CN110189227A - Insider trading discriminating conduct based on principal component analysis and reverse transmittance nerve network - Google Patents
Insider trading discriminating conduct based on principal component analysis and reverse transmittance nerve network Download PDFInfo
- Publication number
- CN110189227A CN110189227A CN201910472187.8A CN201910472187A CN110189227A CN 110189227 A CN110189227 A CN 110189227A CN 201910472187 A CN201910472187 A CN 201910472187A CN 110189227 A CN110189227 A CN 110189227A
- Authority
- CN
- China
- Prior art keywords
- principal component
- insider trading
- data set
- neural network
- component analysis
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000513 principal component analysis Methods 0.000 title claims abstract description 35
- 210000005036 nerve Anatomy 0.000 title claims abstract description 33
- 238000002834 transmittance Methods 0.000 title claims abstract description 33
- 238000013528 artificial neural network Methods 0.000 claims abstract description 33
- 238000012360 testing method Methods 0.000 claims abstract description 26
- 230000009467 reduction Effects 0.000 claims abstract description 9
- 238000012549 training Methods 0.000 claims description 21
- 238000000034 method Methods 0.000 claims description 14
- 239000002075 main ingredient Substances 0.000 claims 1
- 210000004218 nerve net Anatomy 0.000 claims 1
- 238000003062 neural network model Methods 0.000 abstract 1
- 230000008707 rearrangement Effects 0.000 description 8
- 239000011159 matrix material Substances 0.000 description 6
- 230000000875 corresponding effect Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000002596 correlated effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000005284 excitation Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000033228 biological regulation Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000012850 discrimination method Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000006798 recombination Effects 0.000 description 1
- 238000005215 recombination Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000005303 weighing Methods 0.000 description 1
Classifications
-
- 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
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/04—Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange
Abstract
The insider trading discriminating conduct based on principal component analysis and reverse transmittance nerve network that the invention discloses a kind of obtains the principal component of the characteristic index collection of insider trading using principal component analysis dimensionality reduction；Insider trading neural network recognization model is constructed, is trained using principal component data set；Obtain test target, using the principal component data set of test target as the input of neural network recognization model, through neural network recognization model obtain whether the result of insider trading.The present invention is modeled using elastic reverse transmittance nerve network for the non-linear relation of insider trading and characteristic index, neural network model is trained using the related data of securities market, the elastic back propagation artificial neural network model of foundation can the insider trading to securities market accurately identified.
Description
Technical field
The invention belongs to Securities Market Regulation fields, and in particular to one kind is based on principal component analysis and backpropagation neural network
The insider trading discriminating conduct of network.
Background technique
With the continuous development of China's securities market, panoramic swindle also generates therewith.And it is wherein interior
Curtain trading activity quantity is larger, makes a very bad impression to security justice of exchange.The increasingly diversification of insider trading main body composition, its friendship
Easy form has the characteristics that diversification, hiddenization, and supervision department is made to be difficult efficiently to be rapidly performed by accurately identifying for insider trading.
In China stock market insider trading case, rearrangement of assets type insider trading quantity accounting is larger, caused by negatively affect
Also larger.The insider trading discriminating conduct based on principal component analysis and reverse transmittance nerve network that the invention proposes a kind of.
Principal Component Analysis PCA (Principal Component Analysis) is a kind of linear dimension reduction method, it is special
Not Shi He sample data concentrate include more feature, and there is a situation where between these features each other redundancy correlation.It is main at
Point analytic approach can represent the main feature component of the main variation of primitive character collection by the way that feature set is reduced into sub-fraction, come
Realize mapping of the high dimensional data to low-dimensional data space.
Elastic reverse transmittance nerve network is the error feedback learning network based on local auto-adaptive adjustable strategies.It is elastic anti-
Changed to the back-propagation algorithm of Propagation Neural Network according to error energy functional gradient and assigns weight with offset individually to adjust
Synchronizing is long.Compared with using the standard BP of global adaptation strategy (feedforward reverse neural network) algorithm, which does not use biography for this
System error gradient size feedback system, but by error in reading gradient signs, the step-length adjustable strategies of variation, stablize independently
The weight and deviation for adjusting each layer of network, since didactic training mode is utilized in elastic reverse transmittance nerve network algorithm,
And the adaptivity of network parameter is high, so that algorithm while reducing network convergence time, reduces operator by not
Disconnected test obtains the program of optimal value.
Summary of the invention
It is larger the purpose of the present invention is being directed to the rearrangement of assets type insider trading quantity of securities market, market is caused badly
It influences and is not easy identified problem, provide a kind of based on principal component analysis and the insider trading of reverse transmittance nerve network discrimination
Method filters out the principal component of rearrangement of assets type insider trading correlated characteristic index, to the insider trading of rearrangement of assets type and feature
The non-linear relation of index is modeled, and the insider trading identification model of foundation can carry out the insider trading of securities market efficient
And it accurately distinguishes.
The technical scheme is that the insider trading discriminating conduct based on principal component analysis and reverse transmittance nerve network,
Include the following steps,
Step 1: obtaining the rearrangement of assets type insider trading sample data set under the different time window phase；
Step 2: using principal component analysis dimensionality reduction, obtain the principal component of characteristic index collection and the number of principal components of sample data set
According to collection；
Step 3: building insider trading neural network recognization model is trained using principal component data set；
Step 4: obtaining test target, handled and surveyed according to data set of the principal component of characteristic index collection to test target
Try the principal component data set of target；
Step 5: using the principal component data set of test target as the input of neural network recognization model, knowing through neural network
Other model show whether policy goals are identified as insider trading；
Step 6: whether subsequent supervision verifying recognition result is correct；
Step 6.1: if recognition result is correct, thening follow the steps 8；
Step 6.2: if recognition result is incorrect, thening follow the steps 7；
Step 7: sample data set is added by test target data set and with the presence or absence of the result of insider trading, and training is more
New neural network recognization model；
Step 8: judging whether there is next test target；
Step 8.1: if there is next test target, thening follow the steps 4；
Step 8.2: if terminating without next test target.
Further, the neural network is elastic reverse transmittance nerve network.
Further, in the step 2 of the insider trading discriminating conduct based on principal component analysis and reverse transmittance nerve network,
The principal component analysis dimensionality reduction determines number of principal components using Kaiser standard, i.e. principal component of the selection variance greater than 1 is as most
Mode input data set after final decline dimension.
Further, in the step 3 of the insider trading discriminating conduct based on principal component analysis and reverse transmittance nerve network,
The training method of elastic reverse transmittance nerve network, includes the following steps,
Step 1: initialization neural network parameter, to each weight assign initial change value, setting accelerated factor, deceleration because
Son, the weight variation upper limit, weight change lower limit, setting study number；
Step 2: to learning, calculate and save each layer error energy functional gradient before neural network；
Step 3: calculating the error criterion of excitation output and desired output vector；
Step 4: whether error in judgement index is less than setting index；
Step 4.1: if error criterion is less than setting index, thening follow the steps 9；
Step 4.2: if error criterion thens follow the steps 5 not less than setting index；
Step 5: judging to learn whether number reaches setting number；
Step 5.1: if study number reaches setting number, thening follow the steps 9；
Step 5.2: if study number does not reach setting number, thening follow the steps 6；
Step 6: calculating network variable element to the first-order partial derivative of network error, and calculate power updated value；
Step 7: calculating the adjustment amount of network variable element；
Step 8: network variable element being adjusted, step 2 is executed；
Step 9: terminating training, save neural network weight and offset.
For the marketing enterprises sample that rearrangement of assets occurs, the present invention provides one kind to be passed based on principal component analysis and reversely
The insider trading discriminating conduct for broadcasting neural network, using Principal Component Analysis filter out insider trading correlated characteristic index it is main at
Point, it is modeled using elastic reverse transmittance nerve network for the non-linear relation of insider trading and characteristic index, utilizes security city
The related data of field is trained model, and the elastic back propagation artificial neural network model of foundation can carry out assets to securities market
Whether the enterprise of recombination occurs the discrimination of insider trading progress efficiently and accurately.
Detailed description of the invention
Present invention will be further explained below with reference to the attached drawings and examples.
Fig. 1 is the flow chart of the insider trading discriminating conduct based on principal component analysis and reverse transmittance nerve network.
Fig. 2 is the structural schematic diagram of elastic reverse transmittance nerve network.
Fig. 3 is the flow chart of the training method of elastic reverse transmittance nerve network.
Specific embodiment
As shown in Figure 1, the insider trading discriminating conduct based on principal component analysis and reverse transmittance nerve network, including accept
Step, step 1: obtaining the corresponding difference of stock sample that insider trading occurs during rearrangement of assets that stock supervisory committee announces
Securities market Microscopic Indexes, Corporate Finance and improvement index under event time window phase；Obtain the different event time window phase
Under insider trading sample data set；
Step 2: carrying out the dimensionality reduction of characteristic using principal component analytical method, obtain the principal component and sample of characteristic index collection
The principal component data set of notebook data collection；
Step 3: building insider trading neural network recognization model is trained using principal component data set；
Step 4: obtaining test target, handled and surveyed according to data set of the principal component of characteristic index collection to test target
Try the principal component data set of target；
Step 5: using the principal component data set of test target as the input of neural network recognization model, knowing through neural network
Other model obtain whether the result of insider trading；
Step 6: whether the recognition result of subsequent verifying model is correct；
Step 6.1: if recognition result is correct, thening follow the steps 8；
Step 6.2: if recognition result is incorrect, thening follow the steps 7；
Step 7: by test target data set and whether sample data set is added in the result of insider trading, and training updates this
Invent the identification model of the combination principal component analysis proposed and elastic reverse transmittance nerve network；
Step 8: judging whether there is next test target；
Step 8.1: if there is next test target, thening follow the steps 4；
Step 8.2: if terminating without next test target.
The principal component analysis of step 2 in insider trading discriminating conduct based on principal component analysis and reverse transmittance nerve network
Method uses orthogonal transformation by the Feature Conversion being relative to each other for principal component, to determine variance trend using these principal components,
Principal Component Analysis specifically includes following steps,
Step 1: calculating average vectorWherein xiIndicate the feature of i-th of sample point, n is total of sample
Number；
Step 2: calculating covariance matrix
Step 3: calculating feature vectorAnd corresponding characteristic value；
Step 4: k feature vector before sorting and selecting to feature vector；
Step 5: one d × k dimensional feature vector matrix U of building, d are the original number of dimensions of feature in data set, and k is characterized
The quantity of vector；
Step 6: according to equation y=UTData sample is converted to new subset by x.
In principal component analysis, characteristic value is equal to the variance corresponded in dimension on the coordinate of data after rotation, big
It is small to represent corresponding feature vector after matrix orthogonalization the percentage contribution of entire matrix is usually described with characteristic value
The information content for including in character pair vector direction, and the ratio between a certain characteristic value and all characteristic values then represent this feature vector
Variance contribution ratio, i.e., the ratio of the information content contained under the dimension.In this example, it is calculated in each data set first
Then the corresponding variance of each principal component chooses principal component of the variance greater than 1 as the data set after principal component dimensionality reduction again
The final mask input data set arrived.
Elastic reverse transmittance nerve network algorithm abbreviation RPROP (Resilient backpropagation) algorithm, also known as
For elastic BP (feedforward reverse neural network) algorithm.This is that one kind realizes that batch processing has tutor in Multilayer Feedforward Neural Networks training
The local auto-adaptive learning algorithm of habit.Elastic reverse transmittance nerve network algorithm is different from BP and calculates when being adjusted to network
It is that it will be adjusted network by the way that one " power updated value Δ " is in addition arranged in place of method algorithm, and network weight is repaired
Positive value delta w is then calculated by power updated value Δ, to avoid the inherent limitation of basic BP algorithm.
For elastic reverse transmittance nerve network as shown in Fig. 2, comprising input layer, hidden layer, output layer, X is input vector, C
For output response vector, W is that input layer implies weight matrix, and V is hidden layer to output layer weight matrix.
The objective function of neural network is output layer error energy function:
Y is desired output vector in formula.
As shown in figure 3, the training method of elastic reverse transmittance nerve network, includes the following steps,
Step 1: initialization neural network parameter assigns initial value Δ (0) to each weight, accelerated factor η is arranged+, subtract
Fast factor η-, weight change upper limit Δmax, variation lower limit Δmin, and set study number；
Step 2: to learning, calculate and save each layer error energy functional gradient before neural network；
Step 3: calculating the error criterion of excitation output and desired output vector；
Step 4: whether error in judgement index is less than setting index；
Step 4.1: if error criterion is less than setting index, thening follow the steps 9；
Step 4.2: if error criterion thens follow the steps 5 not less than setting index；
Step 5: judging to learn whether number reaches setting study number；
Step 5.1: if study number reaches setting study number, thening follow the steps 9；
Step 5.2: if study number does not reach setting study number, thening follow the steps 6；
Step 6: calculating network variable element to the first-order partial derivative of network error, and calculate power updated value；
Step 7: calculating the adjustment amount of network variable element；
Step 8: network variable element being adjusted, step 2 is executed；
Step 9: terminating training, save neural network weight and offset.
In the step 6 of the training method of elastic reverse transmittance nerve network, calculating network variable element misses network first
The first-order partial derivative of difference, then calculate power updated value.Wherein, the calculation formula for weighing updated value is as follows:
Wherein, E is error function, and Δ w represents the correction value of network weight, and Δ representation updated value, t is frequency of training.
Elastic reverse transmittance nerve network algorithm uses batch processing training method,For all moulds on the t times training set when trained
Formula corresponds to the sum of adding up for gradient, and i, j respectively indicate j-th of node of i-th of node of input layer and hidden layer, η+Increase for power updated value
Big multiple；η-Reduce multiple for power updated value；T is frequency of training.Wherein, 0 < η-1 < η of <+。
In step 7, the calculation formula of the adjustment amount of network variable element is as follows:
Wherein t is frequency of training.Elastic reverse transmittance nerve network algorithm uses batch processing training method,For t
All modes when secondary trained on training set correspond to the sum of adding up for gradient, and i, j respectively indicate i-th of node of input layer and hidden layer
J-th of node.
In step 8, to network variable element wijThe calculation formula being adjusted is as follows:
In embodiment, inside story occurs during rearrangement of assets between acquisition stock supervisory committee announces first 2001 to 2015 years and hands over
Easy stock sample is 41 total.
It is more comprehensive in terms of subsidiary company equity structure, financial data, improvement system and securities market Microscopic respectively
26 characteristic indexs must be calculated or have chosen, as shown in Table 1.
One characteristic index table of table
With with same type inside news sensitive event, same industry, the same time, to be not affected by punishment former to choose
The white sample then chosen and insider trading stock sample equivalent amount occurs.
Using Principal Component Analysis dimensionality reduction by characteristic index data normalization.Event time window phase was 30 day of trade
Insider trading characteristic index data set obtains 9 principal components, event time window phase after carrying out Principal Component Analysis dimension-reduction treatment
To obtain 9 principal components after the insider trading characteristic index data set progress Principal Component Analysis dimension-reduction treatment of 60 day of trade,
Event time window phase is after the insider trading characteristic index data set of 90 day of trade carries out Principal Component Analysis dimension-reduction treatment
Obtain 10 principal components.
In order to detect recognition effect of the insider trading discriminating conduct proposed by the present invention on unknown stock sample, will be received
The data set of collection is in training set: forecast set=6:4 ratio is divided into training set and forecast set, trains a packet using training set
Include three layers of elastic back propagation artificial neural network model of a hidden layer.The network of multiple hidden layers is although be easier to learn, but be easier to
In sunken people's local minimum, after comprehensively considering, the present embodiment is using three layers with a hidden layer i.e. neural network.So
Afterwards, the recognition detection of insider trading is carried out as the test input of elastic back propagation artificial neural network model using forecast set.
Insider trading totality recognition correct rate when event time window phase was 30 day of trade is 75.76%, such as table
Shown in two.
Two event time window phase of table is the insider trading recognition result table of 30 day of trade
Event time window phase is that the insider trading totality recognition correct rate of 60 day of trade is 72.73%, such as three institute of table
Show.
Three event time window phase of table is the insider trading recognition result table of 60 day of trade
Event time window phase is that the insider trading totality recognition correct rate of 90 day of trade is 84.85%, such as four institute of table
Show.
Four event time window phase of table is the insider trading recognition result table of 90 day of trade
Finally, the method for the present invention is compared in the case where event time window phase is set as 30,60,90 days respectively and is not carried out
The insider trading discrimination of the elastic reverse transmittance nerve network method of principal component analysis, as shown in Table 5.It can find, at three kinds
There is the insider trading discrimination of elastic reverse transmittance nerve network under the time window phase, after progress principal component analysis significantly mentions
It is high.
The insider trading recognition result contrast table of two methods under five or three kinds of event time window phases of table
Claims (3)
1. the insider trading discriminating conduct based on principal component analysis and reverse transmittance nerve network, which is characterized in that including following
Step,
Step 1: obtaining the insider trading sample data set under the different time window phase；
Step 2: using principal component analysis dimensionality reduction, obtain the principal component of characteristic index collection and the number of principal components evidence of sample data set
Collection；
Step 3: building insider trading neural network recognization model is trained using principal component data set；
Step 4: obtaining test target, handled to obtain test mesh according to data set of the principal component of characteristic index collection to test target
Target principal component data set；
Step 5: using the principal component data set of test target as the input of neural network recognization model, through neural network recognization mould
Type obtain whether the result of insider trading；
Step 6: whether subsequent supervision verifying recognition result is correct；
Step 6.1: if recognition result is correct, thening follow the steps 8；
Step 6.2: if recognition result is incorrect, thening follow the steps 7；
Step 7: by test target data set and whether sample data set is added in the result of insider trading, and training updates nerve net
Network identification model；
Step 8: judging whether there is next test target；
Step 8.1: if there is next test target, thening follow the steps 4；
Step 8.2: if terminating without next test target.
2. the insider trading discriminating conduct according to claim 1 based on principal component analysis and reverse transmittance nerve network,
It is characterized in that, the neural network is elastic reverse transmittance nerve network.
3. the insider trading discrimination side according to claim 1 or 2 based on principal component analysis and reverse transmittance nerve network
Method, which is characterized in that in step 2, the principal component analysis dimensionality reduction chooses the biggish principal component of variance contribution degree so that it is determined that main
Ingredient number.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910472187.8A CN110189227A (en) | 2019-05-31 | 2019-05-31 | Insider trading discriminating conduct based on principal component analysis and reverse transmittance nerve network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910472187.8A CN110189227A (en) | 2019-05-31 | 2019-05-31 | Insider trading discriminating conduct based on principal component analysis and reverse transmittance nerve network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110189227A true CN110189227A (en) | 2019-08-30 |
Family
ID=67719714
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910472187.8A Pending CN110189227A (en) | 2019-05-31 | 2019-05-31 | Insider trading discriminating conduct based on principal component analysis and reverse transmittance nerve network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110189227A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110837856A (en) * | 2019-10-31 | 2020-02-25 | 深圳市商汤科技有限公司 | Neural network training and target detection method, device, equipment and storage medium |
| CN110956543A (en) * | 2019-11-06 | 2020-04-03 | 上海应用技术大学 | Method for detecting abnormal transaction |
-
2019
- 2019-05-31 CN CN201910472187.8A patent/CN110189227A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110837856A (en) * | 2019-10-31 | 2020-02-25 | 深圳市商汤科技有限公司 | Neural network training and target detection method, device, equipment and storage medium |
| CN110956543A (en) * | 2019-11-06 | 2020-04-03 | 上海应用技术大学 | Method for detecting abnormal transaction |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105488528B (en) | Neural network image classification method based on improving expert inquiry method | |
| CN107506941A (en) | A kind of enterprise in charge of construction's credit assessment method and system based on big data technology | |
| CN106023195A (en) | BP neural network image segmentation method and device based on adaptive genetic algorithm | |
| CN113538125A (en) | Risk rating method for optimizing Hopfield neural network based on firefly algorithm | |
| CN110009030A (en) | Sewage treatment method for diagnosing faults based on stacking meta learning strategy | |
| CN110321361A (en) | Examination question based on improved LSTM neural network model recommends determination method | |
| CN105931116A (en) | Automated credit scoring system and method based on depth learning mechanism | |
| CN110363230A (en) | Stacking integrated sewage handling failure diagnostic method based on weighting base classifier | |
| CN108009925A (en) | A kind of fund asset configuration system and method based on network clustering | |
| CN106327357A (en) | Load identification method based on improved probabilistic neural network | |
| CN110189227A (en) | Insider trading discriminating conduct based on principal component analysis and reverse transmittance nerve network | |
| CN111460001B (en) | Power distribution network theoretical line loss rate evaluation method and system | |
| CN112215446A (en) | Neural network-based unit dynamic fire risk assessment method | |
| CN103795592B (en) | Online water navy detection method and device | |
| CN107481135A (en) | A kind of personal credit evaluation method and system based on BP neural network | |
| CN112037011A (en) | Credit scoring method based on FOA-RBF neural network | |
| CN111104975B (en) | Credit evaluation method based on breadth learning | |
| CN116526450A (en) | Error compensation-based two-stage short-term power load combination prediction method | |
| CN109919374A (en) | Prediction of Stock Price method based on APSO-BP neural network | |
| CN114841492A (en) | Traditional village protection value evaluation method | |
| CN109408896A (en) | A kind of anerobic sowage processing gas production multi-element intelligent method for real-time monitoring | |
| CN111028086A (en) | Enhanced index tracking method based on clustering and LSTM network | |
| CN110533341A (en) | A kind of Livable City evaluation method based on BP neural network | |
| CN109711450A (en) | A kind of power grid forecast failure collection prediction technique, device, electronic equipment and storage medium | |
| CN108734116A (en) | A kind of face identification method learning depth autoencoder network based on speed change |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190830 |
|
| RJ01 | Rejection of invention patent application after publication |