CN109710648A - A kind of finish message method based on transfer learning - Google Patents
A kind of finish message method based on transfer learning Download PDFInfo
- Publication number
- CN109710648A CN109710648A CN201811648481.1A CN201811648481A CN109710648A CN 109710648 A CN109710648 A CN 109710648A CN 201811648481 A CN201811648481 A CN 201811648481A CN 109710648 A CN109710648 A CN 109710648A
- Authority
- CN
- China
- Prior art keywords
- transfer learning
- information
- mobile terminal
- method based
- finish message
- 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
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Electrically Operated Instructional Devices (AREA)
Abstract
The invention discloses a kind of finish message method based on transfer learning specifically includes and establishes database, establish migration models, information input, arrange result.The invention has the advantages that the efficiency for effectively improving educational information screening and obtaining, and user is made to have more screening selections.
Description
Technical field
The present invention relates to educational informations to arrange related fields, especially a kind of finish message method based on transfer learning.
Background technique
In terms of the information management of educational archive, none suitable, efficient screening technique of current position is arrived, logical
The information for wanting to obtain designated person in normal situation, needs to recognize certain information in advance, which not only adds workload,
And since data are huge, and screening conditions are limited, can not intelligence the unrelated information of shielding, it is inaccurate to finally result in result
Really, there is biggish deviation from the result of anticipation.
If a kind of finish message method based on transfer learning can be designed it is possible to prevente effectively from more than common problem,
To improve working efficiency.
Summary of the invention
The purpose of the present invention is to solve the above problems, devise a kind of finish message method based on transfer learning.
Realize above-mentioned purpose the technical scheme is that, a kind of finish message method based on transfer learning is specific to wrap
Include following steps:
Step 1: establishing database；By the file collection in relation to resource information to server terminal and database is established, database
Information classified by conditions such as age, region, achievements, server terminal is connected by the network port and mobile terminal
It connects.
Step 2: establishing migration models；The transfer learning model of multiple and different attributes is established in server terminal, is migrated
It is in relatively independent space between learning model and transfer learning model, can be learned in special circumstances by designated program in migration
It practises and carries out interspersed analysis between model, this process can be repeated；Computational space construction inside transfer learning model can carry out
The different algorithms such as positive transfer, suitable migration.
Step 3: information input；Corresponding information is inputted in mobile terminal and is saved, and mobile terminal passes through later
The network port uploads onto the server information in terminal, can be by specifying at transfer learning model when information input
Reason, when through different transfer learning mode input information, server terminal is carried out with different migration algorithm
Finish message, server terminal can put question to relevant issues to mobile terminal according to the correlation experience of database during this period, mobile
Terminal is simultaneously answered.
Step 4: arranging result；The associated documents in the information and date library of input are compared, are divided by server terminal
It obtains arranging result after analysis, screening；It arranges result and is equipped with multiple, and be successively ranked up from excellent to bad, every kind of result is attached
Note sequence reason and details factors；When arrangement result error is larger, it can be carried out again by relevant reseting port
It arranges, this process need to provide detailed negative information and correlative factor, can upload again, carry out finish message again.
Arrangement result in the step 4 can be according to the height of average achievement, the range for attending school region and kinsfolk
Quantity etc. because being usually ranked up, details factors be embodied in for example, take an examination in the recent period time, score, place；Kinsfolk's
The details such as number, relationship, age.
When inputting information in the step 3, mobile terminal meeting automated back-up simultaneously uploads to cloud.
Mobile terminal stores writing record when inputting information in the step 3.
The step of database is putd question to mobile terminal in the step 3 can not force to cancel and be subject to time restriction.
The algorithm of transfer learning model is irreversible in the step 2.
The finish message method method based on transfer learning made using technical solution of the present invention, by based on migration
The method of study can make the reasonable operation of screening information, compared and carried out of system intelligence obtain most reasonable knot
Fruit, the method not only increase the precision of calculated result, the time of the user more saved.
Detailed description of the invention
Fig. 1 is a kind of flow diagram of finish message method based on transfer learning of the present invention；
Fig. 2 is the flow diagram of the present invention for establishing migration models；
Fig. 3 is the flow diagram of the present invention for arranging result.
Specific embodiment
The present invention is specifically described with reference to the accompanying drawing, as shown in Figs. 1-3.
In the present embodiment, the first step, by the condition entry to mobile terminal for needing to obtain information, select move later
Learning model is moved to be calculated, screened to determine different algorithms.Transfer learning model can lead to when being calculated
Crossing relevant operation makes calculation change, when transfer learning model is calculated can not timing to mobile terminal
A series of enquirement is carried out, to improve the accuracy for arranging result；
Second step passes through and constantly puts question to and answer, the comparison that server carries out the information of resource and submission in database,
Then proper at screening as a result, this process carry out repeatedly, so as to obtain the result of respective numbers；
Third step, after result comes out, if user is dissatisfied to result is arranged, can also be carried out again by reseting port
It arranges.It needs for unsatisfied opinion to be input in mobile terminal before arrangement, consequently facilitating improving the accuracy for arranging result.
Case study on implementation one,
When carrying out educational information arrangement, such as wish that the student of nearest region comes school and attends school, and also wants to control
Come school in certain score to attend school, score information and certificate address information can be input in mobile terminal when operation, this letter
The screening of breath can carry out finish message using the calculation method of positive transfer, have and input in address information in database
The partial document of information same text, it is the student to live in specified region that just there is a strong possibility, adds one to score again later
Definite limitation may finally filter out suitable arrange as a result, and being ranked up according to the distance of distance and the height of score.
Case study on implementation two,
When carrying out educational information arrangement, such as wish that red shirt comes school and attends school, and score is required not high.
Can be by the information inputs to mobile terminal such as score information and height, weight when operation, the screening of this information can
Finish message is carried out with the calculation method using vertical migration, more one layer of filter information, server on the basis of positive transfer
Terminal compares the information in information and date library, filters out that height is higher, the student of heavier-weight.Because height it is high with
And the exercise for being all conducive to the later period of heavyweight vehicle；The student of this group possess with one middle school student's identical point of case study on implementation, and
Difference is embodied in height and weight.
Case study on implementation three,
When carrying out educational information arrangement, such as wishes that the student in outer school transfers to another school and attend school, and is more demanding to score.?
It can be by the information input mobile terminal in this school when operation.This information screening can using the calculation method far migrated come into
Row information arranges, and server terminal in the relevant information arrangement entirely different with this school information coming out, because of the student in outer school
It is not related for having very big probability with the student in this school.Obtaining can be ranked up after arranging result according to the height of score.
Above-mentioned technical proposal only embodies the optimal technical scheme of technical solution of the present invention, those skilled in the art
The principle of the present invention is embodied to some variations that some of them part may be made, belongs to the scope of protection of the present invention it
It is interior.
Claims (6)
1. a kind of finish message method based on transfer learning, which is characterized in that specifically comprise the following steps:
Step 1: establishing database；By the file collection in relation to resource information to server terminal and database is established, database
Information classified by conditions such as age, region, achievements, server terminal is connected by the network port and mobile terminal
It connects.
Step 2: establishing migration models；The transfer learning model of multiple and different attributes, transfer learning are established in server terminal
Relatively independent space is between model and transfer learning model, it in special circumstances can be by designated program in transfer learning mould
Interspersed analysis is carried out between type, this process can be repeated；Computational space construction inside transfer learning model can be moved just
It moves, along different algorithms such as migrations.
Step 3: information input；Corresponding information is inputted in mobile terminal and is saved, and mobile terminal passes through network later
Port uploads onto the server information in terminal, can be handled by specified transfer learning model when information input,
When through different transfer learning mode input information, server terminal carries out information with different migration algorithm
It arranges, server terminal can put question to relevant issues, mobile terminal to mobile terminal according to the correlation experience of database during this period
And it is answered.
Step 4: arranging result；The associated documents in the information and date library of input are compared, analyzed, sieved by server terminal
It obtains arranging result after choosing；It arranges result and is equipped with multiple, and be successively ranked up from excellent to bad, every kind of result note sequence
Reason and details factors；When arrangement result error is larger, it can be rearranged by relevant reseting port, this
Process need to provide detailed negative information and correlative factor, can upload again, carry out finish message again.
2. a kind of finish message method based on transfer learning according to claim 1, which is characterized in that the step 4
In arrangement result can according to the height of average achievement, attend school the range of region and the quantity of kinsfolk etc. because usually carrying out
Sequence, details factors be embodied in for example, take an examination in the recent period time, score, place；Number, relationship, age of kinsfolk etc. are thin
Section.
3. a kind of finish message method based on transfer learning according to claim 1, which is characterized in that the step 3
When middle input information, mobile terminal meeting automated back-up simultaneously uploads to cloud.
4. a kind of finish message method based on transfer learning according to claim 1, which is characterized in that the step 3
Mobile terminal stores writing record when middle input information.
5. a kind of finish message method based on transfer learning according to claim 1, which is characterized in that the step 3
The step of middle database is putd question to mobile terminal can not force to cancel and be subject to time restriction.
6. a kind of finish message method based on transfer learning according to claim 1, which is characterized in that the step 2
The algorithm of middle transfer learning model is irreversible.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811648481.1A CN109710648A (en) | 2018-12-30 | 2018-12-30 | A kind of finish message method based on transfer learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811648481.1A CN109710648A (en) | 2018-12-30 | 2018-12-30 | A kind of finish message method based on transfer learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109710648A true CN109710648A (en) | 2019-05-03 |
Family
ID=66260455
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811648481.1A Pending CN109710648A (en) | 2018-12-30 | 2018-12-30 | A kind of finish message method based on transfer learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109710648A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106777261A (en) * | 2016-12-28 | 2017-05-31 | 深圳市华傲数据技术有限公司 | Data query method and device based on multi-source heterogeneous data set |
| CN106781769A (en) * | 2016-12-28 | 2017-05-31 | 朱佐昆 | Classroom interactions' system |
| CN107169847A (en) * | 2017-06-21 | 2017-09-15 | 苏州发飚智能科技有限公司 | The system and method for short room rate of renting a house dynamically are adjusted based on machine learning model |
| CN107491484A (en) * | 2017-07-17 | 2017-12-19 | 阿里巴巴集团控股有限公司 | A kind of data matching method, device and equipment |
| CN108090809A (en) * | 2017-12-18 | 2018-05-29 | 赣州欧唯科技有限公司 | A kind of sticking film for mobile phone selection method, system, medium and equipment |
-
2018
- 2018-12-30 CN CN201811648481.1A patent/CN109710648A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106777261A (en) * | 2016-12-28 | 2017-05-31 | 深圳市华傲数据技术有限公司 | Data query method and device based on multi-source heterogeneous data set |
| CN106781769A (en) * | 2016-12-28 | 2017-05-31 | 朱佐昆 | Classroom interactions' system |
| CN107169847A (en) * | 2017-06-21 | 2017-09-15 | 苏州发飚智能科技有限公司 | The system and method for short room rate of renting a house dynamically are adjusted based on machine learning model |
| CN107491484A (en) * | 2017-07-17 | 2017-12-19 | 阿里巴巴集团控股有限公司 | A kind of data matching method, device and equipment |
| CN108090809A (en) * | 2017-12-18 | 2018-05-29 | 赣州欧唯科技有限公司 | A kind of sticking film for mobile phone selection method, system, medium and equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Criado et al. | Non-iid data and continual learning processes in federated learning: A long road ahead | |
| WO2019165673A1 (en) | Reimbursement form risk prediction method, apparatus, terminal device, and storage medium | |
| CN106022568A (en) | Workflow processing method and apparatus | |
| CN110008259A (en) | The method and terminal device of visualized data analysis | |
| KR101732319B1 (en) | IRIS: Goal-oriented big data business analytics framework | |
| CN108809694A (en) | Arranging service method, system, device and computer readable storage medium | |
| Liu et al. | Consensus model based on probability K-means clustering algorithm for large scale group decision making | |
| CN111047563B (en) | Neural network construction method applied to medical ultrasonic image | |
| CN104281615A (en) | Complaint handling method and system | |
| CN105046323B (en) | Regularization-based RBF network multi-label classification method | |
| CN112800115B (en) | Data processing method and data processing device | |
| CN112529184A (en) | Industrial process optimization decision method fusing domain knowledge and multi-source data | |
| CN106033425A (en) | A data processing device and a data processing method | |
| CN109816043A (en) | Determination method, apparatus, electronic equipment and the storage medium of user's identification model | |
| CN113688490A (en) | Network co-construction sharing processing method, device, equipment and storage medium | |
| US11915120B2 (en) | Flexible parameter sharing for multi-task learning | |
| Sun | Study on application of data mining technology in university computer network educational administration management system | |
| CN109993026A (en) | The training method and device of relatives' identification network model | |
| CN109710648A (en) | A kind of finish message method based on transfer learning | |
| Shraga et al. | Inductive context-aware process discovery | |
| CN113762514A (en) | Data processing method, device, equipment and computer readable storage medium | |
| CN110867224A (en) | Multi-granularity Spark super-trust fuzzy method for large-scale brain pathology segmentation | |
| JP4780668B2 (en) | Traffic analysis model construction method, apparatus, construction program, and storage medium thereof | |
| Gias et al. | Samplehst: Efficient on-the-fly selection of distributed traces | |
| CN115292361A (en) | Method and system for screening distributed energy abnormal data |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190503 |
|
| WD01 | Invention patent application deemed withdrawn after publication |