CN109711436A - A kind of artificial intelligence training pattern construction method, device and storage medium - Google Patents
A kind of artificial intelligence training pattern construction method, device and storage medium Download PDFInfo
- Publication number
- CN109711436A CN109711436A CN201811481871.4A CN201811481871A CN109711436A CN 109711436 A CN109711436 A CN 109711436A CN 201811481871 A CN201811481871 A CN 201811481871A CN 109711436 A CN109711436 A CN 109711436A
- Authority
- CN
- China
- Prior art keywords
- training pattern
- information
- artificial intelligence
- model
- built
- 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)
Abstract
The embodiment of the invention discloses a kind of artificial intelligence training pattern construction method, device and storage mediums, the present invention relates to field of computer technology, wherein, which comprises obtain the modeling solicited message corresponding with artificial intelligence training pattern to be built of active user；According to modeling solicited message, corresponding training pattern building task is generated；Modeling solicited message is analyzed according to the analysis model of default training pattern, obtains analyzing result accordingly, wherein analysis result includes at least the model application message and model training information of artificial intelligence training pattern to be built；According to model application message and model training information, executes training pattern and construct task, construct and obtain corresponding training pattern.The solution of the present invention, can intellectual analysis go out the modeling solicited message of user, and targetedly construct corresponding training pattern, simplify existing artificial intelligence training pattern building process, improve user experience.
Description
Technical field
The present invention relates to field of computer technology, and in particular to a kind of artificial intelligence training pattern construction method, device and
Storage medium.
Background technique
With the development of science and technology, artificial intelligence technology is also developed.Now, artificial intelligence technology is widely answered
For the every aspect of people's daily life, for example, the smart home being widely used now, for example, Intelligent oven, intelligent machine
Device people, pilotless automobile, intelligent washing machine, intelligent air condition etc..With the development of artificial intelligence technology, more and more intelligence
Can domestic applications in daily life, it is convenient to provide for daily life, allows more people from daily many and diverse
It frees in housework, there are more times to lie fallow.
The building process of existing artificial intelligence training pattern is the training sample and correspondence that user chooses pending training
Test sample, then model is trained using corresponding algorithm, tests, constantly improve, finally obtain and meet user's need
The artificial intelligence training pattern asked, in above process, algorithm are to be familiar with based on user in existing algorithm, or even be proficient in grasp
On the basis of, further corresponding one be related in existing algorithm or several parameters are improved, it is above-mentioned to train
Journey, constantly existing algorithm optimizes, in order to finally obtain the algorithm for more fitting in user demand, and it is excellent based on this
Algorithm after change is trained the building and optimization process of model.
Above-mentioned artificial intelligence training pattern and the process of subsequent optimization, are extremely complex processes, need user for tool
There are the professional person of software development capabilities, and the stock of knowledge with corresponding different business field.How to simplify existing
Artificial intelligence training pattern building process is technical problem to be solved.
Summary of the invention
Be designed to provide a kind of artificial intelligence training pattern construction method, device and the storage of the embodiment of the present invention are situated between
Matter, to solve the problems, such as that existing artificial intelligence training pattern building process is excessively complicated.
To achieve the above object, the embodiment of the present invention provides a kind of artificial intelligence training pattern construction method, the method
It include: the modeling solicited message corresponding with artificial intelligence training pattern to be built for obtaining active user；According to the modeling
Solicited message generates corresponding training pattern building task；The modeling is requested according to the analysis model of default training pattern
Information is analyzed, and obtains analyzing result accordingly, wherein the analysis result includes at least the artificial intelligence to be built
The model application message and model training information of training pattern；According to the model application message and the model training information,
The training pattern building task is executed, constructs and obtains corresponding training pattern.
Optionally, the model application message is included at least with the next item down: the artificial intelligence training pattern to be built
Corresponding input paramete information, wherein the input paramete information includes at least text input parameter information, video file inputs
It is one or more in parameter information, audio file input paramete information；The artificial intelligence training pattern to be built is corresponding
Output parameter information, wherein the output parameter information include at least text output parameter information, video file output parameter
It is one or more in information, audio file output parameter information；The artificial intelligence training pattern to be built corresponding
One model object information, wherein the first model object information includes at least the learning framework for carrying out model training and believes
At least one of in breath, the first training pattern information corresponding with the learning framework.
Optionally, the model training information is included at least with the next item down: the artificial intelligence training pattern to be built
Used material file information when corresponding training pattern, wherein the material file information includes at least the material text
The file name information of part, the file type information of the material file, the document content information of the material file, the element
It is one or more in the storage location information of material file；The corresponding configuration file of artificial intelligence training pattern to be built
Information, wherein the profile information includes at least the file structure letter of the file structure for describing the material file
Breath, for describing training pattern when used frame frame information at least one of；The artificial intelligence to be built
Training pattern is corresponding for adjusting the hyper parameter information of training pattern；The artificial intelligence training pattern to be built is corresponding
Second model object information, wherein the second model object information includes at least the second training pattern information trained.
Optionally, the method also includes: according to preset evaluation rule, the training pattern is evaluated, is obtained
Corresponding evaluation result, and according to the evaluation result, the training pattern is optimized, the training mould after being optimized
Type.
Match with the above method, another aspect of the present invention provides a kind of artificial intelligence training pattern construction device, described
Device includes: acquiring unit, obtains the modeling solicited message corresponding with artificial intelligence training pattern to be built of active user；
Task generation unit, the modeling solicited message according to the information acquired by the acquisition unit generate corresponding training pattern building
Task；Analytical unit analyzes the modeling solicited message according to the analysis model of default training pattern, obtains corresponding
Analyze result, wherein the analysis result includes at least the model application message of the artificial intelligence training pattern to be built
With model training information；Training pattern construction unit, the model application message analyzed according to the analytical unit and institute
Model training information is stated, the training pattern building task that the task generation unit generates is executed, constructs and obtain corresponding
Training pattern.
Optionally, the model application message that the analytical unit analyzes is included at least with the next item down: described to be built
The corresponding input paramete information of artificial intelligence training pattern, wherein the input paramete information includes at least text input parameter
It is information, video file input paramete information, one or more in audio file input paramete information；It is described to be built artificial
The corresponding output parameter information of intelligent training model, wherein the output parameter information include at least text output parameter information,
It is one or more in video file output parameter information, audio file output parameter information；The artificial intelligence to be built
The corresponding first model object information of training pattern, wherein the first model object information is included at least for carrying out model
At least one of in trained learning framework information, the first training pattern information corresponding with the learning framework.
Optionally, the model training information that the analytical unit analyzes is included at least with the next item down: described to structure
Used material file information when the corresponding training pattern of the artificial intelligence training pattern built, wherein the material file letter
Breath includes at least the file name information of the material file, the file type information of the material file, the material file
Document content information, the material file storage location information in it is one or more；The artificial intelligence to be built
The corresponding profile information of training pattern, wherein the profile information is included at least for describing the material file
File structure document structure information, for describing training pattern when used frame frame information at least one
?；The artificial intelligence training pattern to be built is corresponding for adjusting the hyper parameter information of training pattern；It is described to be built
The corresponding second model object information of artificial intelligence training pattern, wherein the second model object information include at least instruction
The the second training pattern information practised.
Optionally, described device further include: evaluation unit constructs the training pattern according to preset evaluation rule
The training pattern that unit obtains is evaluated, and corresponding evaluation result is obtained；Training pattern optimizes unit, according to institute's commentary
The evaluation result that valence unit obtains, the training pattern obtained to the training pattern construction unit are optimized, are obtained excellent
Training pattern after change.
Match with the above method, further aspect of the present invention provides a kind of electronic equipment, and institute's electronic equipment includes: memory
And processor, the processor and the memory complete mutual communication by bus；The memory is stored with can quilt
The program instruction that the processor executes, the processor call described program instruction to be able to carry out method as described above.
Match with the above method, another aspect of the invention provides a kind of computer readable storage medium, is stored thereon with
The step of computer program, the computer program realizes any of the above-described the method when being executed by processor.
The embodiment of the present invention has the advantages that a kind of artificial intelligence training pattern building side provided in an embodiment of the present invention
Method, device and storage medium, can intellectual analysis go out the modeling solicited message of user, and the targetedly corresponding training mould of building
Type simplifies existing artificial intelligence training pattern building process, improves user experience.
Detailed description of the invention
Fig. 1 is a kind of flow diagram for artificial intelligence training pattern construction method that the embodiment of the present invention 1 provides；
Fig. 2 is a kind of structural schematic diagram for artificial intelligence training pattern construction device that the embodiment of the present invention 2 provides.
In figure: 201- acquiring unit；202- task generation unit；203- analytical unit；204- training pattern construction unit.
Specific embodiment
Embodiments of the present invention are illustrated by particular specific embodiment below, those skilled in the art can be by this explanation
Content disclosed by book is understood other advantages and efficacy of the present invention easily.
Embodiment 1
Embodiment according to the present invention 1 provides a kind of artificial intelligence training pattern construction method, as shown in Figure 1, for this
A kind of flow diagram for artificial intelligence training pattern construction method that inventive embodiments 1 provide.This method includes at least following
Step:
S101 obtains the modeling solicited message corresponding with artificial intelligence training pattern to be built of active user；
S102 generates corresponding training pattern building task according to modeling solicited message；
S103 analyzes modeling solicited message according to the analysis model of default training pattern, is analyzed accordingly
As a result, wherein analysis result includes at least the model application message and model training letter of artificial intelligence training pattern to be built
Breath；
S104 executes training pattern and constructs task, construct and obtain phase according to model application message and model training information
The training pattern answered；In this way, 1 construction method provided through the embodiment of the present invention, can intellectual analysis go out the modeling of user and ask
Information is sought, and targetedly constructs corresponding training pattern, simplifies existing artificial intelligence training pattern building process, is improved
User experience.
It should be noted that presetting the analysis of training pattern in the scheme for the construction method that the embodiment of the present invention 1 provides
Model is the analysis model set up using conventional analysis method and the corresponding algorithm of the analysis model is also conventional
Algorithm, details are not described herein.The analysis model is mainly used for the modeling solicited message to active user and carries out detailed analysis, obtains
To corresponding analysis as a result, and based on the analysis as a result, corresponding training pattern is targetedly constructed, in such manner, it is possible to meet
The demand of different user.
In an optional example, model application message is included at least with the next item down: artificial intelligence training to be built
The corresponding input paramete information of model, wherein input paramete information includes at least text input parameter information, video file inputs
It is one or more in parameter information, audio file input paramete information；Artificial intelligence training pattern to be built is corresponding defeated
Parameter information out, wherein output parameter information includes at least text output parameter information, video file output parameter information, sound
It is one or more in frequency file output parameter information；The corresponding first model object letter of artificial intelligence training pattern to be built
Breath, wherein the first model object information includes at least the learning framework information, corresponding with learning framework for carrying out model training
The first training pattern information at least one of；In this way, according to model application message, it will be able to so that finally constructing and obtaining
Training pattern be more in line with the demand of each different user, improve user experience.
It should be noted that above-mentioned only enumerate common model application message, it is used only for example, other models
Application message also in the protection scope for the construction method scheme that the embodiment of the present invention 1 provides, will not enumerate herein.
In an optional example, model training information is included at least with the next item down: artificial intelligence training to be built
Used material file information when the corresponding training pattern of model, wherein material file information includes at least material file
File name information, the file type information of material file, the document content information of material file, material file storage location
It is one or more in information；The corresponding profile information of artificial intelligence training pattern to be built, wherein configuration file letter
Breath include at least for describing the document structure information of the file structure of material file, for describing training pattern when it is used
At least one of in the frame information of frame；Artificial intelligence training pattern to be built is corresponding for adjusting the super of training pattern
Parameter information；The corresponding second model object information of artificial intelligence training pattern to be built, wherein the second model object information
Including at least the second training pattern information trained；In this way, according to model training information, it will be able to so that final construct and obtain
To training pattern be more in line with the demand of each different user, improve user experience.
In practical applications, no matter for which kind of model training, entire process flow is all identical or similar.
Such as dichotomy model training, two class file of material A, B is written with material structure and frame used, Yong Hutong in configuration file
Overregulating hyper parameter can be carried out model training；In this way, process flow formation mode, finally simplifies existing training
The building process of model.
It should be noted that above-mentioned only enumerate common model training information, it is used only for example, other models
Training information also in the protection scope for the construction method scheme that the embodiment of the present invention 1 provides, will not enumerate herein.
In an optional example, the method also includes: according to preset evaluation rule, training pattern is commented
Valence obtains corresponding evaluation result, and according to evaluation result, optimizes to training pattern, the training pattern after being optimized；
In this way, can constantly be optimized to the training pattern initially obtained, the training pattern after being optimized.
It should be noted that preset evaluation rule is normal in the scheme for the construction method that the embodiment of the present invention 1 provides
The rule for being used to evaluate the training pattern constructed of rule, details are not described herein.
In practical applications, corresponding to comment in addition to by conventional use of every evaluation index and corresponding judgement schematics
Except the evaluation rule of valence model composition, new evaluation index can also be introduced, and to introducing according to the demand of different user
The weight of evaluation index carries out assignment, so that finally obtained evaluation result is more accurate, and evaluation result is also with more sticking on
Together in the demand of active user, user experience is improved.
In addition, in the scheme for the construction method that the embodiment of the present invention 1 provides, side that training pattern is continued to optimize
Method is also conventional method, and details are not described herein, is exactly based on and constantly optimizes to the training pattern initially obtained, finally obtains
Performance more preferably optimize after training pattern.
In conclusion a kind of artificial intelligence training pattern construction method that the embodiment of the present invention 1 provides, has beneficial below
Effect: can intellectual analysis go out the modeling solicited message of user, and targetedly construct corresponding training pattern, simplify existing
Artificial intelligence training pattern building process, improve user experience.
In addition, the construction method that the embodiment of the present invention 1 provides, carries out software development without user, use very succinctly,
The professional threshold for reducing building training pattern, allows more users that can experience the building enjoyment of artificial intelligence training pattern,
User experience is improved, the training pattern constructed is also more abundant, can satisfy the different demands of different user, is applicable in model
It encloses wide.
Furthermore model is applied and the service layer of model training carries out height by the construction method that the embodiment of the present invention 1 provides
Degree is abstract, includes wherein, unified interface to be provided a user, to reach the building of simplified training pattern by each learning framework
Journey reduces the professional threshold of building training pattern.
Embodiment 2
Embodiment according to the present invention 2 additionally provides a kind of artificial intelligence training pattern construction device, as shown in Fig. 2, being
A kind of structural schematic diagram for artificial intelligence training pattern construction device that the embodiment of the present invention 2 provides.The embodiment of the present invention 2 provides
A kind of artificial intelligence training pattern construction device include acquiring unit 201, task generation unit 202, analytical unit 203 and instruction
Practice model construction unit 204.
Specifically, acquiring unit 201, obtains the corresponding with artificial intelligence training pattern to be built of active user and builds
Mould solicited message；
Task generation unit 202, the modeling solicited message got according to acquiring unit 201 generate corresponding training mould
Type constructs task；
Analytical unit 203 is analyzed modeling solicited message according to the analysis model of default training pattern, is obtained corresponding
Analysis result, wherein analysis result include at least artificial intelligence training pattern to be built model application message and model
Training information；
Training pattern construction unit 204, the model application message analyzed according to analytical unit 203 and model training letter
Breath executes the training pattern that task generation unit 202 generates and constructs task, constructs and obtain corresponding training pattern；In this way, logical
The construction device for crossing the offer of the embodiment of the present invention 2, can accomplish: can intellectual analysis go out the modeling solicited message of user, and have
Specific aim constructs corresponding training pattern, simplifies existing artificial intelligence training pattern building process, improves user experience
Degree.
In an optional example, the model application message that analytical unit 203 analyzes is included at least with the next item down: to
The corresponding input paramete information of artificial intelligence training pattern of building, wherein input paramete information is joined including at least text input
It is number information, video file input paramete information, one or more in audio file input paramete information；Artificial intelligence to be built
The corresponding output parameter information of energy training pattern, wherein output parameter information includes at least text output parameter information, video text
It is one or more in part output parameter information, audio file output parameter information；Artificial intelligence training pattern pair to be built
The the first model object information answered, wherein the first model object information includes at least the learning framework for carrying out model training
At least one of in information, the first training pattern information corresponding with learning framework.
In an optional example, the model training information that analytical unit 203 analyzes is included at least with the next item down: to
Used material file information when the corresponding training pattern of the artificial intelligence training pattern of building, wherein material file information
Including at least the file name information of material file, the file type information of material file, material file document content information,
It is one or more in the storage location information of material file；The corresponding configuration file letter of artificial intelligence training pattern to be built
Breath, wherein profile information is included at least for describing the document structure information of the file structure of material file, for describing
When training pattern in the frame information of used frame at least one of；The corresponding use of artificial intelligence training pattern to be built
In the hyper parameter information for adjusting training pattern；The corresponding second model object information of artificial intelligence training pattern to be built,
In, the second model object information includes at least the second training pattern information trained.
In an optional example, described device further includes that evaluation unit (being not shown in Fig. 2) and training pattern are excellent
Change unit (being not shown in Fig. 2)；Evaluation unit obtains training pattern construction unit 204 according to preset evaluation rule
Training pattern is evaluated, and corresponding evaluation result is obtained；Training pattern optimizes unit, the evaluation knot obtained according to evaluation unit
Fruit, the training pattern obtained to training pattern construction unit 204 optimize, the training pattern after being optimized.
Part in the partial content in scheme that the embodiment of the present invention 2 provides and the scheme of the offer of the embodiment of the present invention 1
The same or similar part of content, please be referring to the description of the corresponding portion for the embodiment of the present invention 1, and details are not described herein.
In conclusion a kind of artificial intelligence training pattern construction device that the embodiment of the present invention 2 provides, has beneficial below
Effect: can intellectual analysis go out the modeling solicited message of user, and targetedly construct corresponding training pattern, simplify existing
Artificial intelligence training pattern building process, improve user experience.
In addition, the construction device that the embodiment of the present invention 2 provides, carries out software development without user, use very succinctly,
The professional threshold for reducing building training pattern, allows more users that can experience the building enjoyment of artificial intelligence training pattern,
User experience is improved, the training pattern constructed is also more abundant, can satisfy the different demands of different user, is applicable in model
It encloses wide.
Furthermore model is applied and the service layer of model training carries out height by the construction device that the embodiment of the present invention 2 provides
Degree is abstract, includes wherein, unified interface to be provided a user, to reach the building of simplified training pattern by each learning framework
Journey reduces the professional threshold of building training pattern.
Embodiment 3
Embodiment according to the present invention 3, additionally provides a kind of electronic equipment, and the electronic equipment includes: memory and place
Device is managed, the processor and the memory complete mutual communication by bus；The memory is stored with can be described
The program instruction that processor executes, the processor call described program instruction to be able to carry out following method: obtaining active user
Modeling solicited message corresponding with artificial intelligence training pattern to be built；According to modeling solicited message, corresponding instruction is generated
Practice model construction task；Modeling solicited message is analyzed according to the analysis model of default training pattern, is divided accordingly
Analyse result, wherein analysis result includes at least the model application message and model training of artificial intelligence training pattern to be built
Information；According to model application message and model training information, executes training pattern and construct task, construct and trained accordingly
Model.
Part in the partial content in scheme that the embodiment of the present invention 3 provides and the scheme of the offer of the embodiment of the present invention 1
The same or similar part of content, please be referring to the description of the corresponding portion for the embodiment of the present invention 1, and details are not described herein.
In conclusion a kind of electronic equipment that the embodiment of the present invention 3 provides, having the advantages that intelligently to divide
The modeling solicited message of user is precipitated, and targetedly constructs corresponding training pattern, simplifies existing artificial intelligence training
Model construction process improves user experience.
In addition, a kind of electronic equipment that the embodiment of the present invention 3 provides, carries out software development without user, uses very
Succinctly, the professional threshold for reducing building training pattern, allows more users that can experience the building of artificial intelligence training pattern
Enjoyment improves user experience, and the training pattern constructed is also more abundant, can satisfy the different demands of different user,
It is applied widely.
Furthermore the embodiment of the present invention 3 provide a kind of electronic equipment, by model apply and model training service layer into
Each learning framework is included wherein, unified interface to be provided a user, to reach the structure of simplified training pattern by row high abstraction
Process is built, the professional threshold of building training pattern is reduced.
Embodiment 4
Embodiment according to the present invention 4 additionally provides a kind of computer readable storage medium, is stored thereon with computer journey
Sequence, the computer program realize following method when being executed by processor: obtaining active user with artificial intelligence to be built
The corresponding modeling solicited message of training pattern；According to modeling solicited message, corresponding training pattern building task is generated；According to pre-
If the analysis model of training pattern analyzes modeling solicited message, obtain analyzing result accordingly, wherein analysis result is extremely
It less include the model application message and model training information of artificial intelligence training pattern to be built；According to model application message and
Model training information executes training pattern and constructs task, constructs and obtain corresponding training pattern.
Part in the partial content in scheme that the embodiment of the present invention 4 provides and the scheme of the offer of the embodiment of the present invention 1
The same or similar part of content, please be referring to the description of the corresponding portion for the embodiment of the present invention 1, and details are not described herein.
In conclusion a kind of computer readable storage medium that the embodiment of the present invention 4 provides, has the advantages that
Can intellectual analysis go out the modeling solicited message of user, and targetedly construct corresponding training pattern, simplify existing people
Work intelligent training model construction process, improves user experience.
In addition, a kind of computer readable storage medium that the embodiment of the present invention 4 provides, carries out software development without user,
The professional threshold for very succinctly reducing building training pattern is used, allows more users that can experience artificial intelligence training
The building enjoyment of model, improves user experience, and the training pattern constructed is also more abundant, can satisfy different user
Different demands, it is applied widely.
Furthermore a kind of computer readable storage medium that the embodiment of the present invention 4 provides applies model and model training
Service layer carries out high abstraction, includes wherein, unified interface to be provided a user, to reach simplified instruction by each learning framework
The building process for practicing model reduces the professional threshold of building training pattern.
Although above having used general explanation and specific embodiment, the present invention is described in detail, at this
On the basis of invention, it can be made some modifications or improvements, this will be apparent to those skilled in the art.Therefore,
These modifications or improvements without departing from theon the basis of the spirit of the present invention are fallen within the scope of the claimed invention.
Claims (10)
1. a kind of artificial intelligence training pattern construction method characterized by comprising
Obtain the modeling solicited message corresponding with artificial intelligence training pattern to be built of active user；
According to the modeling solicited message, corresponding training pattern building task is generated；
The modeling solicited message is analyzed according to the analysis model of default training pattern, analyzed accordingly as a result,
Wherein, the analysis result includes at least the model application message and model training of the artificial intelligence training pattern to be built
Information；
According to the model application message and the model training information, the training pattern building task is executed, constructs and obtains
To corresponding training pattern.
2. the method according to claim 1, wherein
The model application message is included at least with the next item down:
The corresponding input paramete information of artificial intelligence training pattern to be built, wherein the input paramete information is at least
Including one or more in text input parameter information, video file input paramete information, audio file input paramete information；
The corresponding output parameter information of artificial intelligence training pattern to be built, wherein the output parameter information is at least
Including one or more in text output parameter information, video file output parameter information, audio file output parameter information；
The corresponding first model object information of the artificial intelligence training pattern to be built, wherein first model object
Information includes at least learning framework information, the first training pattern corresponding with the learning framework for carrying out model training and believes
At least one of in breath.
3. the method according to claim 1, wherein
The model training information is included at least with the next item down:
Used material file information when the corresponding training pattern of the artificial intelligence training pattern to be built, wherein institute
State material file information including at least the file name information of the material file, the material file file type information,
The document content information of the material file, the material file storage location information in it is one or more；
The corresponding profile information of artificial intelligence training pattern to be built, wherein the profile information is at least
Document structure information including the file structure for describing the material file, for describing training pattern when used frame
At least one of in the frame information of frame；
The artificial intelligence training pattern to be built is corresponding for adjusting the hyper parameter information of training pattern；
The corresponding second model object information of the artificial intelligence training pattern to be built, wherein second model object
Information includes at least the second training pattern information trained.
4. the method according to claim 1, wherein the method also includes:
According to preset evaluation rule, the training pattern is evaluated, obtains corresponding evaluation result, and according to institute's commentary
Valence is as a result, optimize the training pattern, the training pattern after being optimized.
5. a kind of artificial intelligence training pattern construction device characterized by comprising
Acquiring unit obtains the modeling solicited message corresponding with artificial intelligence training pattern to be built of active user；
Task generation unit, the modeling solicited message according to the information acquired by the acquisition unit, generates corresponding training pattern
Building task；
Analytical unit analyzes the modeling solicited message according to the analysis model of default training pattern, obtains corresponding
Analyze result, wherein the analysis result includes at least the model application message of the artificial intelligence training pattern to be built
With model training information；
Training pattern construction unit, the model application message analyzed according to the analytical unit and model training letter
Breath executes the training pattern building task that the task generation unit generates, constructs and obtain corresponding training pattern.
6. device according to claim 5, which is characterized in that
The model application message that the analytical unit analyzes is included at least with the next item down:
The corresponding input paramete information of artificial intelligence training pattern to be built, wherein the input paramete information is at least
Including one or more in text input parameter information, video file input paramete information, audio file input paramete information；
The corresponding output parameter information of artificial intelligence training pattern to be built, wherein the output parameter information is at least
Including one or more in text output parameter information, video file output parameter information, audio file output parameter information；
The corresponding first model object information of the artificial intelligence training pattern to be built, wherein first model object
Information includes at least learning framework information, the first training pattern corresponding with the learning framework for carrying out model training and believes
At least one of in breath.
7. device according to claim 5, which is characterized in that
The model training information that the analytical unit analyzes is included at least with the next item down:
Used material file information when the corresponding training pattern of the artificial intelligence training pattern to be built, wherein institute
State material file information including at least the file name information of the material file, the material file file type information,
The document content information of the material file, the material file storage location information in it is one or more；
The corresponding profile information of artificial intelligence training pattern to be built, wherein the profile information is at least
Document structure information including the file structure for describing the material file, for describing training pattern when used frame
At least one of in the frame information of frame；
The artificial intelligence training pattern to be built is corresponding for adjusting the hyper parameter information of training pattern；
The corresponding second model object information of the artificial intelligence training pattern to be built, wherein second model object
Information includes at least the second training pattern information trained.
8. device according to claim 5, which is characterized in that described device further include:
Evaluation unit carries out the training pattern that the training pattern construction unit obtains according to preset evaluation rule
Evaluation, obtains corresponding evaluation result；
Training pattern optimization unit obtains the training pattern construction unit according to the evaluation result that the evaluation unit obtains
To the training pattern optimize, the training pattern after being optimized.
9. a kind of electronic equipment characterized by comprising
Memory and processor, the processor and the memory complete mutual communication by bus；The memory
It is stored with the program instruction that can be executed by the processor, the processor calls described program instruction to be able to carry out right such as and wants
Seek 1 to 4 any method.
10. a kind of computer readable storage medium, which is characterized in that be stored thereon with computer program, the computer program
The step of the method as any such as Claims 1-4 is realized when being executed by processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811481871.4A CN109711436A (en) | 2018-12-05 | 2018-12-05 | A kind of artificial intelligence training pattern construction method, device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811481871.4A CN109711436A (en) | 2018-12-05 | 2018-12-05 | A kind of artificial intelligence training pattern construction method, device and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109711436A true CN109711436A (en) | 2019-05-03 |
Family
ID=66255432
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811481871.4A Pending CN109711436A (en) | 2018-12-05 | 2018-12-05 | A kind of artificial intelligence training pattern construction method, device and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109711436A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113516200A (en) * | 2021-07-30 | 2021-10-19 | 盛景智能科技(嘉兴)有限公司 | Method and device for generating model training scheme, electronic equipment and storage medium |
| CN113516200B (en) * | 2021-07-30 | 2024-06-04 | 盛景智能科技(嘉兴)有限公司 | Model training scheme generation method and device, electronic equipment and storage medium |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101782976A (en) * | 2010-01-15 | 2010-07-21 | 南京邮电大学 | Automatic selection method for machine learning in cloud computing environment |
| WO2015183442A1 (en) * | 2014-05-30 | 2015-12-03 | Apple Inc. | Methods and system for managing predictive models |
| CN106161620A (en) * | 2016-06-29 | 2016-11-23 | 浙江理工大学 | A kind of cloud computing resources Internet of Things supervision and service platform |
| CN106779087A (en) * | 2016-11-30 | 2017-05-31 | 福建亿榕信息技术有限公司 | A kind of general-purpose machinery learning data analysis platform |
| CN107563417A (en) * | 2017-08-18 | 2018-01-09 | 北京天元创新科技有限公司 | A kind of deep learning artificial intelligence model method for building up and system |
| CN107705229A (en) * | 2017-09-27 | 2018-02-16 | 镇江博格信息科技有限公司 | A kind of online training management system and method |
| CN107766940A (en) * | 2017-11-20 | 2018-03-06 | 北京百度网讯科技有限公司 | Method and apparatus for generation model |
| CN108073760A (en) * | 2016-11-10 | 2018-05-25 | 通用电气公司 | For obtaining the method and system that analysis model writes knowledge |
| CN108093030A (en) * | 2017-11-29 | 2018-05-29 | 杭州古北电子科技有限公司 | A kind of artificial intelligence model dispositions method based on Cloud Server |
-
2018
- 2018-12-05 CN CN201811481871.4A patent/CN109711436A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101782976A (en) * | 2010-01-15 | 2010-07-21 | 南京邮电大学 | Automatic selection method for machine learning in cloud computing environment |
| WO2015183442A1 (en) * | 2014-05-30 | 2015-12-03 | Apple Inc. | Methods and system for managing predictive models |
| CN106161620A (en) * | 2016-06-29 | 2016-11-23 | 浙江理工大学 | A kind of cloud computing resources Internet of Things supervision and service platform |
| CN108073760A (en) * | 2016-11-10 | 2018-05-25 | 通用电气公司 | For obtaining the method and system that analysis model writes knowledge |
| CN106779087A (en) * | 2016-11-30 | 2017-05-31 | 福建亿榕信息技术有限公司 | A kind of general-purpose machinery learning data analysis platform |
| CN107563417A (en) * | 2017-08-18 | 2018-01-09 | 北京天元创新科技有限公司 | A kind of deep learning artificial intelligence model method for building up and system |
| CN107705229A (en) * | 2017-09-27 | 2018-02-16 | 镇江博格信息科技有限公司 | A kind of online training management system and method |
| CN107766940A (en) * | 2017-11-20 | 2018-03-06 | 北京百度网讯科技有限公司 | Method and apparatus for generation model |
| CN108093030A (en) * | 2017-11-29 | 2018-05-29 | 杭州古北电子科技有限公司 | A kind of artificial intelligence model dispositions method based on Cloud Server |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113516200A (en) * | 2021-07-30 | 2021-10-19 | 盛景智能科技(嘉兴)有限公司 | Method and device for generating model training scheme, electronic equipment and storage medium |
| CN113516200B (en) * | 2021-07-30 | 2024-06-04 | 盛景智能科技(嘉兴)有限公司 | Model training scheme generation method and device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110198237B (en) | Configuration method of wireless intention driven network | |
| US20190122409A1 (en) | Multi-Dimensional Puppet with Photorealistic Movement | |
| CN107103113A (en) | Towards the Automation Design method, device and the optimization method of neural network processor | |
| CN109213482A (en) | The graphical application platform of artificial intelligence and application method based on convolutional neural networks | |
| CN106777971A (en) | A kind of intelligent hospital guide's method and its device | |
| CN114721833A (en) | Intelligent cloud coordination method and device based on platform service type | |
| CN107451230A (en) | A kind of answering method and question answering system | |
| CN106886408A (en) | It is applied to the intelligent decision system of robot software engineer | |
| WO2020253180A1 (en) | Smart home decision support system and decision support method | |
| CN104866310B (en) | The processing method and system of knowledge data | |
| CN109284386A (en) | Customized intension recognizing method and device | |
| CN106934452A (en) | Robot dialogue method and system | |
| Qian et al. | AIEM: AI-enabled affective experience management | |
| CN109905481A (en) | A kind of building of the Qos model based on RTI-DDS and the runnability prediction technique of Qos strategy protocol | |
| CN109711436A (en) | A kind of artificial intelligence training pattern construction method, device and storage medium | |
| López-Jaquero et al. | A seamless development process of adaptive user interfaces explicitly based on usability properties | |
| Liu et al. | Research on general model of intelligence level for smart home | |
| CN116384240A (en) | Server energy consumption prediction method, device and storage medium | |
| CN110110338A (en) | A kind of Dialogue management model application method based on LSTM and slot filling | |
| CN115878096A (en) | Deep learning model unified application method, device, server and storage medium | |
| CN109949827A (en) | A kind of room acoustics Activity recognition method based on deep learning and intensified learning | |
| CN109492080A (en) | A kind of medical health system and its implementation based on voice response robot | |
| CN107705166A (en) | Information processing method and device | |
| CN110442690A (en) | A kind of query optimization method, system and medium based on probability inference | |
| KR20220016588A (en) | Deep learning inference system based on query, and method thereof |
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: 20190503 |