Systems and methods for generating machine learning applications

Abstract

Systems and methods for generating prebuilt machine learning framework objects comprising sets of prebuilt machine learning components and one or more data mapping requirements. The components are associated with a respective machine learning service. One or more datasets are obtained. A user-specified context for creating a particular machine learning application is obtained. A particular prebuilt object is selected based on the datasets and the context. One more candidate data mappings are identified based on the data mapping requirements and the datasets. A particular data mapping is selected. A particular set of prebuilt components is selected from the plurality of prebuilt components. The particular machine learning application is generated from the particular prebuilt object based on the particular data mapping and the particular set of prebuilt components, the particular machine learning application comprising an executable application. The machine learning application is deployed.

1 . A computing system comprising:
 one or more processors; and   memory storing instructions that, when executed by the one or more processors, cause the computing system to perform:
 generating a plurality of prebuilt machine learning framework objects, each of the prebuilt machine learning framework objects comprising a plurality of sets of prebuilt machine learning components and one or more data mapping requirements, each of the prebuilt machine learning components being associated with a respective machine learning service; 
 obtaining one or more datasets; 
 obtaining a user-specified context for creating a particular machine learning application; 
 selecting a particular prebuilt machine learning framework object from the plurality of machine learning framework objects based on the one or more datasets and the user-specified context for creating the particular machine learning application; 
 identifying one or more candidate data mappings based on the data mapping requirements of the particular prebuilt machine learning framework object and the one or more datasets; 
 determining a score for each of the respective candidate data mappings; 
 selecting a particular data mapping of the one or more candidate data mappings based on the score; 
 selecting a particular set of prebuilt machine learning components from the plurality of sets of prebuilt machine learning components of the particular prebuilt machine learning framework object which comprises:
 selecting at least two sets of prebuilt machine learning components from the plurality of sets of prebuilt machine learning components based on one or more implementation rules, the implementation rules indicating a particular platform associated with a system to execute the machine learning application; 
 scoring each of the at least two sets of prebuilt machine learning components; and 
 selecting the particular set of prebuilt machine learning components based on the scoring of the least two sets; 
 
 generating the particular machine learning application from the particular prebuilt machine learning framework object based on the particular data mapping and the particular set of prebuilt machine learning components, the particular machine learning application comprising an executable application; and 
 deploying the machine learning application. 
   
     
     
         2 . The system of  claim 1 , wherein the respective machine learning services include two or more of a data onboarding service, a data preparation service, a feature service, a model selection service, and a model deployment service. 
     
     
         3 . The system of  claim 1 , wherein the instructions further cause the system to perform:
 selecting a plurality candidate machine learning framework objects from the plurality of machine learning framework objects based on the one or more datasets and the user-specified context for creating the particular machine learning application; and   validating a particular prebuilt machine learning framework object of the plurality of candidate machine learning framework objects, the validated particular prebuilt machine learning framework object comprising the particular prebuilt machine learning framework object from the plurality of machine learning framework objects.   
     
     
         4 . The system of  claim 3 , wherein the validating further comprises:
 instantiating at least one prebuilt machine learning component of the particular prebuilt machine learning framework object of the plurality of candidate machine learning framework objects;   executing an instance of the least one prebuilt machine learning component of the particular prebuilt machine learning framework object of the plurality of candidate machine learning framework objects;   comparing one or more results of the executing the instance of the least one prebuilt machine learning component of the particular prebuilt machine learning framework object of the plurality of candidate machine learning framework objects against one or more threshold conditions; and   determining, based on the comparing, the particular prebuilt machine learning framework object of the plurality of candidate machine learning framework objects is valid.   
     
     
         5 . The system of  claim 1 , wherein the machine learning framework object is platform independent. 
     
     
         6 . A method being implemented by a computing system including one or more physical processors and storage media storing machine-readable instructions, the method comprising:
 generating a plurality of prebuilt machine learning framework objects, each of the prebuilt machine learning framework objects comprising a plurality of sets of prebuilt machine learning components and one or more data mapping requirements, each of the prebuilt machine learning components being associated with a respective machine learning service;   obtaining one or more datasets;   obtaining a user-specified context for creating a particular machine learning application;   selecting a particular prebuilt machine learning framework object from the plurality of machine learning framework objects based on the one or more datasets and the user-specified context for creating the particular machine learning application;   identifying one or more candidate data mappings based on the data mapping requirements of the particular prebuilt machine learning framework object and the one or more datasets;   determining a score for each of the respective candidate data mappings;   selecting a particular data mapping of the one or more candidate data mappings based on the score;   selecting a particular set of prebuilt machine learning components from the plurality of sets of prebuilt machine learning components of the particular prebuilt machine learning framework object which comprises:
 selecting at least two sets of prebuilt machine learning components from the plurality of sets of prebuilt machine learning components based on one or more implementation rules, the implementation rules indicating a particular platform associated with a system to execute the machine learning application; 
 scoring each of the at least two sets of prebuilt machine learning components; and 
 selecting the particular set of prebuilt machine learning components based on the scoring of the least two sets; 
   generating the particular machine learning application from the particular prebuilt machine learning framework object based on the particular data mapping and the particular set of prebuilt machine learning components, the particular machine learning application comprising an executable application; and   deploying the machine learning application.   
     
     
         7 . The method of  claim 6 , wherein the respective machine learning services include two or more of a data onboarding service, a data preparation service, a feature service, a model selection service, and a model deployment service. 
     
     
         8 . The method of  claim 6 , further comprising:
 selecting a plurality candidate machine learning framework objects from the plurality of machine learning framework objects based on the one or more datasets and the user-specified context for creating the particular machine learning application; and   validating a particular prebuilt machine learning framework object of the plurality of candidate machine learning framework objects, the validated particular prebuilt machine learning framework object comprising the particular prebuilt machine learning framework object from the plurality of machine learning framework objects.   
     
     
         9 . The method of  claim 8 , wherein the validating further comprises:
 instantiating at least one prebuilt machine learning component of the particular prebuilt machine learning framework object of the plurality of candidate machine learning framework objects;   executing an instance of the least one prebuilt machine learning component of the particular prebuilt machine learning framework object of the plurality of candidate machine learning framework objects;   comparing one or more results of the executing the instance of the least one prebuilt machine learning component of the particular prebuilt machine learning framework object of the plurality of candidate machine learning framework objects against one or more threshold conditions; and   determining, based on the comparing, the particular prebuilt machine learning framework object of the plurality of candidate machine learning framework objects is valid.   
     
     
         10 . The method of  claim 6 , wherein the machine learning framework object is platform independent. 
     
     
         11 . A non-transitory computer readable medium comprising instructions that, when executed, cause one or more processors to perform:
 generating a plurality of prebuilt machine learning framework objects, each of the prebuilt machine learning framework objects comprising a plurality of sets of prebuilt machine learning components and one or more data mapping requirements, each of the prebuilt machine learning components being associated with a respective machine learning service;   obtaining one or more datasets;   obtaining a user-specified context for creating a particular machine learning application;   selecting a particular prebuilt machine learning framework object from the plurality of machine learning framework objects based on the one or more datasets and the user-specified context for creating the particular machine learning application;   identifying one or more candidate data mappings based on the data mapping requirements of the particular prebuilt machine learning framework object and the one or more datasets;   determining a score for each of the respective candidate data mappings;   selecting a particular data mapping of the one or more candidate data mappings based on the score;   selecting a particular set of prebuilt machine learning components from the plurality of sets of prebuilt machine learning components of the particular prebuilt machine learning framework object which comprises:
 selecting at least two sets of prebuilt machine learning components from the plurality of sets of prebuilt machine learning components based on one or more implementation rules, the implementation rules indicating a particular platform associated with a system to execute the machine learning application; 
 scoring each of the at least two sets of prebuilt machine learning components; and 
 selecting the particular set of prebuilt machine learning components based on the scoring of the least two sets; 
   generating the particular machine learning application from the particular prebuilt machine learning framework object based on the particular data mapping and the particular set of prebuilt machine learning components, the particular machine learning application comprising an executable application; and   deploying the machine learning application.   
     
     
         12 . The non-transitory computer readable medium of  claim 11 , wherein the respective machine learning services include two or more of a data onboarding service, a data preparation service, a feature service, a model selection service, and a model deployment service. 
     
     
         13 . The non-transitory computer readable medium of  claim 11 , further comprising:
 selecting a plurality candidate machine learning framework objects from the plurality of machine learning framework objects based on the one or more datasets and the user-specified context for creating the particular machine learning application; and   validating a particular prebuilt machine learning framework object of the plurality of candidate machine learning framework objects, the validated particular prebuilt machine learning framework object comprising the particular prebuilt machine learning framework object from the plurality of machine learning framework objects.   
     
     
         14 . The non-transitory computer readable medium of  claim 13 , wherein the validating further comprises:
 instantiating at least one prebuilt machine learning component of the particular prebuilt machine learning framework object of the plurality of candidate machine learning framework objects;   executing an instance of the least one prebuilt machine learning component of the particular prebuilt machine learning framework object of the plurality of candidate machine learning framework objects;   comparing one or more results of the executing the instance of the least one prebuilt machine learning component of the particular prebuilt machine learning framework object of the plurality of candidate machine learning framework objects against one or more threshold conditions; and   determining, based on the comparing, the particular prebuilt machine learning framework object of the plurality of candidate machine learning framework objects is valid.