidimensional tensor data D in the embodiment of the invention is K, the number m of dense blocks to be extracted and the size range [ min, max ] of the dense blocks are manually set, min is the lower limit of the size of the dense blocks, and max is the upper limit of the size of the dense blocks.
And B, carrying out suspicious degree measurement on the multi-dimensional tensor data by using a dense block suspicious degree measurement method to obtain a snapnotes list containing a plurality of suspicious data snapshots.
In the method, the dense block suspicious degree measuring method is obtained by deducing the Session distribution including counting and density factors.
In the existing dense block detection method, there is a measure of suspicious degree of dense blocks by using the counted session distribution, that is:
Figure BDA0003372591460000061
where P (Y ═ c) denotes the probability of occurrence of a dense block in the original multidimensional tensor data based on the counting factor under the mission distribution, and c denotes the total count of the dense block, i.e., the sum of all elements in the dense block.
The distribution of the Possions related to counting does not take the density factor of the dense blocks into consideration, the problem that the syndiotactic modules are mixed into a whole block can occur, and the Possions distribution containing the counting and density factors in the method can solve the problem.
The expression of the mission distribution including both count and density factors is as follows:
Figure BDA0003372591460000062
where P (Y ═ qc) denotes the probability of occurrence of a dense block in the original multidimensional tensor data based on the two factors of count and density under the poisson distribution, Q denotes the density of the original multidimensional tensor data, and Q denotes the density of the dense block.
The invention provides a novel dense block suspicious degree measurement method, which can derive a suspicious degree DGCS value of a dense block according to a formula (7), wherein the suspicious degree DGCS value has the following expression:
Figure BDA0003372591460000063
wherein the content of the first and second substances,
Figure BDA0003372591460000064
residual block b representing the ith dimensioniThe DGCS value of the degree of suspicion of,
Figure BDA0003372591460000065
representing the remaining block biThe product of the sizes of (a) and (b),
Figure BDA0003372591460000066
representing the remaining block biN denotes a product of sizes of the multi-dimensional tensor data D, C denotes a total count of the multi-dimensional tensor data D,
Figure BDA0003372591460000071
representing the presence of residual blocks b in the multi-dimensional tensor data DiThe probability of (a) of (b) being,
Figure BDA0003372591460000072
representing the remaining block biI is 1,2, …, K.
According to the calculation formula of the density, the expression of the DGCS value of the suspicious degree satisfies the following equation:
Q＝C/N (9)
Figure BDA0003372591460000073
Figure BDA0003372591460000074
assume that the original multi-dimensional tensor data is a 100 x 100 matrix with a residual block biIs a 50 x 50 matrix, N100 x 100 10000, then C is the sum of 10000 elements in the original multi-dimensional tensor data,
Figure BDA0003372591460000075
then
Figure BDA0003372591460000076
Is the sum of 2500 elements in the remaining block, K ═ 2.
In the embodiment of the present invention, the specific operation of step B is as follows:
and step B01, taking the K-dimensional tensor data D as input data, and establishing an empty list snapnotes for storing snapshots generated in each step of the tensor data D in the optimizing process.
Step B02, deleting the column with the least count in each dimension of the input data to obtain the residual block B in each dimensioni。
The input data is assumed to be a two-dimensional tensor matrix of 100 multiplied by 100, the dimension K of the matrix is 2, the first dimension is a row, and the second dimension is a column; adding 100 elements of each row in the matrix to obtain the count of each row, comparing the counts of 100 rows, selecting the row with the least count, and removing the row with the least count from the 100 rows to obtain a 99-by-100 two-dimensional tensor matrix, which is the residual block b of the first dimension1(ii) a By removing the least counted column from the 100 columns in the same way, a 100 by 99 two-dimensional tensor can be obtainedMatrix, which is the remaining block b of the second dimension2。
Step B03, calculating the residual block B of each dimension by using formula (11)iAnd obtaining K DGCS values according to the suspicious degree DGCS values.
Step B04, comparing K DGCS values, and marking the residual block with the highest DGCS value as Bmax。
Step B05, judging the remaining block BmaxWhether or not the dense block size range, i.e. min, is satisfied<size(bmax)<max, if satisfied, will remain block bmaxAs data snapshot B, otherwise, it is not processed.
And step B06, storing the data snapshot B into a snapshotts list after the data snapshot B is associated with the DGCS value.
Step B07, the remaining blocks B in step B04maxAnd as new input data, measuring the suspicious degree again, extracting a new data snapshot and storing the new data snapshot into a snapshotts list, and stopping circulation until the input data is empty to obtain the snapshotts list containing a plurality of suspicious data snapshots.
Assume the remaining blocks b of the first dimension1The DGCS value of (99 times 100 two-dimensional tensor matrix) is greater than the remaining block b of the second dimension2(100 times 99 two-dimensional tensor matrix) is selected as the DGCS value1As bmax(ii) a When the above steps are repeated, b1(99 times 100 two-dimensional tensor matrix) as the new input data, from b1The least counted row of the 99 rows of (a) is removed to obtain a 98 by 100 two-dimensional tensor matrix from b1The column with the least count is removed from the 100 columns to obtain a 99-by-99 two-dimensional tensor matrix, and the DGCS values of the 98-by-100 two-dimensional tensor matrix and the 99-by-99 two-dimensional tensor matrix are compared; and so on.
And C, extracting a single dense block from the multi-dimensional tensor data according to the snapnotes list, wherein the concrete operations are as follows: and finding out the data snapshot Bmax with the maximum DGCS value from the snapshotts list as a dense block.
And D, removing the extracted single dense block from the multi-dimensional tensor data to obtain the updated multi-dimensional tensor data.
And E, repeating the steps B to D according to the updated multi-dimensional tensor data to generate a new snapshotts list, and extracting new dense blocks until m dense blocks are extracted.
Compared with the prior art, the invention has the following advantages:
1. in the process of detecting the dense blocks, the dense blocks are not spliced or decomposed, so that the independence and the integrity of each dense block are ensured, the recall rate of the dense blocks can be effectively improved, and the method is particularly suitable for detecting a plurality of dense blocks with the same or similar specifications. Assuming that there is a matrix of 100 by 100, in which there are two dense blocks, one is a small block of 20 by 20 with row and column labels both from 1 to 20, and the other is a small block of 21 to 40 with row and column labels, if the dense block detection is performed by using the CrossSpot detection method based on Suspiciousness, it is possible to detect a large dense block (a block of 1 to 40 with row and column labels both from 1 to 40) with 40 by 40 and with two small dense blocks as diagonal matrix, and the upper right corner and lower left corner of the large dense block are non-dense, so the recall rate of the large dense block is only about 50%, whereas the dense block detection performed by using the method of the present invention takes the density problem into full account, and the higher density (small dense block) has the larger DGCS value, and the lower density (large dense block) has the lower DGCS value, so that the method of the present invention can accurately detect two dense blocks of 20 by 20 (not directly multiplying 40), the recall rate of the dense blocks is close to 100 percent and is far greater than that of the prior art.
2. The method has higher detection efficiency and higher detection accuracy. Assuming that a matrix of 100 by 100 has 1 dense block of 30 by 30 and 1 dense block of 15 by 15, and the number of dense block detections is 2, if the dense block detection is performed by using a suspicious block detection method based on binary tree, the method may decompose the dense block of 30 by 30 into 1 small dense block of 20 by 20 and a small dense block of 10 by 10, and after the decomposition, the degree of suspicious of the small dense block of 10 by 10 decreases, and finally retrieve a dense block of 15 by 15 and a small dense block of 20 by 20, which results in insufficient accuracy of dense block detection, and the finally extracted dense block is not the most suspicious. The method does not have the problem, does not split the dense block, directly calculates the DGCS value of the rest blocks, detects the dense block in real time, has higher efficiency, and can accurately find the really most suspicious block.
The above description is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, several modifications and variations can be made without departing from the technical principle of the present invention, and these modifications and variations should also be regarded as the protection scope of the present invention.

Claims (6)

1. A multi-dense block detection and extraction method based on Possion distribution is characterized by comprising the following steps:
acquiring multi-dimensional tensor data, the number m of dense blocks to be extracted and the size range of the dense blocks;
carrying out suspicious degree measurement on the multi-dimensional tensor data by using a dense block suspicious degree measurement method to obtain a snapnotes list comprising a plurality of suspicious data snapshots;
extracting a single dense block from the multi-dimensional tensor data according to a snapshotts list;
removing the extracted single dense block from the multi-dimensional tensor data to obtain updated multi-dimensional tensor data;
generating a new snapshotts list according to the updated multidimensional tensor data, and extracting new dense blocks until m dense blocks are extracted;
the method for measuring the suspicious degree of the dense block is obtained by deducing Possion distribution including counting and density factors.
2. The method as claimed in claim 1, wherein the dimension of the multidimensional tensor data D is K, and the method for measuring the suspicious degree of the multidimensional tensor data by using the dense block suspicious degree measuring method comprises:
taking K-dimensional tensor data D as input data;
each dimension of the input dataDeleting the column with the least count under the degree to obtain the residual block b of each dimensioniWherein i represents dimension, i ═ 1,2, …, K;
calculating residual blocks b of each dimension by using dense block suspicious degree measurement methodiAnd obtaining K DGCS values according to the suspicious degree DGCS values.
3. The method of claim 1, wherein the expression of the Possion distribution including the count and density factors is as follows:
Figure FDA0003372591450000021
where P (Y ═ qc) denotes the probability of occurrence of dense blocks in the original multidimensional tensor data under the poisson distribution, Q denotes the density of the original multidimensional tensor data, Q denotes the density of the dense blocks, and c denotes the total count of the dense blocks.
4. The Possion distribution-based multi-dense-block detection and extraction method as claimed in claim 2 or 3, wherein the DGCS value of the suspicious degree is expressed as follows:
Figure FDA0003372591450000022
wherein the content of the first and second substances,
Figure FDA0003372591450000023
residual block b representing the ith dimensioniThe DGCS value of the degree of suspicion of,
Figure FDA0003372591450000024
representing the remaining block biThe product of the sizes of (a) and (b),
Figure FDA0003372591450000025
representing the remaining blocksbiN denotes a product of sizes of the multi-dimensional tensor data D, C denotes a total count of the multi-dimensional tensor data D,
Figure FDA0003372591450000026
representing the presence of residual blocks b in the multi-dimensional tensor data DiThe probability of (a) of (b) being,
Figure FDA0003372591450000027
representing the remaining block biQ denotes the density of the multi-dimensional tensor data D;
according to the calculation formula of the density, the expression of the DGCS value of the suspicious degree satisfies the following equation:
Q＝C/N
Figure FDA0003372591450000028
Figure FDA0003372591450000029
5. the method of claim 2, wherein the step of obtaining a snapnotes list comprising suspicious data snapshots comprises:
comparing K DGCS values, and recording the residual block with the highest DGCS value as bmax；
Judging the remaining block bmaxWhether the dense block size range is satisfied, if so, the remaining blocks bmaxAs data snapshot B;
the data snapshot B is associated with the DGCS value and then stored in a snapshotts list;
will remain block bmaxAnd as new input data, measuring the suspicious degree again, extracting a new data snapshot and storing the new data snapshot into a snapshotts list until the input data is empty, and obtaining the snapshotts list containing a plurality of suspicious data snapshots.
6. The Possion distribution-based multi-density block detection and extraction method as claimed in claim 1, wherein the data to be detected is converted into multi-dimensional tensor data by data integration, data desensitization, data cleaning and data modeling.
CN202111406839.1A 2021-11-24 2021-11-24 Multi-dense block detection and extraction method based on Possion distribution Active CN114218610B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111406839.1A CN114218610B (en) 2021-11-24 2021-11-24 Multi-dense block detection and extraction method based on Possion distribution

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111406839.1A CN114218610B (en) 2021-11-24 2021-11-24 Multi-dense block detection and extraction method based on Possion distribution

Publications (2)

Publication Number Publication Date
CN114218610A true CN114218610A (en) 2022-03-22
CN114218610B CN114218610B (en) 2023-02-14

Family

ID=80698186

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111406839.1A Active CN114218610B (en) 2021-11-24 2021-11-24 Multi-dense block detection and extraction method based on Possion distribution

Country Status (1)

Country Link
CN (1) CN114218610B (en)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100312797A1 (en) * 2009-06-05 2010-12-09 Xerox Corporation Hybrid tensor-based cluster analysis
CN107832688A (en) * 2017-10-27 2018-03-23 浙江农林大学 A kind of travel pattern of traffic intersection video monitoring and the detection method of abnormal behaviour
CN109064189A (en) * 2018-07-13 2018-12-21 北京亚鸿世纪科技发展有限公司 Brush list detecting and alarm device based on the detection of intensive block
CN110087293A (en) * 2019-05-13 2019-08-02 深圳职业技术学院 A kind of low energy consumption distributed event detection wireless sensor network construction method
CN111523012A (en) * 2019-02-01 2020-08-11 慧安金科（北京）科技有限公司 Method, apparatus, and computer-readable storage medium for detecting abnormal data
US20200285997A1 (en) * 2019-03-04 2020-09-10 Iocurrents, Inc. Near real-time detection and classification of machine anomalies using machine learning and artificial intelligence

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100312797A1 (en) * 2009-06-05 2010-12-09 Xerox Corporation Hybrid tensor-based cluster analysis
CN107832688A (en) * 2017-10-27 2018-03-23 浙江农林大学 A kind of travel pattern of traffic intersection video monitoring and the detection method of abnormal behaviour
CN109064189A (en) * 2018-07-13 2018-12-21 北京亚鸿世纪科技发展有限公司 Brush list detecting and alarm device based on the detection of intensive block
CN111523012A (en) * 2019-02-01 2020-08-11 慧安金科（北京）科技有限公司 Method, apparatus, and computer-readable storage medium for detecting abnormal data
US20200285997A1 (en) * 2019-03-04 2020-09-10 Iocurrents, Inc. Near real-time detection and classification of machine anomalies using machine learning and artificial intelligence
CN110087293A (en) * 2019-05-13 2019-08-02 深圳职业技术学院 A kind of low energy consumption distributed event detection wireless sensor network construction method

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
MENG JIANG 等: "Spotting Suspicious Behaviors in Multimodal Data: A General Metric and Algorithms", 《IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING》 *
陈运文等: "基于异常检测的时间序列研究", 《计算机技术与发展》 *

Also Published As

Publication number Publication date
CN114218610B (en) 2023-02-14

Similar Documents

Publication Publication Date Title
CN107220261B (en) Real-time mining method and device based on distributed data
CN104216349B (en) Utilize the yield analysis system and method for the sensing data of manufacturing equipment
CN115145906B (en) Preprocessing and completion method for structured data
CN110826618A (en) Personal credit risk assessment method based on random forest
CN112949279A (en) Method, system and storage medium for analyzing abnormal data in electronic document
CN114511718B (en) Intelligent management method and system for materials for building construction
CN113077444A (en) CNN-based ultrasonic nondestructive detection image defect classification method
CN113051308A (en) Alarm information processing method, equipment, storage medium and device
CN113254988A (en) High-dimensional sensitive data privacy classified protection publishing method, system, medium and equipment
CN112035621A (en) Enterprise name similarity detection method based on statistics
CN114218610B (en) Multi-dense block detection and extraction method based on Possion distribution
CN111191239B (en) Process detection method and system for application program
CN114997750B (en) Risk information pushing method, system, equipment and medium
CN116610987A (en) Kmeans log classification method and device based on distributed sample screening
CN114285601B (en) Multi-dense-block detection and extraction method for big data
CN105843930A (en) Video search method and device
CN112613521B (en) Multilevel data analysis system and method based on data conversion
CN112333155B (en) Abnormal flow detection method and system, electronic equipment and storage medium
CN104484869A (en) Image matching method and system for ordinal measure features
CN113569879B (en) Training method of abnormal recognition model, abnormal account recognition method and related device
CN114528909A (en) Unsupervised anomaly detection method based on flow log feature extraction
KR101907443B1 (en) Component-based malicious file similarity analysis device and method
CN116502277B (en) Medical data safety processing method, system and device based on blockchain
CN113761034B (en) Data processing method and device
CN115797334B (en) Method and device for rapidly judging quality of large-scale light spots

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
GR01 Patent grant
GR01 Patent grant