Machine Learning Model Repository for Standardized Reuse
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Machine learning models are often custom-designed for specific data sets and problems, making their practical reuse, sharing, or reapplication difficult due to their tailored nature.
Innovation Solution
A machine learning repository system that manages machine learning model envelopes, transforms raw data into model objects conforming to a schema, and selects appropriate ML models for scoring based on parameter models, allowing for the sharing and reuse of ML models across different applications.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning models are custom-designed for specific data sets and problems, then the model performance and accuracy are improved, but the ease of reuse and sharing of models deteriorates
Solution Approach 1:
The patent segments a machine learning model into multiple independent components: a parameter model defining the data structure schema, model objects conforming to that schema, and the actual ML model. This segmentation allows the parameter model and schema to be shared and reused across different applications while the specific model objects can be customized, thus enabling model reuse without sacrificing accuracy.
Solution Approach 2:
The patent uses parameter models to define schemas that can be modified to accommodate different data sets and problems. By changing the parameter definitions in the schema rather than redesigning the entire model, organizations can adapt existing models to new contexts, improving reuse capability while maintaining the structured approach that ensures model accuracy.
2Reliability
If machine learning models are custom-designed for individual data sets, then the reliability for specific problems is improved, but the device complexity for managing and sharing models increases
Solution Approach 1:
The patent creates a universal parameter model schema that can serve multiple functions: defining data structures for different problems, validating model objects, and enabling model sharing. This single schema framework handles diverse machine learning applications, reducing the need for separate management systems for different models and thereby reducing overall complexity.
Solution Approach 2:
The parameter model acts as an intermediary layer between the raw data and the machine learning model. It provides a standardized interface and validation mechanism that simplifies model management by automatically checking conformity to schemas, thus reducing the complexity of managing custom models across different data sets.
3Manufacturing precision
If machine learning models are tailored for specific problems, then the manufacturing precision of model design is improved, but the ease of operation for sharing and reapplying models deteriorates
Solution Approach 1:
The patent performs preliminary actions by defining parameter models and schemas before creating specific machine learning models. This advance structuring of data requirements and validation rules enables models to be shared and reused more easily, as the framework is already in place to accommodate different implementations without requiring redesign.
Data Source
AI summary
Embodiments are directed towards a machine learning repository for managing machine learning (ML) model envelopes, ML models, model objects, or the like. Questions and model objects may be received by a ML model answer engine. Machine learning (ML) model envelopes may be received based on the questions. The model objects may be compared to parameter models associated with the ML model envelopes. ML model envelopes may be selected based on the comparison such that the model objects satisfy the parameter models of each of the selected ML model envelopes. ML models included in each selected ML model envelope may be executed to provide score values for the model objects and the score values may be included in a report.