CN112925514A - Method for expanding custom algorithm component by machine learning platform - Google Patents
Method for expanding custom algorithm component by machine learning platform Download PDFInfo
- Publication number
- CN112925514A CN112925514A CN202110207888.6A CN202110207888A CN112925514A CN 112925514 A CN112925514 A CN 112925514A CN 202110207888 A CN202110207888 A CN 202110207888A CN 112925514 A CN112925514 A CN 112925514A
- Authority
- CN
- China
- Prior art keywords
- algorithm
- data
- port
- parameters
- parameter
- 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
- 238000010801 machine learning Methods 0.000 title claims abstract description 45
- 238000000034 method Methods 0.000 title claims abstract description 32
- 230000008859 change Effects 0.000 claims description 6
- 230000001419 dependent effect Effects 0.000 claims description 6
- 230000000007 visual effect Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 239000000470 constituent Substances 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F8/00—Arrangements for software engineering
- G06F8/30—Creation or generation of source code
- G06F8/31—Programming languages or programming paradigms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Mathematical Physics (AREA)
- Stored Programmes (AREA)
Abstract
The invention discloses a method for expanding a custom algorithm component by a machine learning platform, which relates to the technical field of component expansion; the method comprises the steps that an algorithm component is set to be a minimum operation unit of a machine learning platform, the algorithm component is defined and constructed and comprises a data port, a logic execution unit and algorithm parameters, the data port comprises a data input port and a data output port, the data input port receives data transmitted to the algorithm component and transmits the data to the logic execution unit, the logic execution unit carries out operation on the received data according to algorithm logic by using the algorithm parameters and transmits an operation execution result to the data output port, and the data output port stores the received execution result or transmits the received execution result to the next algorithm component.
Description
Technical Field
The invention discloses a method for expanding a custom algorithm component by a machine learning platform, relates to the technical field of component expansion, and particularly relates to a method for expanding a custom algorithm component by a machine learning platform.
Background
The machine learning platform plays an important role in the artificial intelligence development process, and provides a series of artificial intelligence enabling services such as data insight, machine learning modeling, prediction evaluation, decision support and the like for enterprises from digitalization to intellectualization transformation. A general machine learning platform provides a general machine learning algorithm, covers data access, data conversion, feature engineering, common machine learning algorithms, model evaluation and the like, can meet development requirements of a part of general scenes, but cannot meet all requirements, particularly cannot meet the requirements of specific industry fields on specific algorithms in the fields.
The existing method supports a user to package a custom algorithm in a docker image to be used by an access platform so as to meet the requirement of personalized development, but requires the user to package the custom algorithm by using a container mirror image technology, depends on technical components such as container mirror images and the like, and has higher development threshold and cost.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a method for expanding a custom algorithm component by a machine learning platform, so that a custom algorithm developed by a user can be conveniently registered to the platform and used based on the platform.
The specific scheme provided by the invention is as follows:
a method for expanding self-defined algorithm component of machine learning platform is disclosed, the set algorithm component is the minimum operation unit of machine learning platform,
defining and constructing an algorithm component, wherein the algorithm component comprises a data port, a logic execution unit and algorithm parameters, the data port comprises a data input port and a data output port,
the data input port receives data transmitted to the algorithm component and transmits the data to the logic execution unit, the logic execution unit performs operation on the received data by using the algorithm parameters according to the algorithm logic and transmits an operation execution result to the data output port, and the data output port stores the received execution result or transmits the received execution result to the next algorithm component.
Further, in the method for extending the custom algorithm component by the machine learning platform, a data port is defined and constructed by using data port element information, wherein the data port element information comprises a port name, a port serial number, a data type, a port type and a port description.
Further, the method for extending the user-defined algorithm component by the machine learning platform defines algorithm parameters by using algorithm parameter element information, wherein the algorithm parameter element information comprises parameter display names, unique character representation of the parameters, parameter descriptions, parameter default values, parameter types and linkage conditions among the parameters.
Further, in the method for extending the custom algorithm component by the machine learning platform, the linkage condition between the parameters refers to describing a linkage dependency relationship between the parameters by using a linkage tag, and the linkage dependency refers to that when the parameters have a dependency relationship and the dependent parameters meet a certain condition, the current parameters change along with the met condition.
The machine learning platform extends a self-defined algorithm component, the algorithm component is a minimum operation unit of the machine learning platform and comprises a data port, a logic execution unit and algorithm parameters, the data port comprises a data input port and a data output port,
the data input port receives data transmitted to the algorithm component and transmits the data to the logic execution unit, the logic execution unit performs operation on the received data by using the algorithm parameters according to the algorithm logic and transmits an operation execution result to the data output port, and the data output port stores the received execution result or transmits the received execution result to the next algorithm component.
Further, in the machine learning platform extension custom algorithm component, a data port is defined and constructed by using data port element information, wherein the data port element information comprises a port name, a port serial number, a data type, a port type and a port description.
Further, the machine learning platform extension custom algorithm component utilizes algorithm parameter element information to define algorithm parameters, wherein the algorithm parameter element information comprises parameter display names, unique character representation of the parameters, parameter descriptions, parameter default values, parameter types and linkage conditions among the parameters.
Further, the linkage condition between parameters in the algorithm parameter element information in the machine learning platform extended custom algorithm component is to describe a linkage dependency relationship between the parameters by using a linkage tag, and the linkage dependency is to change the current parameter along with a satisfied condition when the parameters have a dependency relationship and the dependent parameters satisfy a certain condition. A registration method of a machine learning platform extension custom algorithm component aims at the machine learning platform extension custom algorithm component, develops an algorithm package, uploads the algorithm package, registers algorithm package information, and analyzes and operates according to the registered algorithm information.
The invention has the advantages that:
the method for expanding the custom algorithm component by the machine learning platform realizes automatic visual presentation of the component at the front end by analyzing the algorithm parameter description and the algorithm port description based on the unified specification, has coding language independence, can support any coding language as long as the component description specification is met, can be visible and available through registration without additional operation, meets the requirement of personalized algorithm expansion, is simple to develop, has a lower threshold, and gives consideration to functions and user experience.
Drawings
FIG. 1 is a schematic flow diagram of algorithm component operation;
FIG. 2 is a flow diagram of a registration method for algorithm components.
Detailed Description
The present invention is further described below in conjunction with the following figures and specific examples so that those skilled in the art may better understand the present invention and practice it, but the examples are not intended to limit the present invention.
The invention provides a method for expanding a self-defined algorithm component by a machine learning platform, wherein the set algorithm component is the minimum operation unit of the machine learning platform,
defining and constructing an algorithm component, wherein the algorithm component comprises a data port, a logic execution unit and algorithm parameters, the data port comprises a data input port and a data output port,
the data input port receives data transmitted to the algorithm component and transmits the data to the logic execution unit, the logic execution unit performs operation on the received data by using the algorithm parameters according to the algorithm logic and transmits an operation execution result to the data output port, and the data output port stores the received execution result or transmits the received execution result to the next algorithm component.
The method of the invention completes the form specification of the algorithm component by defining the algorithm component to comprise three components of a data port, a logic execution unit and an algorithm parameter. Algorithms can be developed according to the specifications, algorithm packages are uploaded, algorithm component information is registered according to the specifications through the method, and the algorithms can be visible and usable on a machine learning platform.
In particular applications, in some embodiments of the present invention, the set algorithm component is the smallest unit of execution of the machine learning platform. The algorithm component comprises a data port, a logic execution unit and an algorithm parameter. The data port comprises a data input port and a data output port.
A specification of a description of the data ports of an algorithm component is defined.
Wherein the data port component includes 5 element information, including: port name (name), port number (sequence), data type (dataType), port type (portType, input or output), port description (description).
Port name (name): the name of the port.
Port type (portType): the port types include an input port (input) for receiving input data and an output port (output) for outputting result data.
Port number (sequence): the algorithm component has one or more ports, and when there are a plurality of ports, the ports are distinguished by marking different sequence numbers (sequences).
Data type (dataType): the data types input or output by the ports are divided into two types of models and data, and different types correspond to different processing and constraint.
Port description (description): the port description is used to explain the port's primary information to the user, such as informing the user what data the port receives or inputs, or whether the port is required, etc.
Port element codes are exemplified as follows:
the method defines a description specification of algorithm parameters in an algorithm component.
The description element information of the algorithm parameters mainly comprises: the method comprises the following steps of parameter display name (name), unique character representation (key) of the parameter, parameter description (description), parameter default value (default), type (type) of the parameter and linkage Conditions (Conditions) among the parameters.
Parameter display name (name): the name of the parameter visible to the user, the meaning of the parameter is expressed using a short word.
Unique character representation (key) of parameter: the unique character representation of the parameter is used to distinguish different parameters.
Description of parameters (description): the more detailed description of the meaning of the parameters is convenient for the user to understand.
Parameter default (default): default reference values for parameter presetting.
The parameter types are as follows: refers to the data type of the parameter, including int, double, long, text, char, date, enumeration, etc. Wherein, when the parameter type is the enumeration, an enumeration value is set through a Values tag. The parameter values can be checked and constrained according to the specified parameter types.
Linkage Conditions (Conditions) between parameters: setting a linkage dependency relationship between parameters described by a Conditions label, wherein a plurality of linkage Conditions can be described in the Conditions label. Linkage means that when there is a dependency relationship between parameters and the dependent parameters satisfy a certain condition, the current parameters will change accordingly. A Condition tag is set to describe a specific linkage Condition. In the Condition tag, a type attribute tag is set to specify a Condition type (for example, equal to, not equal to, greater than or equal to, or less than or equal to). Within the Condition tag, a referrkey attribute tag is set to specify which parameter has a dependency relationship. In the Condition tag, a FulfillingOptions tag is set for setting the value of the dependent parameter of the applicable Condition. In the following example, the Condition tag for the parameter "sigma" describes that when the value of the parameter "model" is equal to "1", the "sigma" parameter will need to be populated with the corresponding value.
An example algorithm parameter element code is as follows:
<parameter default＝"0"description＝"select model type"key＝"model"name＝""type＝"enumeration">
<Values>
<Value enName＝"SIR"value＝"0"zhName＝""/>
<Value enName＝"SEIR"value＝"1"zhName＝""/>
</Values>
</parameter>
<parameter default＝"0.14285714285714285"description＝"Define sigma in SEIR model."key＝"sigma"max＝"1.0"min＝"0.0"name＝""type＝"double">
<Conditions>
<Condition type＝"EqualTypeCondition"referkey＝"model">
<FulfillingOptions>
<FulfillingOption value＝"1"/>
</FulfillingOptions>
</Condition>
</Conditions>
</parameter>
the invention provides a machine learning platform extending self-defining algorithm component by using the above description specification algorithm component, the algorithm component is the minimum operation unit of the machine learning platform, and comprises a data port, a logic execution unit and algorithm parameters, the data port comprises a data input port and a data output port,
the data input port receives data transmitted to the algorithm component and transmits the data to the logic execution unit, the logic execution unit performs operation on the received data by using the algorithm parameters according to the algorithm logic and transmits an operation execution result to the data output port, and the data output port stores the received execution result or transmits the received execution result to the next algorithm component.
The interaction process and other contents between the constituent units in the algorithm component are based on the same concept as the method embodiment of the present invention, and specific contents can be referred to the description in the method embodiment of the present invention, and are not described herein again.
The invention provides a registration method of a machine learning platform extension self-defined algorithm component on the basis of the algorithm component, which aims at the machine learning platform extension self-defined algorithm component, develops an algorithm package, uploads the algorithm package, registers the information of the algorithm package, and analyzes and operates according to the registered algorithm information.
In a specific application, the method can comprise the following processes:
step 1: developing an algorithm package according to the specification: developing a custom algorithm according to the algorithm component form specification as described in "define an algorithm component form specification", comprising: data access processing, algorithm logic, data output processing and the like, and compiling to form an executable code packet.
Step 2: uploading algorithm package: and uploading the executable code package to a machine learning platform through a visual interface.
And step 3: registration algorithm package information: the name, unique logo, version number, encoding language, and other kits that depend on the algorithm package are filled out.
And 4, step 4: registering algorithm information according to the specification: filling in an algorithm name, a unique mark, an entry class and an algorithm description, and organizing and filling in input port description information and output port description information of an algorithm component according to a data port specification defined in a description specification defining a data port in the algorithm component; organizing and filling parameter description information of the algorithm component according to the algorithm parameter specification defined in the description specification defining the algorithm parameters in the algorithm component; and selecting an executable algorithm package corresponding to the algorithm, wherein the executable algorithm package is the algorithm package uploaded in the step 2, and the executable algorithm package is selected by a name or a unique mark filled when the algorithm package information is registered in the step 3. After the registration is finished, the newly registered algorithm can be seen in the algorithm directory, no additional operation is needed, the registration can be seen, and the registration can be seen and used.
Step 6: and (4) analysis operation: and dragging the newly registered algorithm to the canvas, and analyzing the filled algorithm component information such as input port description, output port description, parameter description and the like during algorithm registration according to the specification to form a corresponding visual component form and a parameter setting interface for presentation. And the user configures algorithm parameters through a visual interface, connects the input port and the output port, and clicks to operate. When the operator is operated, the algorithm execution entry class and the executable algorithm package are found through the unique mark filled in the algorithm registration, the execution command is started, the data is received, the algorithm logic is executed, and the data is output to the appointed storage medium or the connected next algorithm component.
It should be noted that not all steps and modules in the flows and system structures of the preferred embodiments are necessary, and some steps or modules may be omitted according to actual needs. The order of execution of the steps is not fixed unless explicitly stated and can be adjusted as desired. The system structure described in the above embodiments may be a physical structure or a logical structure, that is, some modules may be implemented by the same physical entity, or some modules may be implemented by a plurality of physical entities, or some components in a plurality of independent devices may be implemented together.
The above-mentioned embodiments are merely preferred embodiments for fully illustrating the present invention, and the scope of the present invention is not limited thereto. The equivalent substitution or change made by the technical personnel in the technical field on the basis of the invention is all within the protection scope of the invention. The protection scope of the invention is subject to the claims.
Claims (9)
1. A method for expanding self-defined algorithm component of machine learning platform is characterized by that the set algorithm component is the minimum operation unit of machine learning platform,
defining and constructing an algorithm component, wherein the algorithm component comprises a data port, a logic execution unit and algorithm parameters, the data port comprises a data input port and a data output port,
the data input port receives data transmitted to the algorithm component and transmits the data to the logic execution unit, the logic execution unit performs operation on the received data according to the algorithm logic by using the algorithm parameters and transmits an operation execution result to the data output port, and the data output port stores the received execution result or transmits the received execution result to the next algorithm component.
2. The method as claimed in claim 1, wherein the data port is defined and constructed by using data port element information, and the data port element information includes port name, port serial number, data type, port type and port description.
3. The method for extending self-defined algorithm component of machine learning platform according to claim 1 or 2, wherein algorithm parameter element information is used to define algorithm parameters, and the algorithm parameter element information includes parameter display name, unique character representation of parameters, parameter description, parameter default value, parameter type and linkage condition between parameters.
4. The method for extending the custom algorithm component by the machine learning platform according to claim 3, wherein the linkage condition between the parameters is to use a linkage tag to describe a linkage dependency relationship between the parameters, and the linkage dependency is to change the current parameter with the satisfied condition when the parameters have a dependency relationship and the dependent parameters satisfy a certain condition.
5. An extended custom algorithm component of a machine learning platform is characterized in that the algorithm component is a minimum operation unit of the machine learning platform and comprises a data port, a logic execution unit and algorithm parameters, the data port comprises a data input port and a data output port,
the data input port receives data transmitted to the algorithm component and transmits the data to the logic execution unit, the logic execution unit performs operation on the received data by using the algorithm parameters according to the algorithm logic and transmits an operation execution result to the data output port, and the data output port stores the received execution result or transmits the received execution result to the next algorithm component.
6. The machine learning platform extension custom algorithm component of claim 1, wherein a data port is constructed using data port element information definitions, the data port element information comprising port name, port number, data type, port type, and port description.
7. The machine learning platform extension self-defined algorithm component of claim 5 or 6, wherein algorithm parameter element information is used to define algorithm parameters, and the algorithm parameter element information comprises parameter display names, unique character representation of parameters, parameter descriptions, parameter default values, parameter types and linkage conditions among parameters.
8. The machine learning platform extension self-defined algorithm component of claim 7, wherein the linkage condition between parameters in the algorithm parameter element information is that linkage dependency relationship between parameters is described by using linkage tags, and the linkage dependency is that when there is dependency relationship between parameters and the dependent parameters satisfy a certain condition, the current parameters change along with the satisfied condition.
9. A registration method of a machine learning platform extension custom algorithm component is characterized in that aiming at the machine learning platform extension custom algorithm component of any one of claims 5 to 8, an algorithm package is developed and uploaded, the information of the algorithm package is registered, and the registration algorithm is analyzed and operated according to the information of the registration algorithm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110207888.6A CN112925514A (en) | 2021-02-25 | 2021-02-25 | Method for expanding custom algorithm component by machine learning platform |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110207888.6A CN112925514A (en) | 2021-02-25 | 2021-02-25 | Method for expanding custom algorithm component by machine learning platform |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112925514A true CN112925514A (en) | 2021-06-08 |
Family
ID=76171624
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110207888.6A Pending CN112925514A (en) | 2021-02-25 | 2021-02-25 | Method for expanding custom algorithm component by machine learning platform |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112925514A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113610242A (en) * | 2021-08-10 | 2021-11-05 | 中国工商银行股份有限公司 | Data processing method and device and server |
| CN114117162A (en) * | 2021-07-29 | 2022-03-01 | 鱼快创领智能科技(南京)有限公司 | Vehicle data visualization system and method |
-
2021
- 2021-02-25 CN CN202110207888.6A patent/CN112925514A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114117162A (en) * | 2021-07-29 | 2022-03-01 | 鱼快创领智能科技(南京)有限公司 | Vehicle data visualization system and method |
| CN113610242A (en) * | 2021-08-10 | 2021-11-05 | 中国工商银行股份有限公司 | Data processing method and device and server |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11822906B2 (en) | Industrial programming development with a converted industrial control program | |
| US20220129251A1 (en) | Industrial automation domain-specific language programming paradigm | |
| CN106371825B (en) | A kind of mobile terminal application interface list dynamic creation method and device | |
| US7676478B2 (en) | Data warehousing systems and methods having reusable user transforms | |
| CN109918607A (en) | Page building method and device, medium and calculating equipment | |
| CN112925514A (en) | Method for expanding custom algorithm component by machine learning platform | |
| US11256755B2 (en) | Tag mapping process and pluggable framework for generating algorithm ensemble | |
| CN113139063B (en) | Intention recognition method, device, equipment and storage medium | |
| CN110348201A (en) | A kind of configuration method and device of device security policy | |
| US7962895B2 (en) | Language for binding scalable vector graphics elements to java classes | |
| CN109471773A (en) | A kind of method, apparatus, server and the storage medium of the output of front end log multichannel | |
| CN111581920A (en) | Document conversion method, device, equipment and computer storage medium | |
| CN103109264A (en) | Creating a configuration file template | |
| US20020129330A1 (en) | System for configuration programming | |
| CN115495069B (en) | Model-driven coal industry software process implementation method, device and equipment | |
| CN113268227A (en) | Zero-code visualization software development platform and development method | |
| US10747941B2 (en) | Tag mapping process and pluggable framework for generating algorithm ensemble | |
| CN109656453A (en) | Rule editing method, apparatus and equipment | |
| CN115080046B (en) | Multi-component abstract association fusion method and device in page design | |
| Kotula | Using patterns to create component documentation | |
| CN114371762B (en) | Method, device and storage medium for printing custom labels | |
| CN115618838A (en) | Report generation method and equipment | |
| CN103488477A (en) | Visual editing system and visual editing method of JAVA interface | |
| US10628131B2 (en) | Device and method of generating programs for an information code reading device and host computer | |
| CN113949438B (en) | Unmanned aerial vehicle communication method, device, equipment and storage medium |
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 |