AutoML Component Determination via Pipeline Subset Evaluation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automated machine learning (AutoML) technologies face high calculation costs when determining whether to add a new component to the search range, as they evaluate the performance of all pipelines formed by combinations of multiple components before and after the addition.
Innovation Solution
The proposed solution involves narrowing down the pipelines to be evaluated by identifying one or more machine learning pipelines with components of the same type as the target component, generating replacement pipelines with the target component, and comparing their performances to determine if the target component should be added to the search range.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If all pipelines formed by combinations of multiple components are evaluated to determine whether to add a new component, then the determination accuracy is improved, but the calculation cost increases
Solution Approach 1:
The patent segments the evaluation process into two stages: first evaluating a subset of representative pipelines (those containing components of the same type as the target component) to determine whether the new component should be added to the search range, and then potentially evaluating all pipelines in a subsequent stage. This segmentation allows the system to achieve reasonable determination accuracy while significantly reducing the calculation cost by avoiding comprehensive evaluation of all possible pipeline combinations in the initial decision-making process.
2Reliability
If all pipelines formed by combinations of multiple components are evaluated to determine whether to add a new component, then the reliability of component selection is improved, but the processing time increases
Solution Approach 1:
The patent performs preliminary evaluation of a selected subset of pipelines before committing to comprehensive evaluation of all pipelines. By first assessing representative pipelines that include components of the same type as the target component, the system can make an informed preliminary decision about whether the new component deserves inclusion in the search range, thereby saving processing time while maintaining adequate selection reliability.
Solution Approach 2:
The patent applies partial evaluation by assessing only a portion of the total possible pipelines (those containing same-type components) rather than performing excessive comprehensive evaluation of all pipeline combinations. This partial action approach provides sufficient information for component selection decisions while avoiding the time cost of evaluating every possible pipeline configuration.
Data Source
AI summary
A machine learning pipeline component determination program causes a computer to execute a process. The process including: obtaining a type of a first component; identifying one or more first machine learning pipelines including a component of the same type as the type of the first component among a plurality of machine learning pipelines outputted for a plurality of datasets by a program that generates machine learning pipelines including components selected from among a plurality of components depending on a task; generating one or more second machine learning pipelines in which the component of the same type is changed to the first component; and determining whether or not to add the first component to the plurality of components based on a result of comparison between a performance of each of the first machine learning pipelines and a performance of each corresponding second machine learning pipelines