CN109218432A - A kind of distributed approach and device of intelligence computation cloud service - Google Patents
A kind of distributed approach and device of intelligence computation cloud service Download PDFInfo
- Publication number
- CN109218432A CN109218432A CN201811145896.7A CN201811145896A CN109218432A CN 109218432 A CN109218432 A CN 109218432A CN 201811145896 A CN201811145896 A CN 201811145896A CN 109218432 A CN109218432 A CN 109218432A
- Authority
- CN
- China
- Prior art keywords
- task
- algorithm
- classification
- intelligent
- characteristic information
- 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.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/50—Network services
- H04L67/60—Scheduling or organising the servicing of application requests, e.g. requests for application data transmissions using the analysis and optimisation of the required network resources
Abstract
This application involves a kind of distributed approaches of intelligence computation cloud service, comprising: receives task processing request；Match the task generic；It is called according to the task generic and is best suitable for the intelligent algorithm for handling the task requests in the algorithm classification for handling the task requests.Due to being classified according to the characteristic information of intelligent algorithm to different intelligent algorithms, after receiving task requests, some intelligent algorithm is not called directly, but first call the intelligent algorithm classification for handling the task requests, the intelligent algorithm that processing task requests are best suitable in the algorithm classification of processing task requests is called according to task requests generic, thus more preferably, more flexible, more intelligent provide most suitable service for client.
Description
Technical field
This application involves intelligent algorithm field more particularly to a kind of distributed treatment sides of intelligence computation cloud service
Method and device.
Background technique
Mostly software function service is in online or local area network for the client of calling service at present
It is fully transparent sightless, according to the operating condition of each real server, determine the server for really carrying out processing operation, place
After having managed, client is returned to.
The service of identical content can often be realized by various intelligent algorithm, however these algorithms exist
There is very big differences in performance and in accuracy, respectively have superiority and inferiority, there is also difference in consumption, client exists various algorithms
Algorithm that is more preferable, being more suitable for oneself request can not be selected during calling service.
Summary of the invention
To be overcome the problems, such as present in the relevant technologies at least to a certain extent, the application provides a kind of intelligence computation cloud
The distributed approach and device of service.
The scheme of the application is as follows:
According to the embodiment of the present application in a first aspect, provide a kind of distributed approach of intelligence computation cloud service,
Include:
The processing request of reception task；
Match the task generic；
It is called according to the task generic and is best suitable for processing described in the algorithm classification for handling the task requests
The intelligent algorithm of business request.
Preferably, in conjunction with the above content, in a kind of possible implementation of the application, in the matching task
Before generic, further includes:
Judge whether cloud has the corresponding classification of the task；
If establishing new classification of task no, using the task as template；
If so, then matching the task generic.
Preferably, in conjunction with the above content, in a kind of possible implementation of the application,
The matching task generic includes:
Judge whether the task function classification is consistent with existing task category calling interface；
If consistent, it is included into and has classification；
If inconsistent, shared interface and peculiar interface are concluded.
Preferably, in conjunction with the above content, in a kind of possible implementation of the application, the reception task requests
Before, the method also includes: classify to task.
Preferably, described to classify to task in a kind of possible implementation of the application in conjunction with the above content
Include:
Classified according to the definition of the function of the task to the task；And/or
Classified according to the composition definition of the parameter of the task to the task；And/or
Classified according to the characteristic information of the task to the task.
Preferably, in conjunction with the above content, in a kind of possible implementation of the application, the reception task requests
Before, the method also includes: algorithm classification is carried out to different intelligent algorithms according to the characteristic information of intelligent algorithm.
Preferably, described according to intelligent algorithm in a kind of possible implementation of the application in conjunction with the above content
After characteristic information carries out algorithm classification to different intelligent algorithms, the method also includes: monitor the feature of the intelligent algorithm
Information.
Preferably, described according to intelligent algorithm in a kind of possible implementation of the application in conjunction with the above content
After characteristic information carries out algorithm classification to different intelligent algorithms, the method also includes: to the feature letter of the intelligent algorithm
The maintenance and publication of breath.
Preferably, described to the intelligent algorithm in a kind of possible implementation of the application in conjunction with the above content
Characteristic information maintenance and publication include: by centralized architecture, or, feature of the decentralization structure to the intelligent algorithm
Information is safeguarded and is issued.
According to the second aspect of the embodiment of the present application, a kind of intelligence computation cloud service device is provided, comprising: processing mould
Block, receiving module, calling module；
The processing module, for carrying out algorithm classification to different intelligent algorithms according to the characteristic information of intelligent algorithm；
The receiving module, for receiving task requests；
The calling module is most accorded with for being called in the algorithm classification for handling the task requests according to the task requests
Close the intelligent algorithm for handling the task requests.
Technical solution provided by the present application can include the following benefits: due to the characteristic information pair according to intelligent algorithm
Different intelligent algorithms are classified, and after receiving task requests, do not call directly some intelligent algorithm, but first call processing should
The intelligent algorithm of task requests is classified, and is called according to task requests generic and is best suitable in the algorithm classification of processing task requests
Handle the intelligent algorithm of task requests, thus more preferably, more flexible, more intelligent provide most suitable service for client.
It should be understood that above general description and following detailed description be only it is exemplary and explanatory, not
The application can be limited.
Detailed description of the invention
The drawings herein are incorporated into the specification and forms part of this specification, and shows the implementation for meeting the application
Example, and together with specification it is used to explain the principle of the application.
Fig. 1 is a kind of process of the distributed approach for intelligence computation cloud service that the application one embodiment provides
Figure；
Fig. 2 is a kind of stream of the distributed approach for intelligence computation cloud service that another embodiment of the application provides
Cheng Tu；
Fig. 3 is a kind of stream of the distributed approach for intelligence computation cloud service that another embodiment of the application provides
Cheng Tu；
Fig. 4 is a kind of structure chart for intelligence computation cloud service device that the application one embodiment provides.
Specific embodiment
Example embodiments are described in detail here, and the example is illustrated in the accompanying drawings.Following description is related to
When attached drawing, unless otherwise indicated, the same numbers in different drawings indicate the same or similar elements.Following exemplary embodiment
Described in embodiment do not represent all embodiments consistent with the application.On the contrary, they be only with it is such as appended
The distributed approach of the consistent intelligence computation cloud service of some aspects be described in detail in claims, the application
And the example of device.
Fig. 1 is the flow chart of the distributed approach for the intelligence computation cloud service that the application one embodiment provides.
Intelligent algorithm has its unique characteristic.In the application of every field, we can conclude intelligent algorithm
At some feature contents for having very strong versatility, the intelligent task of identical function content, corresponding various realization side
Method.For example: face recognition algorithms there is many technology schools and realize version.All versions can be realized
The function of recognition of face.However these algorithms in performance and accuracy on there is very big difference, respectively have superiority and inferiority；Various calculations
There is also differences in consumption for method.
The embodiment of the present invention provides a kind of distributed approach of intelligence computation cloud service, by intelligent algorithm mould
For block as a kind of intelligent Service, how more preferable deployment is beyond the clouds, then, more flexible, more intelligent to provide Intellectual garment for client
Business.
Referring to Fig. 1, this method comprises:
101: receiving task processing request；
It is triggered according to user's operation, client sends task processing request to cloud, and cloud receives appointing for client transmission
Business processing request, task processing request include task feature and task parameters etc..
102: matching task generic；
Parameter composition can be divided into call parameter, optional parameters.Such as:
The task type that client is sent be it is diversified, the task type that cloud is sent according to client matches this
The business classification can preferably handle the task processing request of client transmission.
103: being called according to task generic and be best suitable for processing task requests in the algorithm classification of processing task requests
Intelligent algorithm.
The task requests of identical content can often be realized by various intelligent algorithm, however these are calculated
Method there is very big difference, respectively there is a superiority and inferiority in performance and in accuracy, various algorithms in consumption there is also difference, due to
Classified according to the characteristic information of intelligent algorithm to different intelligent algorithms, after receiving task requests, does not call directly some
Intelligent algorithm, but the intelligent algorithm classification for handling the task requests is first called, it is handled according to task requests generic calling
Be best suitable in the algorithm classification of task requests processing task requests intelligent algorithm, thus more preferably, it is more flexible, more intelligent for visitor
Family end provides most suitable service.
As a kind of improvement of above-described embodiment, the embodiment of the present invention provides the distribution of another intelligence computation cloud service
Formula processing method, referring to Fig. 2-Fig. 3, this method comprises:
201: classifying to task.
Specifically, it may include one of following three kinds of mode classifications or a variety of for carrying out classification to task:
The first mode classification classifies to task according to the definition of the function of task；
Second of mode classification classifies to task according to the composition definition of the parameter of task；
The third mode classification classifies to task according to the characteristic information of task.
It should be noted that this case, in specific implementation process kind, mode classification is not limited to above-mentioned three kinds, it can be by this field
Technical staff carries out other definition according to the actual situation.
In order to be assigned to suitable server to each intelligent task that client is sent, beyond the clouds, we are to various intelligence
Energy task is all classified, and is collected and is established many task types.For specific intelligent task, the foundation of classification
It may include: function definition, the definition of parameter composition and characteristic information composition.
Function, which defines, mainly to be existed in the form of text, while the definition of each concrete function can have unique function fixed
Justice number, such as:
| Definition number | Function definition |
| 7429472231122 | Whether include the face in default face database in identification input still image. |
Parameter group ingredient is call parameter, optional parameters.Whether to include default face database in identification input still image
In face for:
The characteristic information of intelligent task classification is used to the different intelligent tasks within description same category and calculates after definition
The otherness of method module from each other.
The characteristic information of classification of task, including parameter attribute, algorithm characteristics, generic features.
Parameter attribute derives from all nonessential optional parameters, including outputs and inputs；
Algorithm characteristics derive from the specific business tine of intelligent task algorithm, such as face recognition algorithms, optional algorithm
Parameter includes mAP, FPS etc..
Generic features are features common to different intelligent classification of task.It mainly include service response time, carrier
Load evaluation is the case where running the calculation resources on the hardware of intelligent task, and the assessment of carrier operation efficiency is that intelligent algorithm is carrying
It is run on body, whether the consumption of bearer resource is matched, for example stressed the algorithm of central processing unit resource and stressing graphics process
It is run on the hardware of device just more unreasonable.
Cloud has been completed the classification to task, and different tasks needs different algorithms to be handled, according to algorithm
Function and the task category of processing classify to algorithm, so that each algorithm is belonged to some classification of task, i.e., often
A classification of task has with the algorithm that can handle its corresponding task requests.
Algorithm characteristics information is used to adequately describe the otherness of each algoritic module, and this otherness information is provided
To client, client is allowed to wish to be analyzed by the way of using any its, chosen.
After carrying out algorithm classification to different intelligent algorithms according to the characteristic information of intelligent algorithm, method further include: monitoring
The characteristic information of intelligent algorithm.
There is mistake in the characteristic information for occurring the intelligent algorithm stored in order to prevent, when solving task requests, calls wrong
The problems such as intelligent algorithm missed, after carrying out algorithm classification to different intelligent algorithms, cloud also needs the spy to intelligent algorithm
Reference breath is monitored, to guarantee that the characteristic information of each intelligent algorithm of storage is accurate.
The characteristic information of algorithm can be divided into static types and regime type.
The characteristic information of the algorithm of static types does not need ceaselessly to monitor, it is only necessary to guarantee to complete the original definition of algorithm
?；
The characteristic information of the algorithm of regime type, needs cloud according to configuration, is regularly monitored or monitor log
It is true each time to call record, the maintenance features value of information.
Further, when being monitored to the characteristic information of intelligent algorithm, it is also necessary to be maintained and issue.
The characteristic information of each intelligent task algorithm by client query and will determine final calling object.Beyond the clouds
In service system, how storage system maintenance these characteristic informations, and the mode for being distributed to all clients is broadly divided into two classes: in
Heart structure, decentralization structure.
The maintenance of centralized architecture and published method belong to the traditional mode of comparison.System centre includes feeling concerned about in one
System after some intelligent algorithm of all node deployments, obtains specific algorithm characteristics information, node sends out the characteristic information of algorithm
Give centring system.Centring system is responsible for safeguarding these information, and is distributed to all clients.All clients are unified from center
System obtains the characteristic information of all algorithms.
Decentralization structure be using the consistency algorithm based on Paxos algorithm ensure distributed storage information it is consistent
Property.The characteristic information of algorithm by it is distributed storage in the entire system, decentralization structure theoretically, can permit whether there is or not
A server access port is limited, client can connect any one server access port therein, or the help in service
Under, selection accesses the most suitable server of client, to obtain the characteristic information of all algorithms.
The difference of centralized architecture and decentralization structure is that centralization will lead to only one server publication letter
Breath, the information big for real-time height and data volume can not be handled in time, and effect is more preferable at this time for decentralization structure,
But cost is higher.
Intelligent algorithm is safeguarded and issued by the way of being suitble to according to actual needs in cloud.
101: receiving task processing request；
It is triggered according to user's operation, client sends task processing request to cloud, and cloud receives appointing for client transmission
Business processing request, task processing request include task feature and task parameters etc..
202: judging that the whether existing corresponding classification of task in cloud thens follow the steps 203 if not having；If so, then executing step
Rapid 204.
203: using task as template, establishing new classification of task；
204: matching task generic；
The task type that client is sent be it is diversified, the task type that cloud is sent according to client matches this
The business classification can preferably handle the task processing request of client transmission.
Matching task generic includes:
Judge whether task function classification is consistent with existing task category calling interface；
If consistent, it is included into and has classification；
If inconsistent, shared interface and peculiar interface are concluded.
The calling interface of similar intelligent algorithm needs consistent, i.e. intelligent algorithm itself, input data and output data
Content-form needs unification.
The feature content feature of similar intelligent algorithm needs unanimously, i.e., intelligent algorithm solves the problems, such as it is very close
, different places is implementation method, carrier, the place difference of function.
Wherein, implementation method difference can be, but not limited to as algorithm difference；
Carrier difference can be, but not limited to as differences such as server, mobile devices；
Place difference can be, but not limited to as specific client end difference.
Although cloud has the classification of most of task requests, it is not excluded for client and sends new task requests, institute
After receiving task requests, to judge whether cloud has the corresponding classification of the task, if cloud is not present and the task function
The energy consistent classification of content establishes new classification of task then using the task as template；
If cloud exist with the consistent classification of task function content, match the task generic.
Further,
103: being called according to task generic and be best suitable for processing task requests in the algorithm classification of processing task requests
Intelligent algorithm.
A kind of intelligence computation cloud service device, referring to Fig. 4, comprising: processing module 31, receiving module 32, calling module
33；
Processing module 31, for carrying out algorithm classification to different intelligent algorithms according to the characteristic information of intelligent algorithm；
Receiving module 32, for receiving task requests；
Calling module 33 is best suitable for processing task for calling in the algorithm classification for handling task requests according to task requests
The intelligent algorithm of request.
It is understood that same or similar part can mutually refer in the various embodiments described above, in some embodiments
Unspecified content may refer to the same or similar content in other embodiments.
It should be noted that term " first ", " second " etc. are used for description purposes only in the description of the present application, without
It can be interpreted as indication or suggestion relative importance.In addition, in the description of the present application, unless otherwise indicated, the meaning of " multiple "
Refer at least two.
Any process described otherwise above or method description are construed as in flow chart or herein, and expression includes
It is one or more for realizing specific logical function or process the step of executable instruction code module, segment or portion
Point, and the range of the preferred embodiment of the application includes other realization, wherein can not press shown or discussed suitable
Sequence, including according to related function by it is basic simultaneously in the way of or in the opposite order, to execute function, this should be by the application
Embodiment person of ordinary skill in the field understood.
In the description of this specification, reference term " one embodiment ", " some embodiments ", " example ", " specifically show
The description of example " or " some examples " etc. means specific features, structure, material or spy described in conjunction with this embodiment or example
Point is contained at least one embodiment or example of the application.In the present specification, schematic expression of the above terms are not
Centainly refer to identical embodiment or example.Moreover, particular features, structures, materials, or characteristics described can be any
One or more embodiment or examples in can be combined in any suitable manner.
Although embodiments herein has been shown and described above, it is to be understood that above-described embodiment is example
Property, it should not be understood as the limitation to the application, those skilled in the art within the scope of application can be to above-mentioned
Embodiment is changed, modifies, replacement and variant.
Claims (10)
1. a kind of distributed approach of intelligence computation cloud service characterized by comprising
The processing request of reception task；
Match the task generic；
It is asked according to the processing task is best suitable in the algorithm classification of the task generic calling processing task requests
The intelligent algorithm asked.
2. the method according to claim 1, wherein also being wrapped before the matching task generic
It includes:
Judge whether cloud has the corresponding classification of the task；
If establishing new classification of task no, using the task as template；
If so, then matching the task generic.
3. according to the method described in claim 2, it is characterized in that,
The matching task generic includes:
Judge whether the task function classification is consistent with existing task category calling interface；
If consistent, it is included into and has classification；
If inconsistent, shared interface and peculiar interface are concluded.
4. the method according to claim 1, wherein before the reception task requests, the method also includes: it is right
Task is classified.
5. according to the method described in claim 4, it is characterized in that, it is described to task carry out classification include:
Classified according to the definition of the function of the task to the task；And/or
Classified according to the composition definition of the parameter of the task to the task；And/or
Classified according to the characteristic information of the task to the task.
6. method according to claim 1-5, which is characterized in that before the reception task requests, the method
Further include: algorithm classification is carried out to different intelligent algorithms according to the characteristic information of intelligent algorithm.
7. according to the method described in claim 6, it is characterized in that, the characteristic information according to intelligent algorithm is to different intelligence
After energy algorithm carries out algorithm classification, the method also includes: monitor the characteristic information of the intelligent algorithm.
8. the method according to the description of claim 7 is characterized in that the characteristic information according to intelligent algorithm is to different intelligence
After energy algorithm carries out algorithm classification, the method also includes: maintenance and publication to the characteristic information of the intelligent algorithm.
9. according to the method described in claim 8, it is characterized in that, the maintenance of the characteristic information to the intelligent algorithm with
Publication includes: by centralized architecture, or, decentralization structure is safeguarded and sent out to the characteristic information of the intelligent algorithm
Cloth.
10. a kind of intelligence computation cloud service device characterized by comprising processing module, receiving module, calling module；
The processing module, for carrying out algorithm classification to different intelligent algorithms according to the characteristic information of intelligent algorithm；
The receiving module, for receiving task requests；
The calling module is best suitable for place for calling in the algorithm classification for handling the task requests according to the task requests
Manage the intelligent algorithm of the task requests.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811145896.7A CN109218432B (en) | 2018-09-29 | 2018-09-29 | Distributed processing method and device for intelligent computing cloud service |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811145896.7A CN109218432B (en) | 2018-09-29 | 2018-09-29 | Distributed processing method and device for intelligent computing cloud service |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109218432A true CN109218432A (en) | 2019-01-15 |
| CN109218432B CN109218432B (en) | 2021-05-14 |
Family
ID=64982260
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811145896.7A Active CN109218432B (en) | 2018-09-29 | 2018-09-29 | Distributed processing method and device for intelligent computing cloud service |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109218432B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030033263A1 (en) * | 2001-07-31 | 2003-02-13 | Reel Two Limited | Automated learning system |
| CN101685459A (en) * | 2008-09-28 | 2010-03-31 | 华为技术有限公司 | Multi-media searching method and device |
| CN103283209A (en) * | 2011-04-18 | 2013-09-04 | 北京新媒传信科技有限公司 | Application service platform system and implementation method thereof |
| CN105138532A (en) * | 2015-06-26 | 2015-12-09 | 深圳市腾讯计算机系统有限公司 | Method, device and system for processing service data |
| CN107615275A (en) * | 2015-05-29 | 2018-01-19 | 国际商业机器公司 | Estimate to excavate the computing resource serviced for service data |
| CN108280091A (en) * | 2017-01-06 | 2018-07-13 | 阿里巴巴集团控股有限公司 | A kind of task requests execution method and apparatus |
-
2018
- 2018-09-29 CN CN201811145896.7A patent/CN109218432B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030033263A1 (en) * | 2001-07-31 | 2003-02-13 | Reel Two Limited | Automated learning system |
| CN101685459A (en) * | 2008-09-28 | 2010-03-31 | 华为技术有限公司 | Multi-media searching method and device |
| CN103283209A (en) * | 2011-04-18 | 2013-09-04 | 北京新媒传信科技有限公司 | Application service platform system and implementation method thereof |
| CN107615275A (en) * | 2015-05-29 | 2018-01-19 | 国际商业机器公司 | Estimate to excavate the computing resource serviced for service data |
| CN105138532A (en) * | 2015-06-26 | 2015-12-09 | 深圳市腾讯计算机系统有限公司 | Method, device and system for processing service data |
| CN108280091A (en) * | 2017-01-06 | 2018-07-13 | 阿里巴巴集团控股有限公司 | A kind of task requests execution method and apparatus |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109218432B (en) | 2021-05-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108418862B (en) | Micro-service management method and system based on artificial intelligence service cloud platform | |
| CN104298983B (en) | Tongue fur image with distributed user terminal obtains and analysis system | |
| US10282366B2 (en) | Multi-dimensional decomposition computing method and system | |
| CN109491784A (en) | Reduce method, apparatus, the electronic equipment, readable storage medium storing program for executing of EMS memory occupation amount | |
| CN101414351A (en) | Fingerprint recognition system and control method | |
| US20180276508A1 (en) | Automated visual information context and meaning comprehension system | |
| CN109672750A (en) | A kind of security protection data aggregate management system and its management method based on Edge fabrics | |
| Jin et al. | Specifying and detecting spatio-temporal events in the internet of things | |
| CN107680662A (en) | Database marketing system and method based on the processing of Hadoop cloud big data | |
| US11062224B2 (en) | Prediction using fusion of heterogeneous unstructured data | |
| CN106302742B (en) | A kind of electrical power services resource information interactive system and method | |
| CN110288465A (en) | Object determines method and device, storage medium, electronic device | |
| CN115729683A (en) | Task processing method, device, system, computer equipment and storage medium | |
| CN111461630B (en) | Monitoring method, device, equipment and storage medium for delivering express packages | |
| Taneja et al. | Predictive analytics on IoT | |
| CN109218432A (en) | A kind of distributed approach and device of intelligence computation cloud service | |
| CN108228355A (en) | Task processing method and device, method for scheduling task and device | |
| CN109558463B (en) | Data processing method and device of intelligent report and storage medium | |
| CN112286662B (en) | Task pushing method, device and equipment based on shared resources and storage medium | |
| CN104754040A (en) | A system used for end-to-end cloud service virtualization | |
| CN106570069A (en) | Pesticide electronic monitoring system | |
| CN112669353A (en) | Data processing method, data processing device, computer equipment and storage medium | |
| CN106845795A (en) | A kind of work flow system and its work flow method based on multi-Agent technology | |
| CN106845790A (en) | A kind of local service system and its local service access method based on multi-Agent technology in single operation system | |
| CN109314697A (en) | Automatic network management system and method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |