AutoML Pipeline Conversion to Non-Native Code
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automated machine learning (AutoML) systems primarily focus on model building and data analysis, leaving data preparation tasks under-supported, which limits their effectiveness and efficiency, especially in human-computer interaction and data scientist collaboration.
Innovation Solution
A method to convert AI pipelines generated by AutoML into a non-native format, enabling them to be used outside the AutoML system, reducing computational resources and allowing users to view, edit, and modify the code, thereby increasing the system's usefulness and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If AI pipelines are kept in native AutoML format, then they can be easily managed and executed within the AutoML system, but they cannot be used outside the AutoML system and require unnecessary computations when run within the system
Solution Approach 1:
The patent extracts the essential computational logic from the AutoML system's AI pipeline and separates it from the native AutoML environment. By converting the pipeline to a non-native format (such as Python code or executable script), the core functionality is extracted and can be executed independently outside the AutoML system, eliminating dependency on the AutoML runtime environment and reducing unnecessary computational overhead.
Solution Approach 2:
The patent creates a copy of the AI pipeline in a different format that can be executed outside the original AutoML system. This copy maintains the essential functionality while being compatible with general-purpose programming environments, allowing the pipeline to be deployed and executed in diverse settings without requiring the native AutoML infrastructure.
2Extent of automation
If AI pipelines are automated without user intervention, then the process is fully automated and fast, but users cannot view or modify the generated code, reducing confidence and collaboration
Solution Approach 1:
The patent introduces an intermediary representation layer between the automated pipeline generation and the final executable code. This intermediary format (such as human-readable Python code or structured configuration files) serves as a bridge that preserves the automation benefits while enabling human inspection and modification. Users can view, understand, and edit the pipeline code through this intermediary representation, maintaining both automation and transparency.
Solution Approach 2:
The patent implements a feedback mechanism where users can review the generated pipeline code, make modifications, and have those changes reflected in the final executed model. This feedback loop allows users to provide input to the automated system, enhancing confidence in the generated pipelines while maintaining the efficiency of automation. The system accepts user feedback in the form of code modifications and incorporates them into the final deployment.
Data Source
AI summary
An embodiment of the invention may include a method, computer program product, and system for creating a data analysis tool. The method may include a computing device that generates an AI pipeline based on an input dataset, wherein the AI pipeline is generated using an Automated Machine Learning program. The method may include converting the AI pipeline to a non-native format of the Automated Machine Learning program. This may enable the AI pipeline to be used outside of the Automated Machine Learning program, thereby increasing the usefulness of the created program by not tying it to the Automated Machine Learning program. Additionally, this may increase the efficiency of running the AI pipeline by eliminating unnecessary computations performed by the Automated Machine Learning program.