CN109583659A - User's operation behavior prediction method and system based on deep learning - Google Patents
User's operation behavior prediction method and system based on deep learning Download PDFInfo
- Publication number
- CN109583659A CN109583659A CN201811493939.0A CN201811493939A CN109583659A CN 109583659 A CN109583659 A CN 109583659A CN 201811493939 A CN201811493939 A CN 201811493939A CN 109583659 A CN109583659 A CN 109583659A
- Authority
- CN
- China
- Prior art keywords
- user
- operation behavior
- data
- behavior prediction
- prediction system
- 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
- 230000006399 behavior Effects 0.000 title claims abstract description 157
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000013135 deep learning Methods 0.000 title claims abstract description 19
- 238000012549 training Methods 0.000 claims abstract description 43
- 238000005457 optimization Methods 0.000 claims abstract description 26
- 238000013528 artificial neural network Methods 0.000 claims abstract description 24
- 238000005516 engineering process Methods 0.000 claims abstract description 15
- 238000012360 testing method Methods 0.000 claims abstract description 13
- 238000011156 evaluation Methods 0.000 claims abstract description 5
- 230000007613 environmental effect Effects 0.000 claims description 23
- 230000006870 function Effects 0.000 claims description 7
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 3
- 230000008676 import Effects 0.000 claims description 3
- 230000009897 systematic effect Effects 0.000 claims description 3
- 230000009471 action Effects 0.000 claims description 2
- 230000003542 behavioural effect Effects 0.000 claims description 2
- 229910017435 S2 In Inorganic materials 0.000 claims 1
- 238000013139 quantization Methods 0.000 claims 1
- 239000011159 matrix material Substances 0.000 description 4
- 238000013461 design Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000006386 memory function Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000000306 recurrent effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000004218 nerve net Anatomy 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- 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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The user's operation behavior prediction method based on deep learning that the invention discloses a kind of, comprising: S1: user behavior data is extracted;S2: user behavior data is pre-processed;S3: pretreated user behavior data is divided into training set, verifying collection and test set in proportion;S4: creation user's operation behavior prediction system;S5: training set, verifying collection classification are imported into user's operation behavior prediction system, user's operation behavior prediction system is trained and is verified using window sliding technology;S6: test set is imported into the user's operation behavior prediction system after optimization, carries out performance evaluation;S7: completing training, obtains the user's operation behavior prediction system that optimization is completed.The present invention learns the inherent law of user's Continuous behavior using LSTM length Memory Neural Networks subsystem and using window sliding technology, learn the weight of external factor by full Connection Neural Network subsystem, is finally merged two subsystems to complete the prediction to user's operation behavior.
Description
Technical field
The present invention relates to soft project applied technical fields, in particular to a kind of user's operation based on deep learning
Behavior prediction method and system.
Background technique
With the rapid development of information technology, computer software using more and more extensive, while with user demand
Diversification, a large number of services system also generate therewith, and the big many body system of these systems is huge, function is complicated, cover hundreds and thousands of modules.
User needs to position multiple modules according to their own needs, but since system is excessively huge when operating these systems, finds simultaneously
Positioning service module is very time-consuming, poor user experience.If operation behavior that can be previous according to user, and combine influence operation behavior
External factor, using depth learning technology learning law, to recommend next module most possibly operated for user,
User operation time is saved, it is simple and convenient, substantially increase user experience.
The characteristics of full Connection Neural Network is that all neuron nodes of preceding layer are connected with all nodes of current layer,
That is for each node of n-th layer when being calculated, the input of activation primitive is the weighting of n-1 layers of all nodes, complete to connect
The structure of neural network such as Fig. 1.
It is not both its memory function that Recognition with Recurrent Neural Network (RNN) is maximum with other neural networks, and the network structure of RNN is such as
Fig. 2.
The hidden layer of RNN is not only passed to the output layer at the moment, while also passing down to the hidden layer of subsequent time, from
And realize memory function, therefore the hidden layer calculation formula of t moment:
St=f (U*Xt+W*St-1)
Wherein, StFor the output of t moment hidden layer, f () is activation primitive, and U is input layer parameter matrix, and W is hidden layer
Parameter matrix, St-1It is exported for the hidden layer of last moment.
Length Memory Neural Networks (LSTM) are one kind of Recognition with Recurrent Neural Network (RNN).LSTM is added on the basis of RNN
Forgetting door, input gate, out gate abandon partial information for control system wherein forgeing door, and input gate is updated for determining
Information, out gate then determines the output result of hidden layer.LSTM ratio RNN has further optimization, right by the door of addition
Information is selectively remembered.The detailed construction of LSTM hidden layer such as Fig. 3.
Currently, most user's operation behavior prediction method, be all based on the frequency of the historical operation behavior of user into
Row statistics is to predict.But this method does not account for the inherent law that user operates continuously behavior, does not account for outside others yet
Influence of portion's factor for active user's behavior.
Summary of the invention
It is an object of that present invention to provide a kind of the user's operation behavior prediction method and system based on deep learning, utilization
LSTM length Memory Neural Networks subsystem and the inherent law for being learnt user's Continuous behavior using window sliding technology, are passed through
Full Connection Neural Network subsystem learns the weight of external factor, is finally merged two subsystems to complete to user
The prediction of operation behavior.
To reach above-mentioned purpose, in conjunction with Fig. 5, the present invention proposes a kind of user's operation behavior prediction side based on deep learning
Method, which comprises
S1: extracting user behavior data from historical operation log, and the user behavior data includes at least environmental data
With Continuous behavior data, environmental data includes multiple external environmental factors when data occur, and Continuous behavior data include at least
The operation behavior sequence of user and the time of origin of operation behavior.
Further, the external environmental factor include company ID, department ID, User ID, behavior occur period in
It is several or whole.
For example, environmental data includes the generation source of data, such as company ID, department ID, User ID etc. also includes behavior
Period of right time, such as the morning or afternoon, if festivals or holidays etc..Continuous behavior data, the operation behavior sequence columns and rows including user
For the specific time of generation.
S2: user behavior data is pre-processed.
In further embodiment, in step S2, carrying out pretreated method to user behavior data includes:
S21: the noise data unrelated with business is rejected.
S22: executing normalized to remaining environmental data, carries out pre-training to remaining Continuous behavior data with reality
Existing vectorization.
In further embodiment, the method also includes:
Word2vector technology is used to carry out pre-training to Continuous behavior data to realize vectorization.
In some instances, data are pre-processed and is also had the effect that expand full Connection Neural Network subsystem
There are interfaces for system, if the later period finds holiday influence factor, can directly connect input after factor vectorization, the regularization entirely
Neural network is connect, subsystem can be incorporated.
S3: pretreated user behavior data is divided into training set, verifying collection and test set in proportion.
For example, the data of random division 60% are used for training set, 20% data are used for for verifying collection, 20% data
Test set.
S4: creation user's operation behavior prediction system, the user's operation behavior prediction system includes with operation behavior sequence
Column, operation behavior time of origin be the deep layer LSTM subsystem of parameter and neural as the full connection of parameter using external environmental factor
Network subsystem.
S5: training set, verifying collection classification are imported into user's operation behavior prediction system, using window sliding technology to user
Operation behavior forecasting system is trained and verifies, to obtain the user's operation behavior prediction system after optimization.
In further embodiment, in step S5, training set, verifying collection classification are imported into user's operation behavior prediction system
System, is trained and is verified to user's operation behavior prediction system using window sliding technology, to obtain the user behaviour after optimization
Make the method for behavior prediction system further include:
By training set, verifying concentrate belong to Continuous behavior data part import deep layer LSTM subsystem, by training set,
The part for belonging to environmental data that verifying is concentrated imports full Connection Neural Network subsystem, and the two is spliced to seeking unity of action afterwards together
At least N wheel training, saves the smallest subsystem of penalty values as the subsystem after optimization.
In further embodiment, the N is equal to 50.
User's operation behavior prediction system mentioned by the present invention has used deep layer LSTM subsystem and full connection nerve net
Network Subsystem carries out splicing fusion.Continuous user behavior data is placed into deep layer LSTM subsystem, environmental data is placed
Into full Connection Neural Network subsystem, the two, which is spliced together, unifies training.50 wheel training (EPOCHS) have been carried out in total, have been protected
Deposit the smallest system of penalty values.
In further embodiment, the method also includes:
Selecting size is that 10 sliding window is trained and verifies to user's operation behavior prediction system.
Increase the data volume of user's operation behavior, the data mode present invention uses window sliding technology in training
Also the practical operation behavioral aspect of user can more be met.Fig. 4 illustrates the sliding type that window size is 4, and black window is intracorporal
Operation is used as tag along sort as training independent variable, the intracorporal operation of grey window.By repeatedly training test and comparison accuracy rate, most
Selected window size is 10 eventually.
S6: importing the user's operation behavior prediction system after optimization for test set, if the prediction of the user behavior of output
Accuracy rate is more than or equal to setting accuracy rate threshold value, enters step S7, otherwise, return step S5, the calculation formula of predictablity rate
It is as follows:
Wherein, n is the total sample number of predictive behavior, yi' it is predicted operation behavior serial number, yiIt is practical operation behavior serial number,
αiIt is the accuracy of i-th of predictive behavior.
S7: completing training, obtains the user's operation behavior prediction system that optimization is completed.
In further embodiment, the method also includes:
It is by the way of importing data in batches that training set, verifying collection classification importing user's operation behavior is pre- in step S5
Examining system.
In further embodiment, the method also includes:
In step S5, when to user's operation behavior prediction systematic training, the optimization algorithm used is adam optimization algorithm, damage
It loses function and selects cross entropy loss functions of classifying more.
Based on preceding method, the present invention further mentions a kind of user's operation behavior prediction system based on deep learning, described
System includes following module:
1) for extracting the module of user behavior data from historical operation log, the user behavior data is included at least
Environmental data and Continuous behavior data, environmental data include multiple external environmental factors when data occur, Continuous behavior data
Including at least the operation behavior sequence of user and the time of origin of operation behavior.
2) for carrying out pretreated module to user behavior data.
3) for pretreated user behavior data to be divided into the mould of training set, verifying collection and test set in proportion
Block.
4) for creating the module of user's operation behavior prediction system, the user's operation behavior prediction system includes to grasp
Make behavior sequence, the deep layer LSTM subsystem that the time of origin of operation behavior is parameter and using external environmental factor as the complete of parameter
Connection Neural Network subsystem.
5) for training set, verifying collection classification to be imported user's operation behavior prediction system, using window sliding technology pair
User's operation behavior prediction system is trained and verifies, to obtain the module of the user's operation behavior prediction system after optimization.
6) for test set to be imported to the module of the user's operation behavior prediction system after optimization.
7) for carrying out the module of performance evaluation to the user's operation behavior prediction system after optimization.
The above technical solution of the present invention, compared with existing, significant beneficial effect is,
1) pre-training has been carried out to the behavior of user using Word2Vector algorithm, has improved the receipts of total system
Hold back speed.
2) regularity of behavior on same time point is considered simultaneously, the operating time of each behavior and behavior itself is same
When input LSTM.
3) increase data volume using window sliding technology to learn Continuous behavior rule.
4) inherent laws of continuous user behavior are considered comprehensively and influence the external factor of user behavior.
5) interface is left for the expansion of later period system, if the later period finds holiday influence factor, will can be directly somebody's turn to do
Full Connection Neural Network is inputted after factor vectorization, regularization, can incorporate system.
It should be appreciated that as long as aforementioned concepts and all combinations additionally conceived described in greater detail below are at this
It can be viewed as a part of the subject matter of the disclosure in the case that the design of sample is not conflicting.In addition, required guarantor
All combinations of the theme of shield are considered as a part of the subject matter of the disclosure.
Can be more fully appreciated from the following description in conjunction with attached drawing present invention teach that the foregoing and other aspects, reality
Apply example and feature.The features and/or benefits of other additional aspects such as illustrative embodiments of the invention will be below
Description in it is obvious, or learnt in practice by the specific embodiment instructed according to the present invention.
Detailed description of the invention
Attached drawing is not intended to drawn to scale.In the accompanying drawings, identical or nearly identical group each of is shown in each figure
It can be indicated by the same numeral at part.For clarity, in each figure, not each component part is labeled.
Now, example will be passed through and the embodiments of various aspects of the invention is described in reference to the drawings, in which:
Fig. 1 is the structure chart of full Connection Neural Network of the invention.
Fig. 2 is the network structure of RNN of the invention.
Fig. 3 is the structure chart of LSTM hidden layer of the invention.
Fig. 4 is the sliding type schematic diagram that window size of the invention is 4.
Fig. 5 is the user's operation behavior prediction method flow diagram of the invention based on deep learning.
Fig. 6 is the structure of one of example of the user's operation behavior prediction system of the invention based on deep learning
Figure.
Specific embodiment
In order to better understand the technical content of the present invention, special to lift specific embodiment and institute's accompanying drawings is cooperated to be described as follows.
Below with reference to Fig. 5 and Fig. 6, the present invention is described in further detail.
The first step, the first journal file in collection application system.The content that every log of journal file must include
Have: the information such as date, time, current User ID, affiliated company ID, affiliated function ID for logging in application system of current operation.
All journal files are merged into a file when collecting daily record data, when then by all data with user name and date
Between be ranked up, prevent the phenomenon that user interts.
Second step, data are extracted.What this step was done first is to divide all complete operations, i.e., according to ID pairs of login module
Data are cut, and data are cut into again and again complete operation, then carry out forms sliding for all complete operations,
Generate new data set.The respective data of attributes extraction finally are pressed for new data set, are put into different files, so as to subsequent
Respectively handled.
Third step, data prediction.For external factor, vectorization operation is all carried out, eliminates the number of data itself
It is worth the influence to system, vectorization operation is similarly carried out to the time of each operation, meanwhile, operation itself is used
Word2Vector algorithm pretreatment, since the algorithm is itself is a neural network, result also can tentatively obtain with currently
The recent operation of operating distance, so be equivalent to carry out operation pre-training using Word2Vector algorithm,
It can be improved the convergence rate of whole system.
4th step defines deep layer LSTM subsystem.By after vectorization time and operation behavior correspond and merge, and make
Sliding processing, sliding window size take 10, using final sliding result as the input of deep layer LSTM subsystem, define deep layer LSTM
The structure of subsystem.In order to over-fitting occur in training LSTM, DropOut joined in deep layer LSTM subsystem.
5th step defines the full Connection Neural Network subsystem of external factor.In order to which the external factor after vectorization is melted
Enter full Connection Neural Network subsystem, one-dimensional full Processing with Neural Network is respectively done to all external factor, passes through integral subsystem
Training, obtain the weight matrix of each external factor, and simultaneously training when respective weight matrix is made at L2 regularization
Reason.
6th step merges the output result of both sides subsystem, and executes primary full attended operation, for that will connect entirely
Neural network subsystem is blended with deep layer LSTM subsystem.
7th step, systematic training and optimization.System is exactly to feed data to system after defining, but due to data volume mistake
Greatly, if once feeding total data, a large amount of memory sources of computer can be occupied, result even in computer collapse sometimes.In order to
Solve the problems, such as this, using the method for feeding data in batches, and mostly wheel training, and system is constantly optimized in training,
The optimization algorithm of use is " adam optimization algorithm ", and what loss function was selected is cross entropy loss function of classifying more
(categorical_crossentropy)。
8th step, performance evaluation.After user's operation system of behavior trains, the superiority and inferiority of the system is exactly evaluated.Due to
The system belongs to categorizing system, so using the accuracy rate of classification come the superiority and inferiority of visual evaluation system.
Expression effect of the present invention in test set:
Under the premise of giving 9 behaviors, the accuracy rate of the 10th behavior of prediction is 58.7%.
Use traditional performance results based on frequency statistics method:
Under the premise of giving 9 behaviors, the accuracy rate of the 10th behavior of prediction is 32.2%.
Predictablity rate of the present invention under the scene improves 26.5%.
Various aspects with reference to the accompanying drawings to describe the present invention in the disclosure, shown in the drawings of the embodiment of many explanations.
Embodiment of the disclosure need not be defined on including all aspects of the invention.It should be appreciated that a variety of designs and reality presented hereinbefore
Those of apply example, and describe in more detail below design and embodiment can in many ways in any one come it is real
It applies, this is because conception and embodiment disclosed in this invention are not limited to any embodiment.In addition, disclosed by the invention one
A little aspects can be used alone, or otherwise any appropriately combined use with disclosed by the invention.
Although the present invention has been disclosed as a preferred embodiment, however, it is not to limit the invention.Skill belonging to the present invention
Has usually intellectual in art field, without departing from the spirit and scope of the present invention, when can be used for a variety of modifications and variations.Cause
This, the scope of protection of the present invention is defined by those of the claims.
Claims (10)
1. a kind of user's operation behavior prediction method based on deep learning, which is characterized in that the described method includes:
S1: extracting user behavior data from historical operation log, and the user behavior data includes at least environmental data and company
Continuous behavioral data, environmental data include multiple external environmental factors when data occur, and Continuous behavior data include at least user
Operation behavior sequence and operation behavior time of origin;
S2: user behavior data is pre-processed;
S3: pretreated user behavior data is divided into training set, verifying collection and test set in proportion;
S4: creation user's operation behavior prediction system, the user's operation behavior prediction system include with operation behavior sequence, behaviour
The time of origin for making behavior is the deep layer LSTM subsystem of parameter and using external environmental factor as the full Connection Neural Network of parameter
Subsystem;
S5: training set, verifying collection classification are imported into user's operation behavior prediction system, using window sliding technology to user's operation
Behavior prediction system is trained and verifies, to obtain the user's operation behavior prediction system after optimization;
S6: importing the user's operation behavior prediction system after optimization for test set, if the prediction of the user behavior of output is accurate
Rate is more than or equal to setting accuracy rate threshold value, enters step S7, otherwise, the calculation formula of return step S5, predictablity rate are as follows:
Wherein, n is the total sample number of predictive behavior, y 'iIt is predicted operation behavior serial number, yiIt is practical operation behavior serial number, αiIt is
The accuracy of i-th of predictive behavior;
S7: completing training, obtains the user's operation behavior prediction system that optimization is completed.
2. the user's operation behavior prediction method according to claim 1 based on deep learning, which is characterized in that described outer
Portion's environmental factor include company ID, department ID, User ID, behavior occur period in it is several or whole.
3. the user's operation behavior prediction method according to claim 1 based on deep learning, which is characterized in that step S2
In, carrying out pretreated method to user behavior data includes:
S21: the noise data unrelated with business is rejected;
S22: to remaining environmental data execute normalized, to remaining Continuous behavior data carry out pre-training with realize to
Quantization.
4. the user's operation behavior prediction method according to claim 3 based on deep learning, which is characterized in that the side
Method further include:
Word2vector technology is used to carry out pre-training to Continuous behavior data to realize vectorization.
5. the user's operation behavior prediction method according to claim 1 based on deep learning, which is characterized in that step S5
In, training set, verifying collection classification are imported into user's operation behavior prediction system, using window sliding technology to user's operation behavior
Forecasting system is trained and verifies, the method to obtain the user's operation behavior prediction system after optimization further include:
The part for belonging to Continuous behavior data that training set, verifying are concentrated is imported into deep layer LSTM subsystem, by training set, verifying
The part for belonging to environmental data concentrated imports full Connection Neural Network subsystem, and the two is spliced to seeking unity of action afterwards at least together
N wheel training, saves the smallest subsystem of penalty values as the subsystem after optimization.
6. the user's operation behavior prediction method according to claim 5 based on deep learning, which is characterized in that the N
Equal to 50.
7. the user's operation behavior prediction method according to claim 1 or described in 5 based on deep learning, which is characterized in that
The method also includes:
Selecting size is that 10 sliding window is trained and verifies to user's operation behavior prediction system.
8. the user's operation behavior prediction method according to claim 1 based on deep learning, which is characterized in that the side
Method further include:
In step S5, training set, verifying collection classification are imported into user's operation behavior prediction system by the way of importing data in batches
System.
9. the user's operation behavior prediction method according to claim 1 based on deep learning, which is characterized in that the side
Method further include:
In step S5, when to user's operation behavior prediction systematic training, the optimization algorithm used loses letter for adam optimization algorithm
Number selects cross entropy loss functions of classifying more.
10. a kind of user's operation behavior prediction system based on deep learning, which is characterized in that the system comprises:
For extracting the module of user behavior data from historical operation log, the user behavior data includes at least environment number
According to Continuous behavior data, environmental data includes multiple external environmental factors when data occur, and Continuous behavior data at least wrap
Include the operation behavior sequence of user and the time of origin of operation behavior;
For carrying out pretreated module to user behavior data;
For pretreated user behavior data to be divided into the module of training set, verifying collection and test set in proportion;
For creating the module of user's operation behavior prediction system, the user's operation behavior prediction system includes with operation behavior
Sequence, operation behavior time of origin be parameter deep layer LSTM subsystem and using external environmental factor as parameter full connection mind
Through network subsystem;
For training set, verifying collection classification to be imported user's operation behavior prediction system, user is grasped using window sliding technology
Make behavior prediction system to be trained and verify, to obtain the module of the user's operation behavior prediction system after optimization;
For test set to be imported to the module of the user's operation behavior prediction system after optimization;
For carrying out the module of performance evaluation to the user's operation behavior prediction system after optimization.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811493939.0A CN109583659A (en) | 2018-12-07 | 2018-12-07 | User's operation behavior prediction method and system based on deep learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811493939.0A CN109583659A (en) | 2018-12-07 | 2018-12-07 | User's operation behavior prediction method and system based on deep learning |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109583659A true CN109583659A (en) | 2019-04-05 |
Family
ID=65929049
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811493939.0A Pending CN109583659A (en) | 2018-12-07 | 2018-12-07 | User's operation behavior prediction method and system based on deep learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109583659A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110084150A (en) * | 2019-04-09 | 2019-08-02 | 山东师范大学 | A kind of Automated Classification of White Blood Cells method and system based on deep learning |
CN110110875A (en) * | 2019-05-15 | 2019-08-09 | 常州纺织服装职业技术学院 | Land resources surveying and mapping activity reservation system |
CN110991711A (en) * | 2019-11-19 | 2020-04-10 | 浙江鸿程计算机系统有限公司 | Multi-factor perception terminal switching prediction method based on deep neural network |
CN114154816A (en) * | 2021-11-17 | 2022-03-08 | 鼎捷软件股份有限公司 | Enterprise management system and execution method thereof |
CN114386688A (en) * | 2022-01-07 | 2022-04-22 | 深圳市聚商鼎力网络技术有限公司 | User intention prediction method and system based on multi-data fusion |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107423442A (en) * | 2017-08-07 | 2017-12-01 | 火烈鸟网络(广州)股份有限公司 | Method and system, storage medium and computer equipment are recommended in application based on user's portrait behavioural analysis |
CN108009674A (en) * | 2017-11-27 | 2018-05-08 | 上海师范大学 | Air PM2.5 concentration prediction methods based on CNN and LSTM fused neural networks |
-
2018
- 2018-12-07 CN CN201811493939.0A patent/CN109583659A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107423442A (en) * | 2017-08-07 | 2017-12-01 | 火烈鸟网络(广州)股份有限公司 | Method and system, storage medium and computer equipment are recommended in application based on user's portrait behavioural analysis |
CN108009674A (en) * | 2017-11-27 | 2018-05-08 | 上海师范大学 | Air PM2.5 concentration prediction methods based on CNN and LSTM fused neural networks |
Non-Patent Citations (1)
Title |
---|
张备: "基于多神经网络的混合动态推荐研究", 《优秀硕士学位论文全文库》 * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110084150A (en) * | 2019-04-09 | 2019-08-02 | 山东师范大学 | A kind of Automated Classification of White Blood Cells method and system based on deep learning |
CN110084150B (en) * | 2019-04-09 | 2021-05-11 | 山东师范大学 | Automatic white blood cell classification method and system based on deep learning |
CN110110875A (en) * | 2019-05-15 | 2019-08-09 | 常州纺织服装职业技术学院 | Land resources surveying and mapping activity reservation system |
CN110991711A (en) * | 2019-11-19 | 2020-04-10 | 浙江鸿程计算机系统有限公司 | Multi-factor perception terminal switching prediction method based on deep neural network |
CN114154816A (en) * | 2021-11-17 | 2022-03-08 | 鼎捷软件股份有限公司 | Enterprise management system and execution method thereof |
CN114386688A (en) * | 2022-01-07 | 2022-04-22 | 深圳市聚商鼎力网络技术有限公司 | User intention prediction method and system based on multi-data fusion |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109583659A (en) | User's operation behavior prediction method and system based on deep learning | |
Shen et al. | Wind speed prediction of unmanned sailboat based on CNN and LSTM hybrid neural network | |
CN107423442A (en) | Method and system, storage medium and computer equipment are recommended in application based on user's portrait behavioural analysis | |
CN111754025B (en) | CNN+GRU-based short-time bus passenger flow prediction method | |
CN113905391A (en) | Ensemble learning network traffic prediction method, system, device, terminal, and medium | |
CN106874478A (en) | Parallelization random tags subset multi-tag file classification method based on Spark | |
Wang et al. | Learning performance prediction via convolutional GRU and explainable neural networks in e-learning environments | |
CN113591728A (en) | Electric energy quality disturbance classification method based on integrated deep learning | |
CN110837888A (en) | Traffic missing data completion method based on bidirectional cyclic neural network | |
Wu et al. | Optimized deep learning framework for water distribution data-driven modeling | |
Li et al. | First-order sensitivity analysis for hidden neuron selection in layer-wise training of networks | |
Kumar et al. | Wind speed prediction using deep learning-LSTM and GRU | |
Sangroya et al. | Guided-LIME: Structured Sampling based Hybrid Approach towards Explaining Blackbox Machine Learning Models. | |
CN116542701A (en) | Carbon price prediction method and system based on CNN-LSTM combination model | |
CN111813939A (en) | Text classification method based on representation enhancement and fusion | |
CN115080795A (en) | Multi-charging-station cooperative load prediction method and device | |
Kakade et al. | A neural network approach for text document classification and semantic text analytics | |
CN111783688B (en) | Remote sensing image scene classification method based on convolutional neural network | |
CN112487191A (en) | Text classification method and device based on CNN-BilSTM/BiGRU hybrid combination model | |
CN117131979A (en) | Traffic flow speed prediction method and system based on directed hypergraph and attention mechanism | |
CN116975743A (en) | Industry information classification method, device, computer equipment and storage medium | |
CN114881172A (en) | Software vulnerability automatic classification method based on weighted word vector and neural network | |
CN114818690A (en) | Comment information generation method and device and storage medium | |
Rathee et al. | A machine learning approach to predict the next word in a statement | |
CN113723660A (en) | Specific behavior type prediction method and system based on DNN-LSTM fusion model |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190405 |