CN114936581B - Multi-parameter association mining method based on time sequence data segmentation - Google Patents
Multi-parameter association mining method based on time sequence data segmentation Download PDFInfo
- Publication number
- CN114936581B CN114936581B CN202210621752.4A CN202210621752A CN114936581B CN 114936581 B CN114936581 B CN 114936581B CN 202210621752 A CN202210621752 A CN 202210621752A CN 114936581 B CN114936581 B CN 114936581B
- Authority
- CN
- China
- Prior art keywords
- parameter
- association
- trend
- segmentation
- segment
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 42
- 230000011218 segmentation Effects 0.000 title claims abstract description 42
- 238000005065 mining Methods 0.000 title claims abstract description 30
- 238000004364 calculation method Methods 0.000 claims abstract description 13
- 238000012098 association analyses Methods 0.000 claims abstract description 8
- 239000011159 matrix material Substances 0.000 claims description 9
- 238000000605 extraction Methods 0.000 claims description 7
- 230000000630 rising effect Effects 0.000 claims description 4
- 238000005516 engineering process Methods 0.000 claims description 3
- 230000001360 synchronised effect Effects 0.000 claims description 3
- 238000012800 visualization Methods 0.000 claims description 3
- 230000001174 ascending effect Effects 0.000 claims description 2
- 230000009514 concussion Effects 0.000 claims description 2
- 230000002123 temporal effect Effects 0.000 claims 4
- 238000009412 basement excavation Methods 0.000 abstract description 5
- 238000012545 processing Methods 0.000 abstract description 2
- 230000000694 effects Effects 0.000 description 3
- 238000013139 quantization Methods 0.000 description 3
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000007429 general method Methods 0.000 description 1
- 230000010355 oscillation Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- 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
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2465—Query processing support for facilitating data mining operations in structured databases
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Fuzzy Systems (AREA)
- Signal Processing (AREA)
- Probability & Statistics with Applications (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention relates to a multi-parameter association mining method based on time sequence data segmentation, and belongs to the field of big data processing. The method comprises the steps of initializing parameters needing to be subjected to association analysis, analyzing time sequence data, extracting trend segmentation features of the data to obtain a quantized result in each segmentation feature, calculating association relation based on the quantized result, and finally outputting a multi-parameter association mining result. Meanwhile, expert knowledge is supported by the method to improve accuracy of the association result. The method can improve the calculation performance, and only needs to calculate whether the sectional trends of all the parameters are related or not, so that the speed of an excavation algorithm is greatly improved; according to the parameter association mining algorithm based on trend segmentation, the trend in the segment is quantized and calculated, a threshold value is not required to be set, and the universality of the algorithm is improved; the method can be used for excavating under the condition of fewer historical samples, and can solve the problem that the fewer samples cannot be subjected to multi-parameter association excavation.
Description
Technical Field
The invention belongs to the field of big data processing, and particularly relates to a multi-parameter association mining method based on time sequence data segmentation.
Background
The prior art performs association analysis on sensor time sequence data strictly according to time periods, and calculation and mining on unsegmented data are needed. The most similar prior art to this scheme includes the Apriori method, the FP-Growth method, and the like.
The existing calculation process of multi-parameter association mining is generally as follows:
1. Carrying out standardized preprocessing operation on the original data to reduce the influence of data bias, amplitude scaling, linear trend and noise on calculation;
2. Setting an association threshold value, and utilizing sliding window operation to perform frequent pattern mining on a multi-parameter time sequence;
3. And generating a time sequence association result.
The existing multi-parameter association mining method needs to set association threshold values, performs feature extraction on original data by utilizing operation of a sliding window, and then mines multi-parameter features under the same window.
According to the scheme, the trend segmentation is carried out on the original data, sliding window operation is not needed, the calculated amount can be greatly reduced, the setting of the association threshold value is avoided through the quantization and calculation of the trend, and the universality of the algorithm is improved.
Disclosure of Invention
First, the technical problem to be solved
The invention aims to solve the technical problem of how to provide a multi-parameter association mining method based on time sequence data segmentation so as to solve the problems that the calculation amount of the existing multi-parameter association mining algorithm is large, and the generality of the algorithm cannot be reduced uniformly for thresholds of different scenes.
(II) technical scheme
In order to solve the technical problems, the invention provides a multi-parameter association mining method based on time sequence data segmentation, which comprises the following steps:
Aiming at parameters obtained by a sensor, extracting a patch as a local data point in a downsampling mode by setting a fixed step length, defining a patch set P i＝{Pi,1,Pi,2,…,Pi,n},Pi,l on the parameters D i (t) obtained by the sensor to represent a time sequence data set in the patch, wherein n represents the number of the patches, calculating the characteristic value of each patch, and extracting a trend turning point after calculating the characteristic value of the patch;
obtaining a plurality of segments according to the trend turning points, and marking as follows: s i＝{Si,1,Si,2,…,Si,Ki, wherein K i is the number of segments of parameter D i (t), each intra-segment feature is represented by S i,k, wherein S i,k＝{Loci,k,Trendi,k,Fiti,k; extracting relevant characteristic indexes from the intra-segment data through quantized S i,k: loc i,k,Trendi,k and Fit i,k;
Loc i,k represents the start-stop position and corresponding offset of the kth segment feature of parameter D i (t);
trend i,k = { < flag, num > } represents the overall Trend sign of the current segment, wherein the flag represents the Trend of the current segment, and num is used for representing the number of latches corresponding to the flag;
fit i,k represents the result of linear fitting of the flag tag in the kth segment feature Trend i,k of parameter D i (t) as up, down and horizontal segments;
Based on the segmentation result, describing the association relation and the association degree between the parameters through an association matrix, wherein the association matrix is R= (R ij)N×N, N is the number of parameters participating in association analysis, and R ij is calculated according to the following formula:
Where S i,Sj denotes a consistency feature of the parameter i and the parameter j, and F (S i,Sj) is used to calculate a degree of association of the two parameters, which is defined as follows:
Where K represents the maximum number of feature segments for participating in the association calculation, i.e., k=max (K i,Kj);Ki is the number of segments of parameter i, K j is the number of segments of parameter j; for representing whether synchronization of two parameters occurs according to trend feature statistics of the segmentation result, the method is defined as follows:
wherein, The k 1 th segment for indicating the i-th parameter and the k 2 th segment for the j-th parameter are both ascending and descending, and Λ is an intersection or phase, which is defined as follows:
Indicating whether the corresponding time window satisfies a condition when the k 1 th segment of the i-th parameter and the k 2 th segment of the j-th parameter are trend-related synchronized, which is defined as follows:
Where Φ represents the empty set.
Further, a median (P i,l) is calculated for each P i,l as a feature value for each patch.
Further, each parameter has a plurality of history data, each of which is time series data.
Further, loc i,k＝{start,end,Δstart,Δend, where start represents start time, end represents stop time, Δ start represents offset of start, and Δ end represents offset of end.
Further, the flag includes rising, falling, level and concussion.
Further, fit i,k = { type: linear, slope, bias, fitter error }, where type: linear represents the type of Fit is linear, slope represents the slope of the linear Fit, bias represents the intercept of the linear Fit, and Fitter error represents the fitting error of the linear Fit.
Further, the method comprises the steps of,When calculated, the association synchronization is considered to occur when it appears to be one-up and one-down.
Further, the method comprises the steps of,The corresponding time window is from the Loc feature of the segmentation result.
Further, the method further comprises: after the association matrix R is obtained, the association relation and the association degree between the parameters are displayed by using a visualization technology.
Further, the method comprises the steps of initializing parameters needing to be subjected to association analysis, analyzing time sequence data, extracting trend segmentation features of the data to obtain a quantization result in each segmentation feature, calculating association relation based on the quantization result, and finally outputting a multi-time sequence data association mining result.
(III) beneficial effects
The invention provides a multi-parameter association mining method based on time sequence data segmentation, which has the technical effects that:
1. the method can improve the calculation performance, and only needs to calculate whether the sectional trends of all the parameters are related or not, so that the speed of an excavation algorithm is greatly improved;
2. According to the parameter association mining algorithm based on trend segmentation, the trend in the segment is quantized and calculated, a threshold value is not required to be set, and the universality of the algorithm is improved;
3. The method can be used for excavating under the condition of fewer historical samples, and can solve the problem that the fewer samples cannot be subjected to multi-parameter association excavation.
Drawings
FIG. 1 is a schematic diagram of discrete extraction of Patch points and trend feature points;
Fig. 2 is a flowchart of the multi-parameter association calculation.
Detailed Description
To make the objects, contents and advantages of the present invention more apparent, the following detailed description of the present invention will be given with reference to the accompanying drawings and examples.
The invention carries out multi-parameter association relation mining based on the time sequence data segmentation result, 1) trend segmentation is carried out on the time sequence data; 2) Quantifying and calculating the trend of the data; 3) And extracting and fusing the parameter association relation.
The technical innovation points of the method are mainly divided into three parts, namely:
1. segmenting time series data by utilizing downsampling patch operation and trend turning point extraction
The trend segmentation module in the method improves a trend turning point algorithm, extracts the trend turning points directly from noise dense data and frequent jump data, and can generate a large number of dense trend turning points so as to greatly influence the data segmentation effect. Therefore, the method is based on the mode of taking the patch discretely on the original data, and the extraction of the trend turning points is carried out, as shown in fig. 1.
For parameters obtained by a sensor, extracting a patch as a local data point in a downsampling mode by setting a fixed step length, defining a patch set P i＝{Pi,1,Pi,2,…,Pi,n},Pi,l on the parameters D i (t) obtained by the sensor to represent a time sequence data set in the patch, wherein n represents the number of the patch, performing median calculation media (P i,l) on each P i,l to represent the characteristic value of each patch, and performing trend turning point extraction after calculating the characteristic value of the patch. The trend turning point can be extracted by a general method.
A parameter has a plurality of history data, each of which is time series data.
2. Quantizing and computing data segments
According to the trend turning points, a plurality of segments can be obtained and marked as follows: Where K i is the number of segments of parameter D i (t), here denoted by S i,k for each intra-segment feature, where S i,k＝{Loci,k,Trendi,k,Fiti,k. The related characteristic indexes can be extracted from the data in the segment through the quantized S i,k, and the following three main indexes are mainly extracted: loc i,k,Trendi,k and Fit i,k.
Loc i,k denotes the start-stop position and corresponding offset of the kth segment feature of parameter D i (t), noted: loc i,k＝{start,end,Δstart,Δend, where start represents start time, end represents end time, Δ start represents offset of start, and Δ end represents offset of end.
Trend i,k = { < flag, num > } represents the overall Trend sign of the current segment, where flag represents the Trend of the current segment, such as: rising (UP), falling (DOWN), level (HORIZON) and oscillation (VIBRATE), num is used to indicate the number of latches corresponding to a flag.
Fit i,k represents the result of linear fitting of the kth segment feature Trend i,k of parameter D i (t) with the flag label rising (UP), falling (DOWN) and Horizontal (HORIZON) segments. For example, fit i,k = { type: linear, slope:1.5, bias: -3, fitterror: 0.000001}, where type: linear indicates that the Fit type is linear, slope:1.5 indicates that the slope of the linear Fit is 1.5, bias: -3 indicates that the intercept of the linear Fit is-3,Fitter error:0.000001 indicates that the Fit error of the linear Fit is 0.000001.
3. Extracting and fusing the parameter association relation
Based on the segmentation result of the last step, the association relationship and the association degree between the parameters can be described by an association matrix, wherein the association matrix is r= (R ij)N×N, N is the number of parameters participating in association analysis:
Where S i,Sj denotes a consistency feature of the parameter i and the parameter j, and F (S i,Sj) is used to calculate a degree of association of the two parameters, which is defined as follows:
Where K represents the maximum number of feature segments for participating in the association calculation, i.e., k=max (K i,Kj);Ki is the number of segments of parameter i, K j is the number of segments of parameter j; for representing whether synchronization of two parameters occurs according to trend feature statistics of the segmentation result, the method is defined as follows:
wherein, The k 1 th segment for indicating the i-th parameter and the k 2 th segment for the j-th parameter are both UP (UP) and DOWN (DOWN), and are considered to be associated synchronous when they appear UP and DOWN, Λ being the intersection or phase, defined as follows:
Indicating whether the corresponding time window (Loc feature from the segmentation result) satisfies a condition when the k 1 th segment of the i-th parameter and the k 2 th segment of the j-th parameter are in trend-related synchronization, which is defined as follows:
Where Φ represents the empty set.
4. After the association matrix R is obtained, the association relation and the association degree between the parameters can be displayed by using a visualization technology.
In summary, the calculation flow of the method is shown in fig. 2, the parameters to be subjected to association analysis are initialized, time series data are analyzed, trend segmentation feature extraction is performed on the data to obtain a quantized result in each segmentation feature, association relation is calculated based on the quantized result, and finally a multi-time series data association mining result is output. Meanwhile, expert knowledge is supported by the method to improve accuracy of the association result.
The invention has the technical effects that:
1. according to the method, the computing performance can be improved, and only whether the sectional trend of each time sequence data is associated or not is needed to be computed, so that the speed of an excavating algorithm is greatly improved;
2. According to the parameter association mining algorithm based on trend segmentation, the trend in the segment is quantized and calculated, a threshold value is not required to be set, and the universality of the algorithm is improved;
3. The method can be used for excavating under the condition of fewer historical samples, and can solve the problem that the fewer samples cannot be subjected to multi-parameter association excavation.
The foregoing is merely a preferred embodiment of the present invention, and it should be noted that modifications and variations could be made by those skilled in the art without departing from the technical principles of the present invention, and such modifications and variations should also be regarded as being within the scope of the invention.
Claims (10)
1. The multi-parameter association mining method based on time sequence data segmentation is characterized by comprising the following steps of:
Aiming at parameters obtained by a sensor, extracting a patch as a local data point in a downsampling mode by setting a fixed step length, defining a patch set P i＝{Pi,1,Pi,2,…,Pi,n},Pi,l on the parameters D i (t) obtained by the sensor to represent a time sequence data set in the patch, wherein n represents the number of the patches, calculating the characteristic value of each patch, and extracting a trend turning point after calculating the characteristic value of the patch;
obtaining a plurality of segments according to the trend turning points, and marking as follows: Where K i is the number of segments of parameter D i (t), each segment is characterized by S i,k, where S i,k＝{Loci,k,Trendi,k,Fiti,k; extracting relevant characteristic indexes from the intra-segment data through quantized S i,k: loc i,k,Trendi,k and Fit i,k;
Loc i,k represents the start-stop position and corresponding offset of the kth segment feature of parameter D i (t);
Trend i,k = { < flag, num > } represents the overall Trend sign of the current segment, wherein the flag represents the Trend of the current segment, and num is used for representing the number of latches corresponding to the flag;
fit i,k represents the result of linear fitting of the flag tag in the kth segment feature Trend i,k of parameter D i (t) as up, down and horizontal segments;
Based on the segmentation result, describing the association relation and the association degree between the parameters through an association matrix, wherein the association matrix is R= (R ij)N×N, N is the number of parameters participating in association analysis, and R ij is calculated according to the following formula:
Where S i,Sj denotes a consistency feature of the parameter i and the parameter j, and F (S i,Sj) is used to calculate a degree of association of the two parameters, which is defined as follows:
Where K represents the maximum number of feature segments for participating in the association calculation, i.e., k=max (K i,Kj);Ki is the number of segments of parameter i, K j is the number of segments of parameter j; for representing whether synchronization of two parameters occurs according to trend feature statistics of the segmentation result, the method is defined as follows:
wherein, The k 1 th segment for indicating the i-th parameter and the k 2 th segment for the j-th parameter are both ascending and descending, and Λ is an intersection or phase, which is defined as follows:
Indicating whether the corresponding time window satisfies a condition when the k 1 th segment of the i-th parameter and the k 2 th segment of the j-th parameter are trend-related synchronized, which is defined as follows:
Where Φ represents the empty set.
2. The multi-parameter association mining method based on time series data segmentation according to claim 1, wherein median (P i,l) is calculated for each P i,l as a feature value of each patch.
3. The method of multi-parameter associative mining based on temporal data segmentation of claim 1, wherein each parameter has a plurality of historical data, each historical data being temporal data.
4. The multi-parameter association mining method based on time series data segmentation according to claim 1, wherein Loc i,k＝{start,end,Δstart,Δend, wherein start represents a start time, end represents a stop time, Δ start represents an offset of the start, and Δ end represents an offset of the end.
5. The multi-parameter associative mining method based on time series data segmentation according to claim 4, wherein the flag includes rising, falling, level and concussion.
6. The method of multi-parameter associative mining based on temporal data segmentation of claim 5, wherein Fit i,k = { type: linear, slope, bias, fitter error }, where type: linear represents the type of Fit is linear, slope represents the slope of the linear Fit, bias represents the intercept of the linear Fit, fitter error represents the Fit error of the linear Fit.
7. The multi-parameter association mining method based on time series data segmentation as set forth in claim 6, wherein,When calculated, the association synchronization is considered to occur when it appears to be one-up and one-down.
8. The multi-parameter association mining method based on time series data segmentation as set forth in claim 1, wherein,The corresponding time window is from the Loc feature of the segmentation result.
9. The method of multi-parameter associative mining based on temporal data segmentation of claim 1, further comprising: after the association matrix R is obtained, the association relation and the association degree between the parameters are displayed by using a visualization technology.
10. The multi-parameter association mining method based on time series data segmentation according to any one of claims 1-9, wherein the method is characterized in that firstly, parameters needing to be subjected to association analysis are initialized, time series data are analyzed, then trend segmentation feature extraction is carried out on the data, quantized results in each segmentation feature are obtained, association relation is calculated based on the quantized results, and finally multi-time series data association mining results are output.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210621752.4A CN114936581B (en) | 2022-06-01 | 2022-06-01 | Multi-parameter association mining method based on time sequence data segmentation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210621752.4A CN114936581B (en) | 2022-06-01 | 2022-06-01 | Multi-parameter association mining method based on time sequence data segmentation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114936581A CN114936581A (en) | 2022-08-23 |
| CN114936581B true CN114936581B (en) | 2024-04-26 |
Family
ID=82865691
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210621752.4A Active CN114936581B (en) | 2022-06-01 | 2022-06-01 | Multi-parameter association mining method based on time sequence data segmentation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114936581B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103676645A (en) * | 2013-12-11 | 2014-03-26 | 广东电网公司电力科学研究院 | Mining method for association rules in time series data flows |
| WO2015176565A1 (en) * | 2014-05-22 | 2015-11-26 | 袁志贤 | Method for predicting faults in electrical equipment based on multi-dimension time series |
| CN107562865A (en) * | 2017-08-30 | 2018-01-09 | 哈尔滨工业大学深圳研究生院 | Multivariate time series association rule mining method based on Eclat |
| CN110008253A (en) * | 2019-03-28 | 2019-07-12 | 浙江大学 | The industrial data association rule mining and unusual service condition prediction technique of strategy are generated based on two stages frequent item set |
| CN111046084A (en) * | 2019-12-18 | 2020-04-21 | 重庆大学 | Association rule mining method for multivariate time series monitoring data |
| WO2020177366A1 (en) * | 2019-03-07 | 2020-09-10 | 平安科技（深圳）有限公司 | Data processing method and apparatus based on time sequence data, and computer device |
-
2022
- 2022-06-01 CN CN202210621752.4A patent/CN114936581B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103676645A (en) * | 2013-12-11 | 2014-03-26 | 广东电网公司电力科学研究院 | Mining method for association rules in time series data flows |
| WO2015176565A1 (en) * | 2014-05-22 | 2015-11-26 | 袁志贤 | Method for predicting faults in electrical equipment based on multi-dimension time series |
| CN107562865A (en) * | 2017-08-30 | 2018-01-09 | 哈尔滨工业大学深圳研究生院 | Multivariate time series association rule mining method based on Eclat |
| WO2020177366A1 (en) * | 2019-03-07 | 2020-09-10 | 平安科技（深圳）有限公司 | Data processing method and apparatus based on time sequence data, and computer device |
| CN110008253A (en) * | 2019-03-28 | 2019-07-12 | 浙江大学 | The industrial data association rule mining and unusual service condition prediction technique of strategy are generated based on two stages frequent item set |
| CN111046084A (en) * | 2019-12-18 | 2020-04-21 | 重庆大学 | Association rule mining method for multivariate time series monitoring data |
Non-Patent Citations (2)
| Title |
|---|
| 基于趋势特征聚类的多元相似时间序列的提取;解初;王建东;韩邦磊;王振;;科学技术与工程;20200308(07);全文 * |
| 基于转折点和趋势段的时间序列趋势特征提取;刘意杨;李俊朋;白洪飞;王智凝;;计算机应用;20200710(S1);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114936581A (en) | 2022-08-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109992782B (en) | Legal document named entity identification method and device and computer equipment | |
| CN109741332B (en) | Man-machine cooperative image segmentation and annotation method | |
| CN112732919B (en) | Intelligent classification label method and system for network security threat information | |
| US20220270369A1 (en) | Intelligent cataloging method for all-media news based on multi-modal information fusion understanding | |
| CN113434357B (en) | Log anomaly detection method and device based on sequence prediction | |
| CN107301170B (en) | Method and device for segmenting sentences based on artificial intelligence | |
| CN108985380B (en) | Point switch fault identification method based on cluster integration | |
| US11861925B2 (en) | Methods and systems of field detection in a document | |
| CN112199501B (en) | Scientific and technological information text classification method | |
| CN113312447A (en) | Semi-supervised log anomaly detection method based on probability label estimation | |
| CN111274817A (en) | Intelligent software cost measurement method based on natural language processing technology | |
| CN111611218A (en) | Distributed abnormal log automatic identification method based on deep learning | |
| CN113841168A (en) | Hierarchical machine learning architecture including a primary engine supported by distributed lightweight real-time edge engines | |
| CN116910013A (en) | System log anomaly detection method based on semantic flowsheet mining | |
| CN115565177A (en) | Character recognition model training method, character recognition device, character recognition equipment and medium | |
| CN114936581B (en) | Multi-parameter association mining method based on time sequence data segmentation | |
| Law et al. | Energy-efficient data acquisition by adaptive sampling for wireless sensor networks | |
| CN113806198A (en) | System state diagnosis method based on deep learning | |
| CN116543162A (en) | Image segmentation method and system based on feature difference and context awareness consistency | |
| CN116363460A (en) | High-resolution remote sensing sample labeling method based on topic model | |
| CN113313178B (en) | Cross-domain image example level active labeling method | |
| CN114842301A (en) | Semi-supervised training method of image annotation model | |
| CN114691918A (en) | Radar image retrieval method and device based on artificial intelligence and electronic equipment | |
| CN103927613A (en) | Process retrieval system and method | |
| CN114119492A (en) | Image processing-based thermal protection function gradient material component identification method and system |
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 |