Database Machine Learning Model Deployment via Integrated Functions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Database system environments lack support for machine learning model deployment, leading to increased data processing and storage costs, security concerns, and usability issues for business analysts due to the need to export data for external processing, which complicates integration with existing SQL queries and analytical tools.
Innovation Solution
A method that extracts model data from a trained model, integrates it within a database system environment as a user-defined function, and deploys it for execution, allowing the model to operate within the database system environment using existing SQL queries and tools, thereby avoiding data export and enhancing usability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If machine learning models are deployed externally to database systems, then model training and execution flexibility is improved, but data security deteriorates and data processing costs increase
Solution Approach 1:
The patent merges the machine learning model execution environment with the database system by integrating Python runtime and model artifacts directly into the database server. This allows models to be deployed, stored, and executed within the database system itself, eliminating the need to export data to external systems while maintaining model deployment flexibility.
Solution Approach 2:
The patent introduces an intermediary layer consisting of Python runtime environment and model execution engine that sits within the database system. This intermediary enables machine learning workloads to run inside the database without requiring data to leave the secure database environment, thus maintaining security while providing ML capabilities.
2Adaptability or versatility
If data is exported for external model processing, then model execution capability is improved, but data processing costs and storage requirements increase
Solution Approach 1:
The patent combines data storage and model execution in a single system environment. By integrating the Python runtime and model artifacts within the database server, the system eliminates the need to export data for processing, thereby reducing data transfer overhead, storage requirements, and processing costs while maintaining full model execution capability.
3Adaptability or versatility
If machine learning models are deployed externally, then model functionality is improved, but integration complexity with SQL queries deteriorates
Solution Approach 1:
The patent creates a universal execution environment within the database system that can handle both traditional SQL queries and machine learning model execution through a common interface. The Python runtime and model execution engine are integrated into the database server, allowing users to invoke models using standard SQL syntax while maintaining full model functionality.
4Adaptability or versatility
If data is exported for external processing, then model access flexibility is improved, but usability for business analysts deteriorates
Solution Approach 1:
The patent merges model access capabilities directly into the familiar SQL interface that business analysts already use. By integrating Python runtime and model execution within the database system, analysts can execute models using standard SQL queries without needing to learn new tools or export data, thereby maintaining ease of operation while providing flexible model access.
Data Source
AI summary
Model data comprising a model object and model metadata is extracted from a trained model. The model data is integrated within a function executable from within a database system environment. The integrated function is deployed within the database system environment, the deploying activating the trained model for execution within the database system environment.