CN106610980B - Apparatus and method for classifying/predicting spatio-temporal sequence data - Google Patents
Apparatus and method for classifying/predicting spatio-temporal sequence data Download PDFInfo
- Publication number
- CN106610980B CN106610980B CN201510690684.7A CN201510690684A CN106610980B CN 106610980 B CN106610980 B CN 106610980B CN 201510690684 A CN201510690684 A CN 201510690684A CN 106610980 B CN106610980 B CN 106610980B
- Authority
- CN
- China
- Prior art keywords
- spatial
- weight
- sequence data
- spatio
- matrix
- 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
Images
Classifications
-
- 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
- G06F16/29—Geographical information databases
-
- 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
- G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
Abstract
An apparatus for classifying/predicting spatiotemporal sequence data is provided, comprising: a receiving unit configured to receive spatio-temporal sequence data; a modeling unit configured to generate weight parameters related to geographic heterogeneity based on the spatio-temporal sequence data and construct a model for classification/prediction based on the generated weight parameters; and a classification/prediction unit configured to classify/predict the spatio-temporal sequence data using the constructed model for classification/prediction. A method for classifying/predicting spatiotemporal sequence data is also provided. The geographic weighting extreme learning machine provided by the invention can improve the accuracy of classifying or predicting the time-space sequence data by considering the heterogeneity of geographic space.
Description
Technical Field
The present application relates to the field of data analysis, and in particular, to an apparatus and method for classifying/predicting spatiotemporal sequence data.
Background
A large amount of space-time sequence data are accumulated in a plurality of fields such as environment, weather, traffic, economy and the like. Spatio-temporal sequences generally describe the change of a variable over time at different spatial locations. For example, PM2.5 concentration values for 24 hours a day at 35 air monitoring sites in Beijing. Spatio-temporal sequence modeling aims to describe the correlation relationship between spatio-temporal variables, and then classify some values of the spatio-temporal sequences or predict future values through the relationship.
Extreme Learning Machines (ELMs) are a new type of neural network algorithm, originally proposed by HuangGuangbin 2004. Compared with the traditional neural network, the training speed of the ELM is high, and the human interference is less. The basic flow of ELM algorithms is described in reference 1 ("Huang G B, Zhu Q Y, Siew C K. extreme learning machine: A new learning scheme of fed forward neural networks [ J ]. Proc. int. joint Conf. neural Net, 2006, 2: 985-990.).
However, although the conventional extreme learning machine can model the spatio-temporal sequence, it cannot reflect the spatio-temporal characteristics in the data set, and does not consider the geographic heterogeneity, so that the accuracy of classification or prediction is not good enough.
Disclosure of Invention
The invention provides a technical scheme of GWEELM (geographic Weighted Extreme learning Machine), which takes geographic heterogeneity into consideration in the modeling process and converts geographic positions into local weighting coefficients, so that the GWEELM can be better applied to space-time sequence data.
According to an aspect of the present invention, there is provided an apparatus for classifying/predicting spatio-temporal sequence data, comprising: a receiving unit configured to receive spatio-temporal sequence data; a modeling unit configured to generate weight parameters related to geographic heterogeneity based on the spatio-temporal sequence data and construct a model for classification/prediction based on the generated weight parameters; and a classification/prediction unit configured to classify/predict the spatio-temporal sequence data using the constructed model for classification/prediction.
In one embodiment, the modeling unit includes: a spatial distance matrix construction subunit configured to construct a spatial distance matrix based on the spatio-temporal sequence data; a spatial weight matrix construction subunit configured to select at least two different weight parameters for each spatial position, respectively, to construct a corresponding spatial weight matrix; a model output weight calculation subunit configured to calculate respective model output weights based on the respective spatial weight matrices; and a selecting subunit configured to select an output weight corresponding to the weight parameter that yields the smallest cost value as a final model output weight.
In one embodiment, the spatial distance matrix is a gaussian spatial distance matrix and the spatial weight matrix is a gaussian spatial weight matrix.
In one embodiment, the at least two different weight parameters are selected from a range of minimum spatial distance to maximum spatial distance. Preferably, the at least two different weight parameters may be selected linearly from a range of minimum spatial distance to maximum spatial distance. Alternatively, the at least two different weight parameters may be randomly selected from a range of minimum spatial distances to maximum spatial distances.
According to another aspect of the present invention, there is provided a method for classifying/predicting spatio-temporal sequence data, comprising: receiving space-time sequence data; generating a weight parameter related to geographic heterogeneity based on the spatio-temporal sequence data; constructing a model for classification/prediction based on the generated weight parameters; and classifying/predicting the space-time sequence data by adopting the constructed model for classifying/predicting.
In one embodiment, a spatial distance matrix is constructed based on the spatio-temporal sequence data. At least two different weight parameters are respectively selected for each spatial position to construct a corresponding spatial weight matrix. Respective model output weights are calculated based on the respective spatial weight matrices. Then, the output weight corresponding to the weight parameter that produces the smallest cost value is selected as the final model output weight.
In one embodiment, the spatial distance matrix is a gaussian spatial distance matrix and the spatial weight matrix is a gaussian spatial weight matrix.
In one embodiment, the at least two different weight parameters are selected from a range of minimum spatial distance to maximum spatial distance. Preferably, the at least two different weight parameters may be selected linearly from a range of minimum spatial distance to maximum spatial distance. Alternatively, the at least two different weight parameters may be randomly selected from a range of minimum spatial distances to maximum spatial distances.
The geographic weighting extreme learning machine provided by the invention can improve the accuracy of classifying or predicting the time-space sequence data by considering the heterogeneity of geographic space.
Drawings
The above and other features of the present invention will become more apparent from the following detailed description when taken in conjunction with the accompanying drawings, in which:
FIG. 1 is a block diagram illustrating an apparatus for classifying/predicting spatiotemporal sequence data according to one embodiment of the present invention.
FIG. 2 is a table showing spatiotemporal sequence data according to one embodiment of the present invention.
Fig. 3 is a block diagram illustrating a modeling unit shown in fig. 1.
FIG. 4 is a schematic diagram illustrating spatio-temporal sequence data according to one embodiment of the present invention.
FIG. 5 is a schematic diagram illustrating a model according to one embodiment of the invention.
FIG. 6 is a schematic diagram illustrating spatio-temporal sequence data according to one embodiment of the present invention.
FIG. 7 is a diagram illustrating the construction of a spatial distance matrix according to one embodiment of the invention.
FIG. 8 is a diagram illustrating a model according to one embodiment of the invention.
FIG. 9 is a diagram illustrating a model according to one embodiment of the invention.
FIG. 10 is a schematic diagram illustrating a model according to one embodiment of the invention.
FIG. 11 is a flow diagram illustrating a method for classifying/predicting spatiotemporal sequence data according to another embodiment of the present invention.
Detailed Description
The principles and operation of the present invention will become apparent from the following description of specific embodiments thereof, taken in conjunction with the accompanying drawings. It should be noted that the present invention should not be limited to the specific embodiments described below. In addition, a detailed description of known technologies that are not related to the present invention is omitted for the sake of brevity.
FIG. 1 is a block diagram illustrating an apparatus 10 for classifying/predicting spatiotemporal sequence data according to one embodiment of the present invention. As shown in fig. 1, the apparatus 10 includes a receiving unit 110, a modeling unit 120, and a classification/prediction unit 130.
The receiving unit 110 receives space-time sequence data. In this application, spatio-temporal sequence data refers to a set of data having temporal and spatial characteristics. Spatio-temporal sequence data can be defined as:
Z={zs(t),s∈S,t∈T},
where s denotes spatial position, t denotes time, zs(T) represents an attribute value of a spatial position S at time T, S represents a spatial region relating to a spatio-temporal sequence, and T represents a time range relating to the spatio-temporal sequence. FIG. 2 is a table illustrating spatiotemporal sequence data, as exemplified by uniformly sampled grid data, in accordance with an embodiment of the present invention. For example, as shown in fig. 2, when s is (2, 4) and t is 1, z is 10, i.e., the value of the variable z at the 2 nd row and 4 th column position at the time t is 1 is 10. Therefore, the spatio-temporal sequence data is a time sequence set constructed by attribute values of all spatial units from 1 to T.
The modeling unit 120 generates weight parameters related to geographic heterogeneity based on the spatio-temporal sequence data, and constructs a model for classification/prediction based on the generated weight parameters. Hereinafter, the operation of the modeling unit 120 is described in detail with reference to fig. 3.
Fig. 3 is a block diagram illustrating the modeling unit 120 shown in fig. 1. As shown in fig. 3, the modeling unit 120 includes a spatial distance matrix construction subunit 1210, a spatial weight matrix construction subunit 1220, a model output weight value operator unit 1230, and a selection subunit 1240.
The spatial distance matrix construction subunit 1210 constructs a spatial distance matrix M based on the spatio-temporal sequence data. Referring to fig. 4, a number n-6 of spatio-temporal sequence data is shown, assuming that each time sequence has a length k. Accordingly, the spatial distance matrix M is:
wherein the content of the first and second substances,representing the distance between points i and j
The spatial weight matrix construction subunit 1220 selects at least two different weight parameters for each spatial position, respectively, to construct a corresponding spatial weight matrix. For example, the spatial weight matrix construction subunit 1220 may construct a gaussian spatial weight matrix W under different weight parameters c for spatial position i:
wherein, wij=exp(-(dij/c))
c∈{c1,c2,…,cmD, the value range of c is the minimumijAnd maximum dijIn the meantime.
Since each spatial position is observed k times, a mixed spatial weight matrix can be further defined for the convenience of subsequent calculation
Preferably, the spatial distance matrix may be a gaussian spatial distance matrix, and the spatial weight matrix may be a gaussian spatial weight matrix.
The model output weight value operator unit 1230 calculates the corresponding model output weight value based on the corresponding spatial weight matrix. First, the model output weight value operator unit 1230 obtains the input weights and biases of the model, which may be randomly given. FIG. 5 is a schematic diagram illustrating a model according to one embodiment of the invention. As can be seen in FIG. 5, the input weights of the model include a1、a2、…a1The bias of the model includes b1、b2、…b1. The input weights and offsets may be obtained by reference to a conventional ELM algorithm. In addition, the model output weight value operator unit 1230 calculates the hidden layer input matrix H:
Hnk×lβl=onk
on the basis, the model output weight value calculation operator unit 1230 solves the weighted least square norm solution of the output weight value of the modelThe following cost function is minimized:
Wherein Z is the true value and o is the model output value.
The selection subunit 1240 selects the output weight corresponding to the weight parameter that yields the minimum cost value as the final model output weight. That is, the selection subunit 1240 selects the model corresponding to the weight parameter c corresponding to the best fit as the final model of the spatial position i. For example, for spatial position i, different weight parameters c are calculated1,c2,…,cmMinimum CV value corresponding to (1):
and selects the final model accordingly.
Returning to fig. 1, the classification/prediction unit 130 classifies/predicts the time-space sequence data using the model constructed by the modeling unit 120. Finally, the results of the classification or prediction, such as air quality at a future time, weather conditions, etc., are output.
An example of a particular application of the apparatus 10 shown in fig. 1 is described below in conjunction with fig. 6-10.
FIG. 6 is a schematic diagram illustrating spatio-temporal sequence data according to one embodiment of the present invention. For example, the spatio-temporal sequence data may be:
that is, there are six positions p1 to p6 in space, and each position corresponds to a time series of length 6 (T ═ 1-6). There are two eigenvectors x1, x2 for each position, assuming that their values are the same for all positions:
accordingly, the spatial distance matrix construction subunit 1210 may construct the spatial distance matrix M in the manner shown in fig. 7:
the spatial position i is then equal to 3 (p)3) For example, how to build a GWELM model is illustrated.
For spatial position p3The weight parameter c is respectively selected to be different c1,c2,…,cm. Preferably, the weight parameter is linearly selected in the range from a minimum distance to a maximum distance:
cmin=min(dij),cmax=max(dij) (i=1,…,n;j=1,…,n;i≠j)
where m represents the number of weight parameters. The advantage of this selection strategy is that the optimal solution can be obtained faster.
Alternatively, other strategies may be employed to select the weight parameter c. For example, the weight parameter c may be selected in a random manner from a range of minimum distances to maximum distances.
Assuming that m is 6, the corresponding weight parameter c is calculated as follows:
cmin=1.4,cmax=3.6
c1=1.40;c2=1.84;c3=2.28;c4=2.72;c5=3.16;c6=3.60
the spatial weight matrix construction subunit 1220 constructs gaussian spatial weight matrices under different weight parameters for the spatial position i equal to 3:
Wi,c=diag(wi1,wi2,…,wi6)
wherein, wij=exp(-(dij/c)2)
When c1 is 1.40, W3,1.40=diag(0.08,0.36,1,0.13,0.36,0.36)
When c2 is 1.84, W3,1.84=diag(0.23,0.55,1,0.31,0.55,0.55)
When c3 is 2.28, W3,2.28=diag(0.38,0.68,1,0.46,0.68,0.68)
When c4 is 2.72, W3,2.72=diag(0.51,0.76,1,0.58,0.76,0.76)
When c5 is 3.16, W3,3.16=diag(0.61,0.82,1,0.67,0.82,0.82)
When c6 is 3.60, W3,3.60=diag(0.68,0.86,1,0.73,0.86,0.86)
FIG. 8 is a diagram illustrating a model in which the input weights and biases for the model are randomly given, according to one embodiment of the invention:
a1=[0.8147;0.9058]
a2=[0.1270;0.9134]
a3=[0.6324;0.0975]
b=[0.2785;0.5469;0.9575]
on this basis, the model output weight value operator unit 1230 calculates the hidden layer input matrix H:
the result of the calculation is
Since the variable x is the same for each position point, it is possible to reduce the number of the position points
Then, the model output weight calculation subunit 1230 solves the output weights of the models under different weight parameters:
cv=3.96
cv=4.33
cv=4.59
cv=4.77
cv=4.89
cv=4.97
it can be seen that for spatial position p3When c is 1.40, the corresponding CV value is the smallest. Therefore, the selection subunit 1240 selects the estimated value corresponding to the minimum CV as the output weight of the prediction model:
thereby, a position p for space as shown in fig. 9 can be obtained3The GWELM model of (1).
For spatial position p3When the value at the time t-7 needs to be predicted, the values of the variables x1 and x2 at the time t-7 need to be obtained first. Assuming that x1(7) is 4 and x2(7) is 6, the model shown in fig. 9 is substituted to obtain the final predicted valueAs shown in fig. 10.
Therefore, by adopting the geographical weighting extreme learning machine model provided by the embodiment of the invention, the accuracy of classifying or predicting the time-space sequence data can be improved.
FIG. 11 is a flow diagram illustrating a method for classifying/predicting spatiotemporal sequence data according to another embodiment of the present invention. As shown in fig. 11, the method 1100 begins at step S1110.
In step S1120, spatio-temporal sequence data is received.
In step S1130, weight parameters related to geographic heterogeneity are generated based on the spatio-temporal sequence data. Then, at step S1340, a model for classification/prediction is constructed based on the generated weight parameters. Preferably, a spatial distance matrix is constructed based on the spatio-temporal sequence data, and at least two different weight parameters are respectively selected for each spatial position to construct a corresponding spatial weight matrix. Then, the corresponding model output weights are calculated based on the corresponding spatial weight matrix. And finally, selecting the output weight value corresponding to the weight parameter generating the minimum cost value as the final model output weight value.
The at least two different weight parameters may be selected from a range of minimum spatial distances to maximum spatial distances. Preferably, the at least two different weight parameters may be selected linearly from a range of minimum spatial distance to maximum spatial distance. Alternatively, the at least two different weight parameters may be randomly selected from a range of minimum spatial distance to maximum spatial distance.
In step S1150, spatiotemporal sequence data is classified/predicted using the constructed model for classification/prediction.
Finally, the method 1100 ends at step S1160.
It should be understood that the above-described embodiments of the present invention can be implemented by software, hardware, or a combination of both software and hardware. For example, various components within the systems in the above embodiments may be implemented by a variety of devices, including but not limited to: analog circuits, digital circuits, general purpose processors, Digital Signal Processing (DSP) circuits, programmable processors, Application Specific Integrated Circuits (ASIC), Field Programmable Gate Arrays (FPGA), programmable logic devices (CPLD), and the like.
In addition, those skilled in the art will appreciate that the initial parameters described in the embodiments of the present invention may be stored in a local database, may be stored in a distributed database, or may be stored in a remote database.
Furthermore, embodiments of the invention disclosed herein may be implemented on a computer program product. More specifically, the computer program product is one of the following: there is a computer readable medium having computer program logic encoded thereon that, when executed on a computing device, provides related operations for implementing the above-described aspects of the present invention. When executed on at least one processor of a computing system, the computer program logic causes the processor to perform the operations (methods) described in embodiments of the present invention. Such arrangements of the invention are typically provided as downloadable software images, shared databases, etc. arranged or encoded in software, code and/or other data structures on a computer readable medium such as an optical medium (e.g., CD-ROM), floppy or hard disk or other medium such as firmware or microcode on one or more ROM or RAM or PROM chips or in one or more modules. The software or firmware or such configurations may be installed on a computing device to cause one or more processors in the computing device to perform the techniques described in embodiments of the present invention.
Although the present invention has been described in conjunction with the preferred embodiments thereof, it will be understood by those skilled in the art that various changes, substitutions and alterations can be made herein without departing from the spirit and scope of the invention. Accordingly, the present invention should not be limited by the above-described embodiments, but should be defined by the appended claims and their equivalents.
Claims (10)
1. An apparatus for classifying/predicting spatio-temporal sequence data, comprising:
a receiving unit configured to receive spatio-temporal sequence data;
a modeling unit configured to generate weight parameters related to geographic heterogeneity based on spatio-temporal sequence data and construct a model for classification/prediction based on the generated weight parameters, the modeling unit comprising:
a spatial distance matrix construction subunit configured to construct a spatial distance matrix based on the spatio-temporal sequence data;
a spatial weight matrix construction subunit configured to select at least two different weight parameters for each spatial position, respectively, to construct a corresponding spatial weight matrix;
a model output weight calculation subunit configured to calculate respective model output weights based on the respective spatial weight matrices; and
a selecting subunit configured to select an output weight corresponding to the weight parameter that produces the minimum cost value as a final model output weight;
and
and the classification/prediction unit is configured to classify/predict the spatio-temporal sequence data by adopting the constructed model for classification/prediction.
2. The apparatus of claim 1, wherein the spatial distance matrix is a gaussian spatial distance matrix and the spatial weight matrix is a gaussian spatial weight matrix.
3. The apparatus of claim 1, wherein the at least two different weight parameters are selected from a range of minimum to maximum spatial distances.
4. The apparatus of claim 3, wherein the at least two different weight parameters are linearly selected from a range of minimum to maximum spatial distances.
5. The apparatus of claim 3, wherein the at least two different weight parameters are randomly selected from a range of minimum to maximum spatial distances.
6. A method for classifying/predicting spatio-temporal sequence data, comprising:
receiving space-time sequence data;
generating weight parameters related to geographic heterogeneity based on the spatio-temporal sequence data, and constructing a model for classification/prediction based on the generated weight parameters, wherein the model for classification/prediction is constructed by:
constructing a spatial distance matrix based on the space-time sequence data;
respectively selecting at least two different weight parameters aiming at each spatial position to construct a corresponding spatial weight matrix;
calculating respective model output weights based on the respective spatial weight matrices; and
selecting an output weight value corresponding to the weight parameter generating the minimum cost value as a final model output weight value;
and
and classifying/predicting the space-time sequence data by using the constructed model for classifying/predicting.
7. The method of claim 6, wherein the spatial distance matrix is a Gaussian spatial distance matrix and the spatial weight matrix is a Gaussian spatial weight matrix.
8. The method of claim 6, wherein the at least two different weight parameters are selected from a range of minimum to maximum spatial distances.
9. The method of claim 8, wherein the at least two different weight parameters are selected linearly from a range of minimum to maximum spatial distances.
10. The method of claim 8, wherein the at least two different weight parameters are randomly selected from a range of minimum spatial distance to maximum spatial distance.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510690684.7A CN106610980B (en) | 2015-10-22 | 2015-10-22 | Apparatus and method for classifying/predicting spatio-temporal sequence data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510690684.7A CN106610980B (en) | 2015-10-22 | 2015-10-22 | Apparatus and method for classifying/predicting spatio-temporal sequence data |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106610980A CN106610980A (en) | 2017-05-03 |
CN106610980B true CN106610980B (en) | 2022-03-18 |
Family
ID=58610345
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510690684.7A Active CN106610980B (en) | 2015-10-22 | 2015-10-22 | Apparatus and method for classifying/predicting spatio-temporal sequence data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106610980B (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101127029A (en) * | 2007-08-24 | 2008-02-20 | 复旦大学 | Method for training SVM classifier in large scale data classification |
CN102110365A (en) * | 2009-12-28 | 2011-06-29 | 日电(中国)有限公司 | Road condition prediction method and road condition prediction system based on space-time relationship |
JP2014041547A (en) * | 2012-08-23 | 2014-03-06 | Nippon Telegr & Teleph Corp <Ntt> | Time series data analysis device, method and program |
CN103700255A (en) * | 2013-12-30 | 2014-04-02 | 复旦大学 | Time and space related data mining-based traffic flow prediction method |
-
2015
- 2015-10-22 CN CN201510690684.7A patent/CN106610980B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101127029A (en) * | 2007-08-24 | 2008-02-20 | 复旦大学 | Method for training SVM classifier in large scale data classification |
CN102110365A (en) * | 2009-12-28 | 2011-06-29 | 日电(中国)有限公司 | Road condition prediction method and road condition prediction system based on space-time relationship |
JP2014041547A (en) * | 2012-08-23 | 2014-03-06 | Nippon Telegr & Teleph Corp <Ntt> | Time series data analysis device, method and program |
CN103700255A (en) * | 2013-12-30 | 2014-04-02 | 复旦大学 | Time and space related data mining-based traffic flow prediction method |
Non-Patent Citations (2)
Title |
---|
Decomposing environmental,spatial,and spatiotemporal components of species distributions;Torsten Hothorn等;《ECOLOGICAL MONOGRAPHS》;20110501;329-347页 * |
基于空间自回归和地理加权回归模型的不同尺度下土地利用程度研究;郭龙等;《华中农业大学学报》;20130620;48-53页 * |
Also Published As
Publication number | Publication date |
---|---|
CN106610980A (en) | 2017-05-03 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109492830B (en) | Mobile pollution source emission concentration prediction method based on time-space deep learning | |
US11250308B2 (en) | Apparatus and method for generating prediction model based on artificial neural network | |
CN113313947B (en) | Road condition evaluation method of short-term traffic prediction graph convolution network | |
US20210064999A1 (en) | Multi-scale multi-granularity spatial-temporal traffic volume prediction | |
CN110232444B (en) | Geological monitoring BP neural network optimization method, device, equipment and storage medium | |
CN109977098A (en) | Non-stationary time-series data predication method, system, storage medium and computer equipment | |
CN112437451B (en) | Wireless network flow prediction method and device based on generation countermeasure network | |
US11921697B2 (en) | Methods and systems for detecting spurious data patterns | |
CN110188263B (en) | Heterogeneous time interval-oriented scientific research hotspot prediction method and system | |
CN114565124A (en) | Ship traffic flow prediction method based on improved graph convolution neural network | |
CN108764541B (en) | Wind energy prediction method combining space characteristic and error processing | |
CN113869521A (en) | Method, device, computing equipment and storage medium for constructing prediction model | |
CN114697128A (en) | Big data denoising method and big data acquisition system through artificial intelligence decision | |
Tang et al. | Short-term travel speed prediction for urban expressways: Hybrid convolutional neural network models | |
CN111385037A (en) | Real-time prediction method of indoor available frequency spectrum | |
CN103793438A (en) | MapReduce based parallel clustering method | |
Falahat et al. | Maize tassel detection and counting using a YOLOv5-based model | |
CN111105127B (en) | Modular product design evaluation method based on data driving | |
CN106610980B (en) | Apparatus and method for classifying/predicting spatio-temporal sequence data | |
WO2020261448A1 (en) | Prediction device, prediction method, and prediction program | |
Attanasi et al. | A hybrid method for real-time short-term predictions of traffic flows in urban areas | |
CN111199306A (en) | Prediction device, prediction method, and storage medium | |
CN117036037B (en) | Suspicious transaction risk analysis method and suspicious transaction risk analysis device | |
CN113434295B (en) | Farmland monitoring method, device, equipment and storage medium based on edge calculation | |
CN111785329B (en) | Single-cell RNA sequencing clustering method based on countermeasure automatic encoder |
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 |